The Routledge Handbook of Philosophy of Memory-Routledge (2004) PDF

THE ROUTLEDGE HANDBOOK OF
PHILOSOPHY OF MEMORY
Memory occupies a fundamental place in philosophy, playing a central role not only in the history of
philosophy but also in philosophy of mind, epistemology, and ethics. Yet the philosophy of memory
has only recently emerged as an area of study and research in its own right.
The Routledge Handbook of Philosophy of Memory is an outstanding reference source on the key topics,
problems and debates in this exciting area, and is the first philosophical collection of its kind. The forty-
eight chapters are written by an international team of contributors, and divided into nine parts:
• The nature of memory

• The metaphysics of memory
• Memory, mind, and meaning
• Memory and the self
• Memory and time
• The social dimension of memory
• The epistemology of memory
• Memory and morality
• History of philosophy of memory.
Within these sections, central topics and problems are examined, including: truth, consciousness, imag-
ination, emotion, self-knowledge, narrative, personal identity, time, collective and social memory,
internalism and externalism, and the ethics of memory. The final part examines figures in the history
of philosophy, including Aristotle, Augustine, Freud, Bergson, Wittgenstein, and Heidegger, as well as
perspectives on memory in Indian and Chinese philosophy.
Essential reading for students and researchers in philosophy, particularly philosophy of mind and psy-
chology, the Handbook will also be of interest to those in related fields, such as psychology and anthropology.
Sven Bernecker is Professor of Philosophy at the University of Cologne, Germany and the University
of California, Irvine, USA. He is the author of Reading Epistemology (2006), The Metaphysics of Memory
(2008), and Memory: A Philosophical Study (2010) and is co-editor with Duncan Pritchard of The Routledge
Companion to Epistemology (2011).
Kourken Michaelian is Senior Lecturer in Philosophy at the University of Otago, New Zealand. He
is the author of Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past (2016) and
is co-editor with Stanley B. Klein and Karl K. Szpunar of Seeing the Future: Theoretical Perspectives on
Future-Oriented Mental Time Travel (2016).
ROUTLEDGE HANDBOOKS IN PHILOSOPHY
Routledge Handbooks in Philosophy are state-of-the-art surveys of emerging, newly refreshed, and
important fields in philosophy, providing accessible yet thorough assessments of key problems,
themes, thinkers, and recent developments in research.
All chapters for each volume are specially commissioned, and written by leading scholars in
the field. Carefully edited and organized, Routledge Handbooks in Philosophy provide indispensa-
ble reference tools for students and researchers seeking a comprehensive overview of new and
exciting topics in philosophy. They are also valuable teaching resources as accompaniments to
textbooks, anthologies, and research-orientated publications.
Recently published:
THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF EMPATHY

Edited by Heidi Maibom
THE ROUTLEDGE HANDBOOK OF EPISTEMIC CONTEXTUALISM

Edited by Jonathan Jenkins Ichikawa
THE ROUTLEDGE HANDBOOK OF EPISTEMIC INJUSTICE

Edited by Ian James Kidd, José Medina and Gaile Pohlhaus
THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF PAIN

Edited by Jennifer Corns
THE ROUTLEDGE HANDBOOK OF BRENTANO AND

THE BRENTANO SCHOOL
Edited by Uriah Kriegel
THE ROUTLEDGE HANDBOOK OF COLLECTIVE INTENTIONALITY

Edited by Marija Jankovic and Kirk Ludwig
THE ROUTLEDGE HANDBOOK OF LIBERTARIANISM

Edited by Jason Brennan, Bas van der Vossen and David Schmidtz
THE ROUTLEDGE HANDBOOK OF METAETHICS

Edited by Tristram McPherson and David Plunkett
THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF MEMORY

Edited by Sven Bernecker and Kourken Michaelian
THE ROUTLEDGE
HANDBOOK OF PHILOSOPHY
OF MEMORY
Edited by
Sven Bernecker and
Kourken Michaelian
First published 2017
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
711 Third Avenue, New York City, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
2017 selection and editorial matter, Sven Bernecker and Kourken
Michaelian; individual chapters, the contributors
The right of Sven Bernecker and Kourken Michaelian to be identified as
the authors of the editorial material, and of the authors for their individual
chapters, has been asserted in accordance with sections 77 and 78 of the
Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in
any information storage or retrieval system, without permission in writing
from the publishers.
Trademark notice: Product or corporate names may be trademarks or registered
trademarks, and are used only for identification and explanation without
intent to infringe.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
Names: Bernecker, Sven, editor.
Title: The Routledge handbook of philosophy of memory / edited by Sven
Bernecker and Kourken Michaelian.
Description: 1 [edition]. | New York : Routledge, 2017. |
Series: Routledge handbooks in philosophy | Includes bibliographical
references and index.
Identifiers: LCCN 2016053811| ISBN 9781138909366 (hardback : alk.
paper) | ISBN 9781315687315 (e-book)
Subjects: LCSH: Memory (Philosophy)
Classification: LCC BD181.7 .R68 2017 | DDC 128/.3—dc23
LC record available at https://lccn.loc.gov/2016053811
ISBN: 978-1-138-90936-6 (hbk)

ISBN: 978-1-315-68731-5 (ebk)
Typeset in Bembo Std

by Swales & Willis Ltd, Exeter, Devon, UK
CONTENTS
List of figures and tables xi

Notes on contributors xiii
Acknowledgements xxi
Editors’ introduction: the philosophy of memory today 1

Sven Bernecker and Kourken Michaelian
PART I
The nature of memory 5
1 Taxonomy and unity of memory 7

Markus Werning and Sen Cheng
2 The phenomenology of memory 21

Fabrice Teroni
3 Memory and levels of scientific explanation 34

John Bickle
PART II
The metaphysics of memory 49
4 Memory and truth 51

Sven Bernecker
5 Memory causation 63
Dorothea Debus
v
Contents
6 Memory traces 76
Sarah K. Robins
7 The intentional objects of memory 88

Jordi Fernández
PART III
Memory, mind, and meaning 101
8 Memory and consciousness 103

Paula Droege
9 Memory and perspective 113

Christopher Jude McCarroll and John Sutton
10 Memory and imagination 127

Felipe De Brigard
11 Memory images 141

Elizabeth Irvine
12 Memory and emotion 154

Ronald de Sousa
PART IV
Memory and the self 167
13 Memory and personal identity 169

Shaun Nichols
14 Memory and self-consciousness 180

José Luis Bermúdez
15 Memory and narrativity 192

Daniel D. Hutto
PART V
Memory and time 205
16 Memory and the concept of time 207

Christoph Hoerl
17 Memory and the metaphysics of time 219

Robin Le Poidevin
vi
Contents
18 Memory as mental time travel 228

Denis Perrin and Kourken Michaelian
PART VI
The social dimension of memory 241
19 Extended memory 243

Robert W. Clowes
20 Collective memory 255

Jeffrey Andrew Barash
21 Memory and social identity 268

Robyn Fivush and Matthew Graci
PART VII
The epistemology of memory 281
22 Internalism and externalism 283

Brent J. C. Madison
23 Foundationalism 296
Berit Brogaard
24 Coherentism 310
Erik J. Olsson
25 Preservation and generation 323

Thomas D. Senor
26 Skepticism and memory 335

Andrew Moon
PART VIII
Memory and morality 349
27 A duty to remember 351

Jeffrey Blustein
28 An obligation to forget 364

David Matheson
29 The ethics of memory modification 373

S. Matthew Liao
vii
Contents
PART IX
History of the philosophy of memory 383
30 Plato 385
Sophie-Grace Chappell
31 Aristotle 396
32 Classical Indian philosophy 408

Jonardon Ganeri
33 Indian Buddhist philosophy 416

Monima Chadha
34 Chinese Buddhist philosophy 428

Chung-Ying Cheng
35 Augustine 439
Lilianne Manning
36 Avicenna and Averroes 448

Deborah L. Black
37 Thomas Aquinas 461

John O’Callaghan
38 John Locke and Thomas Reid 470

Rebecca Copenhaver
39 David Hume 480

Daniel E. Flage
40 G. W. F Hegel 487
Valentina Ricci
41 Sigmund Freud and Jacques Lacan 496

Martin Schwab
42 Henri Bergson 510

Trevor Perri
43 Bertrand Russell 519

Paulo Faria
viii
Contents
44 Maurice Halbwachs 528

Dmitri Nikulin
45 Frederic Bartlett 537

Brady Wagoner
46 Ludwig Wittgenstein 546

Andy Hamilton
47 Martin Heidegger 557

Taylor Carman
48 Paul Ricoeur 563

Alexandre Dessingué
Index 572
ix
FIGURES AND TABLES
Figures
1.1 Hierarchical taxonomy of memory according to Squire and
Zola-Morgan (1988) and its relationship to Tulving’s (1985)
phenomenological taxonomy 12
3.1 Levels of scientific explanation of spatial memory in a rodent model 36
11.1 Sperling’s ‘partial report’ paradigm 143
11.2 The basic differences between iconic-cue, retro-cue, and
post-cue conditions 148
23.1 Despite the fact that some of the parts of the coffee mug are
occluded, the whole mug ‘pops out’ and your experience of
it has presentational phenomenology, on Chudnoff’s view 300
23.2 The Müller-Lyer illusion – even when you learn that the line
segments on the left have the same length, they continue to
appear as if they have different length 302
23.3 The Müller-Lyer illusion – illustration of how the outside corners
generate the appearance of the object being further away from us,
whereas the inside corners generate the appearance of the object
being closer to us 303
41.1 Lacan’s Schema L 502
42.1 Bergson’s cone of memory 515
Table
22.1 The reliability table 287
41.1 The Wolf Man’s dream as “Primal Scene” 500
xi
CONTRIBUTORS
Jeffrey Andrew Barash received his PhD at the University of Chicago, USA, and then taught
in the philosophy department of the University of Amiens in France, where he is an Emeritus
professor. He is currently a Fellow at the Institute for Advanced Study in Princeton. On the topic
of memory, he is the author of the book Collective Memory and the Historical Past (2016).
José Luis Bermúdez is Professor of Philosophy at Texas A&M University, USA. His books
include The Paradox of Self-Consciousness (1998), Thinking without Words (2003), Decision Theory
and Rationality (2009), and Understanding “I”: Thought and Language, in addition to Philosophy of
Psychology: A Contemporary Introduction (2005) and Cognitive Science: An Introduction to the Science
of the Mind (2nd edn., 2014).
Sven Bernecker is Humboldt Professor at the University of Cologne, Germany and Professor
of Philosophy at the University of California, Irvine, USA. His main areas of research are epis-
temology, metaphysics and philosophy of mind. He is the author of Reading Epistemology (2006),
The Metaphysics of Memory (2008), and Memory: A Philosophical Study (2010) and is co-editor
with Fred Dretske of Knowledge (2000) and co-editor with Duncan Pritchard of The Routledge
Companion to Epistemology (2011).
John Bickle is Professor of Philosophy, Adjunct Professor of Psychology, Head of the

Department of Philosophy and Religion, and Research Fellow at the Institute for Imaging and
Analytical Technologies (I2AT) at Mississippi State University, USA. He is also Affiliated Faculty
in the Department of Neurobiology and Anatomical Sciences at the University of Mississippi
Medical Center. Bickle works in philosophy of neuroscience, philosophy of science (scientific
reductionism), and cellular and molecular mechanisms of cognition and consciousness. He is the
author of four books from major academic presses, more than eighty publications in philosophy,
neuroscience, and cognitive science journals and edited volumes, and the editor of The Oxford
Handbook of Philosophy and Neuroscience (2009).
xiii
Contributors
Deborah L. Black is Professor of Philosophy and Medieval Studies at the University of

Toronto, Canada. She specializes in medieval philosophy, especially classical Islamic philosophy
and thirteenth-century Western philosophy. Her publications focus on the areas of cognitive
psychology, in particular theories of the internal senses and the intellect, epistemology, and
related topics in metaphysics.
Jeffrey Blustein is the Arthur Zitrin Professor of Bioethics and Professor of Philosophy at
City College of the City University of New York and a member of the doctoral faculty of the
Graduate Center, CUNY, USA. He is the author most recently of two monographs on ethical
and philosophical issues in memory: The Moral Demands of Memory (2008) and Forgiveness and
Remembrance: Remembering Wrongdoing in Personal and Public Life (2014). His research interests
include bioethics, memory studies, and moral psychology. He is currently working on a project
about the nature, role, and significance of collective memory.
Berit Brogaard is Professor of Philosophy at the University of Miami, USA. Her areas of
research include philosophy of perception, philosophy of emotions, and philosophy of language.
She is the author of Transient Truths (2012), On Romantic Love (2015) and The Superhuman
Mind (2015).
Taylor Carman is Professor of Philosophy at Barnard College, Columbia University, USA.

He coedited The Cambridge Companion to Merleau-Ponty (2004) and is the author of Heidegger’s
Analytic: Interpretation, Discourse, and Authenticity in “Being and Time” (2003) and Merleau-Ponty
(2008). He has written articles on various topics in phenomenology and philosophy of mind and
is currently working on a book on Heidegger’s critique of metaphysics.
Monima Chadha is Head of Philosophy of the Philosophy Program at Monash University,

Australia. Her principal research area is the cross-cultural and interdisciplinary philosophy of
mind, specifically the classical Indian and contemporary Western philosophy of mind. She has
published in leading academic journals like Philosophy East and West; Phenomenology and Cognitive
Sciences; and Review of Philosophy and Psychology. Currently she is writing a book on the philo-
sophical evolution of mind in Buddhism and its centrality to the doctrine in the absence of self.
Sophie-Grace Chappell is Professor of Philosophy at The Open University, UK. Her books
include Reading Plato’s Theaetetus (2005) and Knowing What to Do (2014).
Chung-Ying Cheng received his PhD in Philosophy from Harvard University, USA. His
specialization includes logic, ethics, philosophy of language, metaphysics and Chinese philoso-
phy. His recent works cover onto-hermeneutics, Yijing, and Confucianism/Neo-Confucianism.
He has been Professor of Philosophy at the University of Hawaii at Manoa since 1972. He has
lectured around the world in institutions such as Yale University, Oxford University, London
University, and Berlin University. In 1973, Professor Cheng founded the academic quar-
terly Journal of Chinese Philosophy. In 1975, he founded the International Society for Chinese
Philosophy. In 1985, he founded the International Society for the Yijing. He has published
thirty-three books in both English and Chinese and published more than 300 papers in vari-
ous fields of philosophy, including contemporary Chinese philosophy, theory of Confucian
philosophy, creating harmony, ontology and interpretation, philosophy of Yijing ontology, the
collected papers of Chung-Ying Cheng, and onto-aesthetics.
xiv
Contributors
Sen Cheng is Professor of Computational Neuroscience in the Institute for Neural Computation
at the Ruhr University of Bochum, Germany, and co-chair of the Mercator Research Group
Structure of Memory. The goal of his research is to understand the cognitive and neural mecha-
nisms of episodic memory and spatial representations. To this end, he mostly uses computational
methods, including spiking neural networks, cognitive modeling and machine learning. He has
published more than twenty peer-reviewed articles in the field of learning and memory.
Robert W. Clowes received his PhD in Cognitive Science from Sussex University, UK.
Since 2010, he has been based at IFILNOVA (The Philosophy Institute of Nova University,
Lisbon), FCSH where his research project is funded by the personal Portuguese government
FCT grant BPD/70440/2010. He currently chairs the Lisbon Mind and Cognition Group
which is part of the ArgLab. He works on core themes in the philosophy of cognitive science
and philosophy of technology.
Rebecca Copenhaver is a Professor of Philosophy at Lewis & Clark College, USA, where she
has taught since 2001. Her research interests are in modern philosophy and philosophy of mind.
She is writing a book on Thomas Reid’s science of mind.
Felipe De Brigard (PhD) is Assistant Professor in the departments of Philosophy, Psychology

and Neuroscience, and core faculty in the Center for Cognitive Neuroscience and the
Duke Institute for Brain Sciences, USA. He is also Principal Investigator of the Imagination
and Modal Cognition Laboratory, and the associate director of the Center for Societal and
Philosophical Implications of Neuroscience, affiliated with the Science and Society Initiative at
Duke University. His research focuses on the interactions between memory and imagination,
particularly counterfactual thinking, and has been published in philosophical and scientific
venues, including Synthese, Analysis, Psychological Science, Cognition, Neuroimage, and Memory
and Cognition.
Dorothea Debus teaches Philosophy at the University of York, UK. Her main research inter-
ests lie in the philosophy of mind and moral psychology, and her published work considers
philosophical questions relating to memory, imagination, attention, and the emotions.
Ronald de Sousa is Professor Emeritus of Philosophy at the University of Toronto, Canada.

His books include The Rationality of Emotion (1987); Why Think? Evolution and the Rational
Mind (2007), Emotional Truth (2011) and Love, a Very Short Introduction (2015).
Alexandre Dessingué is Professor of Literacy studies at the University of Stavanger in Norway.

He has published several articles and book chapters in the field of memory studies and he is
working now on several projects in the multidisciplinary field of critical literacy and humanities
in education. His last publication Beyond Memory: Silence and the Aesthetics of Remembrance (2016)
has been co-edited with Prof. Jay Winter.
Paula Droege is currently a Fellow at the Institute for Advanced Study/Wissenschaftskolleg

in Berlin, Germany, where she is part of an interdisciplinary Focus Group on Animal Pain. Her
home institution is in the Department of Philosophy at Pennsylvania State University. She is
the author of Caging the Beast: A Theory of Sensory Consciousness (2003) and articles on the role of
consciousness in memory, free will, and delusions such as confabulation. Her research proposes
a temporal representation theory of consciousness.
xv
Contributors
Paulo Faria is Professor of Philosophy at the Federal University of Rio Grande do Sul, Brazil.
He works in metaphysics and epistemology, the philosophies of logic and language, and the
history of analytical philosophy.
Jordi Fernández is Associate Professor of Philosophy at the University of Adelaide, Australia.

His teaching and research interests are in philosophy of mind, epistemology, and metaphysics.
He has published articles on the philosophy of memory in the Australasian Journal of Philosophy,
Philosophy and Phenomenological Research, Philosophical Studies and Synthese. He is the author of
Transparent Minds (2013).
Robyn Fivush is the Samuel Candler Dobbs Professor of Psychology at Emory University,
USA, where she has been on the faculty since 1984. She received her PhD from the Graduate
Center of the City University of New York in 1983 and was a Postdoctoral Fellow at the
Center for Human Information Processing, University of California at San Diego from 1983 to
1984. She is associated faculty with the Department of Women’s Studies and a Senior Fellow
in the Center for the Study of Law and Religion. Her research focuses on early memory with
an emphasis on the social construction of autobiographical memory and the relations among
memory, narrative, identity, trauma, and coping. She has published over 150 books, book chapters,
and articles.
Daniel E. Flage is Professor of Philosophy at James Madison University, USA. Since complet-
ing his PhD at the University of Iowa in 1977, his work has focused on Hume, Berkeley, and
Descartes.
Jonardon Ganeri’s research interests are in consciousness, self, attention, the epistemology of
inquiry, intellectual affinities between India, Greece and China, and early Buddhist philosophy
of mind. He teaches courses in the philosophy of mind, the nature of subjectivity, Buddhist
philosophy, the history of Indian philosophical traditions, and supervises PhDs on Indian philo-
sophical texts in classical Sanskrit. His books include Philosophy in Classical India: The Proper
Work of Reason (2001), The Concealed Art of the Soul (2007), The Lost Age of Reason: Philosophy in
Early Modern India 1450–1700 (2011), and The Self: Naturalism, Consciousness, and the First-Person
Stance (2012). He is currently editing The Oxford Handbook of Indian Philosophy. He is a Fellow
of the British Academy, and laureate of the Infosys Prize in the Humanities.
Matthew Graci is a graduate student at Emory University, USA. He received his BA in

Psychology and Philosophy from the University of Dayton in 2011. He received his MEd in
Human Development Counseling from Vanderbilt University in 2013. His research focuses
autobiographical memory with an emphasis on narrative, meaning-making, emotion-regulation,
stress and coping, and attachment.
Andy Hamilton teaches philosophy at Durham University, UK, specializing in aesthetics,

philosophy of mind, political philosophy and the history of nineteenth- and twentieth-century
philosophy, especially Wittgenstein. His monographs include Aesthetics and Music (2007), The Self
in Question: Memory, the Body and Self-Consciousness (2013), and The Routledge Philosophy Guidebook
to Wittgenstein and On Certainty (2014).
xvi
Contributors
Christoph Hoerl is Professor of Philosophy at the University of Warwick, UK. His research is
mainly in the philosophy of mind, with a particular interest in philosophical questions about the
nature of temporal experience, memory, and our ability to think about time.
Daniel D. Hutto is Professor of Philosophical Psychology at the University of Wollongong,

Australia, and member of the Australian Research Council College of Experts. His most recent
books include Wittgenstein and the End of Philosophy (2006) and Folk Psychological Narratives (2008).
He is co-author of the award-winning Radicalizing Enactivism (2013) and editor of Narrative and
Understanding Persons (2007) and Narrative and Folk Psychology (2009). A special yearbook, Radical
Enactivism, focusing on his philosophy of intentionality, phenomenology and narrative, was
published in 2006. He regularly speaks at conferences and expert meetings for anthropologists,
clinical psychiatrists, educationalists, narratologists, neuroscientists, and psychologists.
Elizabeth Irvine (PhD) is Lecturer in Philosophy at Cardiff University, UK. Her research
interests are in the philosophy of the mind sciences and philosophy of science, and cover top-
ics such as research methodology in consciousness science, modeling and simulation, and the
evolution of language.
Robin Le Poidevin is Professor of Metaphysics at the University of Leeds, UK, where he

has taught since 1989. His interests include the metaphysics of space and time, and their inter-
sections with other areas of philosophy, in particular philosophy of mind, religion and aesthetics.
He is the author of Change, Cause and Contradiction (1991), Arguing for Atheism (1996), Travels
in Four Dimensions (2003), The Images of Time (2007), and Agnosticism: A Very Short Introduction
(2010). He is one of the co-editors of The Routledge Companion to Metaphysics (2009).
S. Matthew Liao is Arthur Zitrin Professor of Bioethics, Director of the Center for Bioethics,
and Affiliated Professor in the Department of Philosophy at New York University, USA. He is
the author or editor of The Philosophical Foundations of Human Rights (2015); The Right to Be Loved
(2015); Current Controversies in Bioethics (2016); Moral Brains: The Neuroscience of Morality (2016)
and over fifty articles in philosophy and bioethics. He is the editor-in-chief for the Journal of
Moral Philosophy, a peer-reviewed international journal of moral, political, and legal philosophy.
Christopher Jude McCarroll is a research assistant in the Department of Cognitive Science at

Macquarie University, Australia. He works on perspective in episodic memory and imagination.
His PhD, Point of View in Personal Memory: A Philosophical Investigation, was awarded in 2015.
Brent J. C. Madison is Assistant Professor of Philosophy at the United Arab Emirates University.
His main research interests lie in epistemology, and in related issues in the philosophy of reli-
gion and the philosophy of mind. His papers in these areas have appeared in The Australasian
Journal of Philosophy, Pacific Philosophical Quarterly, Philosophy Compass, and Erkenntnis, among other
places. The primary focus of his current research is on the internalism/externalism distinction in
epistemology.
Lilianne Manning is Professor of Neuropsychology and founder of the Master of Cognitive

Neuropsychology, Strasbourg University, France. She is author of seventy peer-reviewed sci-
entific papers, five books, and eighteen book chapters. Her research focuses on normal and
xvii
Contributors
pathological autobiographical memory, and fMRI. Recent investigation: Interpretative

Phenomenological Analysis in clinical settings.
David Matheson (PhD, Brown University) is an associate professor in the Department of

Philosophy at Carleton University, Canada. His research interests include social epistemology,
value-theoretic issues at the intersection of epistemology and ethics, and the philosophy of life.
Kourken Michaelian is Senior Lecturer in Philosophy at the University of Otago, New

Zealand. He is the author of Mental Time Travel: Episodic Memory and Our Knowledge of the
Personal Past (2016) and is co-editor with Stanley B. Klein and Karl K. Szpunar of Seeing the
Future: Theoretical Perspectives on Future-Oriented Mental Time Travel (2016).
Andrew Moon’s area of specialty is epistemology, and he has interests in philosophy of mind,
metaethics, and philosophy of religion. He has recently published on memory, the internalism/
externalism debate, and evolutionary moral debunking arguments. He is also working on a book
on the relationship between knowledge and the doxastic attitudes (belief, doubt, confidence,
and certainty). He is currently a religious experience research fellow at the University of Notre
Dame, USA.
Shaun Nichols is Professor of Philosophy at the University of Arizona, USA. He works at the
intersection of philosophy and cognitive science.
Dmitri Nikulin is Professor of Philosophy at the New School for Social Research in New York,
USA. His interests range from ancient philosophy and early modern science to the philosophy
of memory and philosophy of history. He is the author of a number of books including Matter,
Imagination and Geometry (2002), On Dialogue (2006), Dialectic and Dialogue (2010), Comedy, Seriously
(2014), and The Concept of History (2017). He is also the editor of Memory: A History (2015).
John O’Callaghan is Associate Professor of Philosophy at the University of Notre Dame,

USA. He is the author of Thomist Realism and the Linguistic Turn: Towards a More Perfect Form of
Existence (2003).
Erik J. Olsson is Professor and Chair in Theoretical Philosophy at Lund University, Sweden.
His areas of research include epistemology, philosophical logic, pragmatism, and, more recently,
philosophy of the Internet. Olsson has published extensively on epistemic coherence, the value
of knowledge, the logic of belief revision, and social epistemology. Recent books include
Against Coherence: Truth, Probability, and Justification (2005), Knowledge and Inquiry: Essays on the
Pragmatism of Isaac Levi (2006), and Belief Revision Meets Philosophy of Science (2011).
Trevor Perri earned a PhD in philosophy from the University of Leuven, Belgium, in 2013.
He has published articles on phenomenology and Bergson’s philosophy and currently works in
the manuscript editing department at the University of Chicago Press.
Denis Perrin is Maître de conférences at the University Grenoble Alpes, France, where he
teaches logic and analytic philosophy. He has published papers on the question of memory
and mental time travel and edited a special issue of the Review of Philosophy and Psychology on
episodic memory.
xviii
Contributors
Valentina Ricci is Mellon Humanities Faculty Fellow at the University of California, USA.
Her current research focuses on the ontology and ethics of violence. She is the editor, with
Federico Sanguinetti, of the volume Hegel on Recollection (2013).
Sarah K. Robins is Assistant Professor of Philosophy at the University of Kansas (KU),

USA, and an affiliate of the Cognitive and Brain Sciences PhD Program in KU’s Psychology
Department. She earned her PhD from the Philosophy-Neuroscience-Psychology Program at
Washington University in St Louis, USA. Her research interests are at the intersection of phi-
losophy and psychology. Her primary focus is on memory – specifically, she’s interested in
explaining the many kinds of memory error uncovered by contemporary memory science and
how they are best integrated with philosophical theories of remembering.
Martin Schwab is Professor of Philosophy at the University of California, USA. He special-

izes in European Continental Philosophy in the nineteenth and twentieth centuries. He has
published monographs on speech acts and Samuel Beckett and has written articles on Nietzsche,
phenomenology, deconstruction, film, painting, and literature. Presently he is working on a
Nietzsche book.
Thomas D. Senor (PhD, Arizona) is Professor of Philosophy and Director of Religious

Studies at the University of Arkansas, USA. Most of his publications are in the areas of episte-
mology and philosophy of religion. Senor was the 2014–15 Plantinga Fellow at the University
of Notre Dame.
John Sutton is Professor and Deputy Director of the Department of Cognitive Science at
Macquarie University, Australia. He is author of Philosophy and Memory Traces: Descartes to
Connectionism (1998), and co-editor of Descartes’ Natural Philosophy (2000) and Embodied
Cognition and Shakespeare’s Theatre: The Early Modern Body-Mind (2014). His current research
addresses autobiographical memory and shared remembering, skilled movement and embodied
cognition, and cognitive history.
Fabrice Teroni is Associate Professor in Philosophy at the University of Geneva, Switzerland,

and project leader at CISA, the Swiss Centre for Affective Sciences. He works in the philoso-
phy of mind and epistemology. His background is in the philosophy of memory, of percep-
tion and of affective states. He has published several articles and monographs on the general
theory of emotions (The Emotions: A Philosophical Introduction, 2012), on the nature of shame
(In Defense of Shame: The Faces of an Emotion, 2011) and on memory. He is also interested in the
nature of our reactions to fiction, and in the way the self is manifest in affective states, as well
as in epistemological issues related to emotions and memory.
Brady Wagoner is a professor and Director of the MA and PhD programmes in Cultural
Psychology at Aalborg University, Denmark. He received his PhD from the University of
Cambridge, where he was also co-founder of the F. C. Bartlett Internet Archive. His research
focuses on the cultural and constructive dimensions of the mind, in particular in relation to
memory and cultural diffusion. Some of his edited volumes include Symbolic Transformation
(2010), Development as a Social Process (2013), and Integrating Experiences (2015). His most recent
books are The Constructive Mind: Frederic Bartlett’s Psychology in Reconstruction (2017) and The
Oxford Handbook of Culture and Memory (2017).
xix
Contributors
Markus Werning is Professor of Philosophy of Language and Cognition at the Ruhr

University of Bochum, Germany, and co-chair of the Mercator Research Group Structures
of Memory. He previously held academic positions at the universities of Erfurt and Düsseldorf
and was research scholar at Rutgers University (RUCCS), USA. Aside from the philosophy
of language and mind as well as epistemology, his research interests address the interface
between philosophy and neuroscience. He has published extensively on episodic memory,
the possibility of introspection, epistemological reliabilism, semantic compositionality, lexical
decomposition, the semantics-pragmatics interface, and the neural realization of meaning.
xx
ACKNOWLEDGEMENTS
We wish to thank all chapter authors for their contributions, as well as Adam Johnson at
Routledge for his guidance during the preparation of the book. Thanks also to Chloe Wall and
André Sant’Anna for their assistance with the final preparation of the manuscript.
xxi
EDITORS’
INTRODUCTION
The philosophy of memory today
Why a philosophy of memory handbook? Why now?

Memory is a fundamental cognitive capacity and as such interacts with virtually all other basic
cognitive capacities. Given its centrality to the mind, it is surprising neither that theorizing
about memory is as old as philosophy itself nor that memory continues to be an active area
of philosophical research. Important early ideas about memory were developed by Plato and
Aristotle, as well as in the Chinese and Indian philosophical traditions. In the early modern
period, key ideas were developed by figures such as Hume, Reid, and Locke. More recently,
continental philosophers have made a number of valuable contributions, sometimes drawing
on psychoanalytic insights. In the contemporary analytic tradition, research on memory is the-
matically oriented, clustering around a number of topics in philosophy of mind, epistemology,
and ethics.
Despite this long tradition of inquiry, the philosophy of memory was until recently not
recognized as an area of research in its own right. In recent years, however, the situation has
changed markedly, and an increasing number of philosophers now count themselves as specialists
in or active contributors to the philosophy of memory. The philosophy of memory is now well
on its way to taking form as a distinct, coherent area of research, with a recognized set of prob-
lematics and theories. Many of the questions that have driven this development stem from a new
interdisciplinarity, and philosophers of memory have often interacted as closely with colleagues
in other disciplines, particularly psychology, as they have with other philosophers. Crucially,
philosophers of memory, particularly those working in the analytic tradition, which is the focus
of the present volume, increasingly recognize that they have as much to say to each other as they
do to colleagues in psychology and other disciplines.
Philosophers of memory, in other words, increasingly think of themselves as philosophers
of memory, and the area is in the process of developing its own infrastructure, as books, special
issues, conferences, and workshops on all aspects of the philosophy of memory become regular
occurrences. The aim of this handbook is to build on and contribute to this trend by providing
a critical piece of infrastructure for the field: a comprehensive overview of the key concepts,
debates, theories, and figures in the philosophy of memory. The handbook is designed to be a
comprehensive reference work, accessible both to researchers and advanced students, that will
be of use to the field for many years to come.
1
The present handbook

The handbook consists of eight thematically oriented parts and one part on the history of phi-
losophy of memory.
• The nature of memory. This part covers highly general issues in everyday and scientific think-
ing about memory. For example, while we tend to take it for granted that memory is a
unitary phenomenon, reflection on the variety of things that we can remember – things
as different as facts, events, skills, to start with – suggests that it may in fact be irreducibly
multiple, a suggestion which receives some support from current psychology and neuro-
science. Is memory fundamentally a unified capacity, or is the term ambiguous among a
number of essentially distinct cognitive capacities? The chapters in this part deal with this
question, the phenomenology of memory, and memory and levels of scientific explanation.
• The metaphysics of memory. This part deals with core questions about what memory is. Does
memory imply truth? What is it for someone to remember something? Does memory
presuppose a causal connection with the past? Does it necessarily involve stored traces
originating in past experience? When we remember, do we stand in cognitive contact with
the past itself or only with internal representations of the past?
• Memory, mind, and meaning. Remembering is intimately bound up with a wide range of
other mental phenomena. Consider imagination: the ability to imagine possible events
clearly depends on our ability to remember past events (with past events providing the
raw materials for imagined events). Recent research, however, suggests that memory itself
might be best understood as a form of imagination. The chapters in this part provide a
survey of these connections, looking at memory in relation to consciousness, perspective,
imagination, images, and emotion.
• Memory and the self. The connection between memory and the self has long been appre-
ciated, with memory providing one of the standard answers to the puzzle of personal
identity. In recent years, other connections between memory and the self have come to the
fore, including the role of memory in constituting the psychological self. The chapters in
this part look both at the traditional question of memory and personal identity and at the
relationships between memory and self-consciousness and memory and narrativity.
• Memory and time. An adequate understanding of memory presupposes an understanding of
its relationship to time. The chapters in this part look at a number of connections between
memory and time, including memory and the concept of time, memory and the metaphys-
ics of time, and the idea, prominent in current psychology and playing an increasing role
in philosophy, that memory for past events amounts to mental time travel, an imaginative
process in which the agent projects himself into the past, much as he projects himself into
the future when imagining future events.
• The social dimension of memory. Social influences on individual memory and remembering
as a social phenomenon are key themes of recent research on memory. What is the role of
memory in constituting collective identities (for example, a nation’s memory for its past)?
What impact do cultural practices of remembering have on the shape of the individual’s
memory? What is the relationship between internal memory and external “memory”? The
chapters in this part draw on philosophical and interdisciplinary resources to survey answers
to these and related questions.
• The epistemology of memory. Epistemologists recognize that memory is a core epistemic
source: without memory, we would be deprived of nearly all of our knowledge, both of
the past and of things in general. But different epistemological theories (externalist theories,
2
Editors’ introduction
as well as internalist theories such as foundationalism and coherentism) account for mem-
ory knowledge in different ways, and all theories must deal with certainly highly general
questions about the nature of memory knowledge: is memory capable of generating new
knowledge, or does it merely preserve existing knowledge? Given that we can’t rule out
the possibility that memory systematically misleads us about the past, can we really claim
to know the past?
• Memory and morality. Given its centrality to our mental lives, memory is bound to have an
important ethical dimension. There are a number of emerging technologies that promise
either to enhance memory (allowing superior recall) or to selectively inhibit it (allowing,
for example, the forgetting of traumatic experiences). The chapters in this part will survey
the thorny ethical questions raised by these technologies, as well as looking at more general
questions: might we have a duty to remember (or to forget) certain people, events, or facts?
• History of philosophy of memory. The final part of the volume is devoted to the history of the
philosophy of memory, with chapters on figures and currents including Plato, Aristotle,
classical Indian philosophy, Indian Buddhist philosophy, Chinese Buddhist philosophy,
Augustine, Avicenna and Averroes, Aquinas, Locke, Reid, Hume, Hegel, Freud, Lacan,
Bergson, Russell, Halbwachs, Bartlett, Wittgenstein, Heidegger, and Ricoeur.
Reflecting the diversity of topics that fall under the heading of philosophy of memory, this
handbook is long, with 48 chapters (not including this introduction). It could, however, easily
have been longer. There are, for example, active programmes of philosophical research on habit
memory, nondeclarative memory, working memory, and memory in nonhuman animals, to
give but a few examples of topics that, for one reason or another, could not be given chapters of
their own here. If the philosophy of memory continues to grow and develop at its present rate,
it will not be long before a new edition of this handbook is called for. Hopefully, these topics –
and others that have yet to emerge – will be included there.
3
PART I
The nature of memory

1
TAXONOMY AND UNITY
OF MEMORY
1. Introductory linguistic considerations

In the most general way of speaking, people use the noun “memory” to refer to instances
where information of the past is made available for present purposes. In this minimal sense the
rings of trees are memories of the climatic conditions in the seasonal succession of years during
certain periods of the past. The characteristic features of tree rings make available this informa-
tion for the present purposes of dendrologists. Likewise, hieroglyphs inside the Cheops pyramid
make information about political events in the lifetime of the pharaoh Cheops available to
Egyptologists and can justly be called memories of that time. Making available information of
the past for present purposes is also the function of certain psychological states of humans and
animals that we refer to by the noun “memory.” For the psychological domain, in English, we
also have the verb “remember” along with the verbs “recall” and “recollect” (as well as less fre-
quently used or more remotely related verbs like “reminisce,” “memorize,” “commemorate,”
“think of”). The usages of the three verbs differ slightly, but—at least, most of the time—refer
to instances of memory. Syntactically speaking, the verb “remember” alone can figure in a
great variety of grammatical constructions taking noun phrases and that-clauses or interroga-
tive (wh-)clauses as well as infinitival and gerundival constructions as dependent arguments:
1 At his wife’s funeral the widower vividly remembered their wedding.

2 The teacher remembers his best pupil well.
3 John remembers his first car whenever he visits his parents.
4 The dog remembered his way home.
5 The history student remembered that Napoleon was exiled in Elba.
6 The client remembered her PIN number.
7 The math professor remembered that 311 is a prime number.
8 The tourist remembered where Caesar besieged Vercingetorix and visited the place.
9 Remember to send in your application by the end of the week!
10 The victim remembered hiding in the basement.
11 The waitress remembered the burglar carrying a gun.
12 At the front door the woman remembers that her latchkey is still in the car.
13 The football coach remembers that the tournament will take place on Labor Day.
7
14 Standing in front of the locked entrance door, the customer remembered that the bank
closes at 2 o’clock every Friday.
15 In the English language tutorial the student remembered that “feline” means cat.
16 After the stroke the patient did not remember how to ride a bike.
Semantically speaking, there seems to be little restriction on the argument of “remember”: if it

is a noun phrase, it can refer to an event (as in 1 and 10), a person (2, 11), a concrete object (3),
or an abstract object (4, 6). The argument may also denote a complete proposition (5, 7, 12, 13,
14, 15; also 9 with the infinitival construction resulting from syntactic raising), or an incomplete
proposition (8; also 16). The proposition often pertains to a particular event or object in the past
(5, 8), but can also be about something particular in the present (12) or future (13). However,
the proposition need not target a concrete particular at all, but may have a general (14) or
abstract (7, 15) content. How-to constructions (16), it has been argued, have at least one reading
where “remember” refers to a procedural capacity without an intentional content fully accessible
to the subject (see also Chapter 8).
In some cases the intentional object of the mental state denoted by “remember” was directly
experienced by the subject and may even be part of his or her autobiography (as in 1). In
other cases (e.g., as in 5 and 8) it was not directly experienced and the memory is indirect.
In some cases, there might be no relation to a particular experience at all (e.g., as in 7). It is,
however, widely agreed that “remember” comes with a presupposition of the factivity of the
intentional object that the state referred to by “remember” is directed to. That is, the existence
of the object or person in question, the occurrence of the event, or the truth of the proposi-
tion is presupposed when the embedding sentence is asserted (see also Chapter 5). Whether
the presupposition of factivity is regarded more as a matter of pragmatics than one of semantics
depends, a.o., on whether one is ready to allow for its subsequent cancellation. The more you
are inclined to regard the following discourse as outright contradictory, the more you are in
favor of the stronger semantic rather than the softer pragmatic view:
17 The waitress remembered that the burglar carried a gun. In fact, he carried only a knife.
Making a presupposition of factivity, the verb “remember” aligns with verbs like “regret,”
“accept” and “forget.” Assertions of
18 The coach remembered/regretted/accepted/forgot that Bayern Munich lost against

Manchester United
all presuppose that Munich indeed lost against Manchester. It is a characteristic feature of
presuppositions as opposed to (semantic) entailments or (pragmatic) implicatures that they are
upheld in negated contexts. Assertions of
19 The coach did not remember/regret/accept/forget that Bayern Munich lost against
Manchester United
also (normally) presuppose the truth of the that-clause. What exactly underlies presupposi-
tions, how widespread they are, under which conditions they can be canceled, and whether
they are a matter of pragmatics or semantics is an ongoing debate in linguistics (Sauerland and
Stateva 2007).
8
Taxonomy and unity of memory
Many philosophers have appealed to grammatical features of the verb “remember” to make
categorical distinctions in the domain of memory. One, for instance, quite frequently finds
authors who base their distinction between episodic and semantic memory on the grammatical
distinction between gerundival (20, 22) and that-clause constructions (21, 23):
20 The victim remembered hiding in the basement.

21 The victim remembered that he hid in the basement.
22 The waitress remembered the burglar carrying a gun.
23 The waitress remembered that the burglar carried a gun.
However, this grammatical variation seems to be rather particular to English and not at all
universal. In a language as closely akin as German, the gerundival construction does not exist
(or is strongly marked) and all cases have to go with the that(dass)-clause construction. It seems
questionable if anything at all can justly be inferred for the categorization of memories from a
grammatical variation that one language offers and another does not.
Other authors (e.g., Bernecker 2010) have proposed to distinguish between propositional
and non-propositional memories depending on whether the psychological state in question
is denoted by the verb “remember” taking a that-clause or taking a noun phrase as argument.
Compare:
24 John remembers that his wife was wearing a blue hat at their wedding.
25 John remembers the blue hat his wife was wearing at their wedding.
26 John remembers his wife wearing a blue hat at their wedding.
According to this proposal, the mental state referred to in (24) should belong to the category
of propositional memories and the one in (25) to the category of non-propositional memories.
However, to which of the two categories does the state referred to in (26) belong? Grammatically,
the gerundival construction is a noun phrase. It should therefore count as non-propositional
memory. Semantically, though, (26) has (exactly or nearly) the same truth-conditions as (24).
How could the mental state denoted in (26) then be categorically different (if at all different) from
the state of propositional memory denoted in (24)? Liefke and Werning (2017) indeed argue that
the fact that certain verbs allow for both that-clauses and noun phrases as arguments does not
license the (type-theoretic) inference that the verb itself is polysemous. In the light of these prob-
lems, basing a taxonomy of memory on grammatical consideration does not seem favorable to us.
We will, therefore, now turn to non-linguistic approaches towards taxonomizing memories.
In the literature, a variety of taxonomies with different taxonomical maxims have been offered.
In this chapter, we will systematize and evaluate the most prominent taxonomical approaches
towards memory along their general structure by distinguishing scalar, hierarchical, and natural-
kind based taxonomies. Scalar taxonomy divides up the various types of memories along a linear
scale: the time span of memory. Hierarchical taxonomy follows the Aristotelian method of definitio
per genus proximum et differentia specifica. This approach has mainly been applied to cases of long-term
memories. Here, memories are first categorized into declarative and non-declarative memories.
In a second step, the category of declarative memories is further subdivided into semantic and
episodic memories. For this hierarchy of categories content-based, phenomenological, and merely
descriptive criteria have been proposed. Natural-kind based taxonomies, in contrast, try to identify
a category with a maximal class whose members are likely to share a set of properties for rel-
evant inductive and explanatory purposes because of some underlying uniform causal mechanism.
9
So far this approach has been developed in a promising way, as we argue, for episodic memory.
We think that it is less promising for other cases of memories.
2. Scalar taxonomy
We would like to begin with as broad a view on memory as possible to provide a context for the
narrower memory phenomena that we will discuss later in this chapter. Initially, we will discuss
phenomena that the uninitiated might not think of as memory, but that would fit under the min-
imal definition given in the beginning of this chapter. Atkinson and Shiffrin (1968) proposed that
human memory could be divided into three classes depending on their persistence in time: ultra-
short-term memory, short-term memory, and long-term memory. However, even though these
three types of memory are identified according to their duration, they in fact differ in several
other aspects as well, that together suggest that they are qualitatively different types of memory.
Ultra-short-term memory, or sensory memory, makes sensory information that does not exist
anymore in the physical world available for processing for less than 1 second (Coltheart 1980).
This kind of memory is specific to a particular sensory system and does not appear to be shared
across sensory systems. In the visual system, ultra-short-term memory is the reason, why, for
example, we perceive apparent, continuous motion when watching a movie instead of 25 indi-
vidual frames per second. To perceive apparent motion, our visual system has to store at least one
previous frame and compare it to the current frame. The offset of an object in two consecutive
images is then interpreted by our perceptual system as motion. Atkinson and Shiffrin suggested
that ultra-short-term memory is limited to only a few hundred milliseconds in duration because
the memory traces decay naturally. The timescale is so short that we are not aware of it as a form
of memory. We also cannot consciously control ultra-short-term memory since it is entirely
driven by external stimuli. For these reasons, some researchers suggest alternatively that ultra-
short-term memory is a component of the sensory system rather than a memory system.
Short-term memory can be used to store information for up to 30 seconds (Cowan 2008). It
is filled with sensory information without a conscious effort, such as when we are remembering
the beginning of a long sentence while listening to a speaker to make sense of the sentence.
However, short-term memory is less dependent on sensory inputs than ultra-short-term mem-
ory and can be controlled consciously. For instance, we can make a mental shopping list for our
groceries while driving on the highway. Most importantly, through covert rehearsal, we can
prevent the spontaneous decay of short-term memories and therefore sustain them for more
than the 30 seconds, after which they would normally have decayed. For instance, we might
hold on to our mental shopping list for minutes by covertly articulating the items repeated.
Experimental studies have shown that short-term memory has a limited capacity, perhaps only
four chunks (Cowan 2001), although the exact capacity remains contentious. Even though the
capacity may appear low, working memory can store more information than this number sug-
gests, because several items can be combined into a single chunk, e.g., remembering a phone
number in three chunks of numbers 234-322-7136 is much easier than remembering the string
of ten digits 2343227136.
Once memories enter the long-term memory store, they are no longer vulnerable to spon-
taneous decay and therefore could endure for a lifetime. Nonetheless, some memories may
be forgotten due to interference from memories that are stored later. Other key differences to
short-term memory are the larger, potentially unlimited, capacity and the storage of arbitrary
combinations of different kinds of information in long-term memory. For instance, when we
recall our last party, we might remember the names of participants, the images of their faces, the
sound of their voices, the smell of their perfumes, and the feel of the host’s sofa. Most importantly,
10
we remember these things not in isolation, but including the context in which they are embedded
and the relationship between them.
In addition to identifying the three different memory types, Atkinson and Shiffrin also
suggested that they stand in a rather fixed relationship to one another. They proposed that
stored information is processed sequentially entering the brain through the sensory system into
ultra-short-term, passes through short-term memory and ends up in long-term memory. This
processing chain can, and frequently does, terminate when information is forgotten from one
of the memory stores.
Over the years, several aspects of Atkinson and Shiffrin’s taxonomy have been questioned.
The main criticisms can be divided into two classes. First, their taxonomy artificially splits
memories of the same kind into different taxa. Several authors have proposed that short-term
memory and long-term memory are instances of the same unitary memory that happen to have
different properties. Second, Atkinson and Shiffrin lump together disparate kinds of memories
into a single taxon. This criticism concerns mostly long-term memory. Despite these contro-
versies, Atkinson and Shiffrin’s taxonomy has been and remains one of the most influential
taxonomies of memory.
3. Hierarchical taxonomy
While a scalar taxonomy is appealing due to its simplicity, a tremendous number of empiri-
cal studies in psychology and neuroscience have revealed that many instances of long-term
memory appear quite distinct from one another. However, it remains highly controversial how
to organize long-term memories into taxa. A frequently suggested approach is the hierarchical
taxonomy according to Squire and Zola-Morgan (1988). In a first step, memory is split into
declarative and non-declarative memory (Figure 1.1). Memories of the former type are those
that we can articulate, while we cannot articulate memories of the latter type. Sometimes these
types of memory are also called explicit and implicit memory. The sentences (1) to (15) above
give examples of declarative memory.
Examples of non-declarative memory include motor skills such as riding a bicycle. We
are aware of our skill and can articulate at a superficial level how we perform those skills,
e.g., “you sit on the saddle, put your feet on the pedals, hold on to the handle bar, and pedal
with your legs.” Anyone who has tried to teach a child to ride a bicycle can attest to the
incompleteness of these instructions. What we are unable to describe are the multitudes of
observations, computations, and manipulations that we carry out while riding the bicycle,
e.g., how we detect deviations and counteract them to keep the balance, how to predict
whether we will collide with another rider who is on a collision course. Everyone who knows
how to ride a bicycle has stored this information, and much more, since she can do the right
thing when required to do so. That is why this type of memory is called motor memory.
However, this information is not consciously accessible to us (see also Chapter 3).
In a second step, two subordinate categories are introduced—first by Tulving (1972)—
within the superordinate category of declarative memory, namely, semantic memory and
episodic memory. Tulving conceived of semantic memory as general knowledge about oneself
and the world, and of episodic memory as memory of personally experienced events. At first
glance, the distinction between episodic and semantic memory seems straightforward, but
research in the subsequent decades revealed that this distinction is difficult to draw. Since this
area of research is the most active and episodic memory is the closest to the everyday notion
of memory, we will focus on episodic memory and its distinction from semantic memory in
the following.
11
memory
declarative non-declarative
episodic semantic motor priming habits . . .

(events) (facts) skills
auto-noetic noetic anoetic

consciousness consciousness consciousness
Figure 1.1 Hierarchical taxonomy of memory according to Squire and Zola-Morgan (1988) and its
relationship to Tulving’s (1985) phenomenological taxonomy.
Content-based taxonomy
Early differentiations between semantic and episodic memory were based on their content.
When Tulving (1972) introduced the distinction between episodic and semantic memory, he
suggested that episodic memories were unique in that they included information about the
what-where-when of an event. Importantly, the three different types of information would
have to be represented jointly in a single memory, not separately in different memories.
Since the what-where-when criterion refers to the content of a memory that can be tested in
non-human animals, it has been frequently employed in animal cognition studies. Memory
of joint what-where-when information of an event has been reported in various avian and
mammalian species.
While what-where-when information is undoubtedly frequently part of the content of
episodic memory, the WWW criterion is insufficient to fully discern episodic from semantic
memory. In some cases, the WWW criterion is too rigid. For instance, in episodic memories
one of the WWW components can be poorly encoded or missing (Bauer et al. 2012; Friedman
1993). In other cases, the WWW criterion is too liberal. For instance, semantic memory of an
event that was not personally experienced may also contain all WWW components. To dis-
tinguish such semantic memories from episodic memories, Clayton and colleagues (2003) have
suggested to add two more conditions: the structural and the flexibility condition. However,
even with these additions Clayton and colleagues acknowledge that the combined criteria
describe episodic-like memory, not proper episodic memory. In summary, even if the WWW
criterion is convenient in nonhuman animal studies and supplemented by other conditions, it
does not appear to appropriately capture the very idea of episodic memory in humans.
Phenomenological taxonomy
Having realized the difficulty in distinguishing episodic from semantic memory on the basis
of content alone, Tulving (1985) instead suggested a criterion based on the type of subjec-
tive experience during retrieval: anoetic, noetic, and autonoetic consciousness. He suggested
that non-declarative memory is associated with anoetic consciousness, meaning that we are
aware of being capable of a certain skill, but not aware of the content of our memory of that
skill. He further suggested that “Semantic memory is characterized by noetic consciousness.
12
Noetic consciousness allows an organism to be aware of, and to cognitively operate on, objects
and events, and relations among objects and events, in the absence of these objects and events”
(Tulving 1985: 3).
Finally, Tulving suggested that autonoetic consciousness was associated with episodic mem-
ory and enables the subject to be aware of having personally experienced the event (see also
Chapter 2). He likened the experience during retrieval of episodic memories to mental time
travel into the past, a reliving of the past. However, the evolutionary purpose of traveling back
in time can only be to inform future behavior. Suddendorf and Corballis (1997), therefore,
suggested that the episodic memory system sustains mental time travel into the future. This
suggestion is supported by experimental studies that show that amnesics have deficits in con-
structing imaginary scenes (Hassabis et al. 2007) and that the hippocampus is activated when
healthy subjects imagine a future event (Weiler et al. 2010).
Suddendorf and Corballis made a strong argument that mental time travel is unique to
humans. Although a number of animal cognition studies have sought to prove them wrong,
Suddendorf and Corballis (2008) are not convinced because of methodological concerns and
because the studied behaviors are limited to ecological niche behaviors, such as food-caching in
scrub jays. Planning for the future in a narrow behavioral context, they argue, is not equivalent
to mental time travel into the future.
Although Suddendorf and Corballis originally proposed the mental time-travel idea to study
foresight, it is arguably the currently most widely used account of episodic memory in humans
(see also Chapter 19). Despite its popularity, we believe that this approach is unsatisfactory at
several levels. At a practical level, mental time travel is difficult, if not impossible, to study in
nonhuman animals (Clayton et al. 2003), since it relies on the subjective experience during
retrieval, which perhaps cannot be shared between different species. While it may turn out in
the end that nonhuman animals do not possess episodic memory, that conclusion would be
much stronger if episodic memory was not construed in such a way as to preclude, or severely
bias against, this possibility. At a theoretical level, it seems unsatisfactory that the nature of a
memory would depend predominantly on the subjective experience during recall, since the
memory persists even when not being recalled. One possible remedy might be to specify that
a memory is episodic memory if an autonoetic recall could be cued (Klein 2013). However,
at a conceptual level, doubts are emerging about the association between different levels of
consciousness and the retrieval of certain memories. For instance, it has been suggested that
consciousness does not always accompany episodic memory retrieval (Hannula and Ranganath
2009; Henke 2010; see also Chapter 9).
Descriptive approaches
An alternative approach to explicating episodic memory is to list the properties of episodic
memory and show that the sum of these properties distinguishes it from other forms of mem-
ory, in particular, semantic memory. This approach, too, has been pioneered by Tulving,
who proposed that semantic memory differs from episodic memory in 28 properties (Tulving
1983). For example, one contrast is: episodic memories are memories of events or episodes
that are organized temporally, whereas semantic memories are memories of facts, ideas, or
concepts that are organized conceptually. This distinction is not as clear as it may seem at first
glance, since information about an event or episode that occurred is information about a fact,
too. Some properties remain vague, such as the temporal organization of the information in
episodic memory—a notion that plays a central role in our analysis below. Other properties are
properties of the neural system supporting memory, rather than of individual memories, such
13
as appearing early or late in development. It is conceivable that the sum of all 28 properties
would clearly distinguish between semantic and episodic memory, but the descriptive approach
faces two fundamental challenges. First, it is difficult to ascertain that the list of properties is
complete. In other words, are all properties listed? Second, it remains unclear which properties
are important characteristics of episodic memory and which ones are inconsequential. In other
words, do all properties have to be on the list?
4. Natural-kind based taxonomy

It has recently become a central topic in the philosophy of psychology to ask whether certain
notions used in psychology correspond to natural kinds and how this might assure that psychol-
ogy has the inductive and explanatory potential we generally expect from sciences (Machery
2009). The identification of natural kinds is neither the principal goal of the scalar and hierar-
chical taxonomies of memory, nor do these taxonomical approaches guarantee that the resulting
classification reflects natural kinds. In the philosophy of science, the dominant view of a natural
kind is due to Boyd (1991) and commonly labeled “the homeostatic property cluster view.”
The core idea is that, in science, entities should be clustered together in a way that (i) optimizes
the inductive and explanatory potential of theories that make reference to those clusters, and
(ii) that this inductive and explanatory potential should rest on uniform causal mechanisms
underlying each cluster (see also Chapter 4). In this spirit, the notion of a natural kind can be
defined as follows:
A class C of entities is a natural kind if and only if there is a large set of properties
that subserve relevant inductive and explanatory purposes such that C is the maximal
class whose members are likely to share these properties because of some uniform
causal mechanism.
The question whether memory, in general, is a natural kind has been addressed by Michaelian
(2010), who argues for a negative answer. Furthermore, Bedford (1997) has argued that implicit
memory is the result of a fallacy and should not be considered a category of memory. On the
positive side, we have argued that episodic memory under a certain explication, indeed, is likely
to be a natural kind (Cheng and Werning 2016; Werning and Cheng 2014). So far episodic
memory, thus, is the only kind of memory for which an explicit case for it being a natural kind
has been made.
In our approach, the explication of episodic memory and its identification with a natural
kind go hand in hand. The claim is that under a specific explication—the Sequence Analysis—
and only under this explication, episodic memory constitutes a natural kind. The approach aims
at a number of desiderata:
(i) The explication needs to clarify what is potentially stored in episodic memory, that is, its
content.
(ii) Despite subjective experiences of recalling detailed episodic memories, numerous experi-
mental studies have consistently found that episodic memory in humans often preserves little
more than the gist of the experienced episode. Therefore, the explication has to integrate two
competing requirements. On the one hand, the memory of an episode E must be allowed to
differ in content (even significantly) from the experience of the grounding episode E′. On the
other hand, one has to enforce a sufficiently stringent relationship between the experiential
base E′ and the mnemonic content E to justify that the memory is based in the experience.
14
(iii) Even though overwhelming evidence indicates that subjects frequently retrieve inac-
curate information when asked to recall episodic memories, these cases are regarded as
improper episodic memory. The aim is an explication of memory that presupposes its
factivity (see above).
The key to fulfilling these desiderata, we believe, is to emphasize the sequential nature of epi-
sodic memory (see also Chapters 17 and 18). Sequentiality is a distinguishing structural feature of
both the content of episodic memory and its underlying neural realization, as has been suggested
by various experimental studies and computation models (for review, see Cheng 2013; Cheng
and Werning 2013; Hasselmo 2012).
According to the Sequence Analysis a subject S has episodic memory with content E at a
time t1 if and only if the following conditions are fulfilled:
(S1) E is an episode with E = e1, . . . , en being a temporal sequence of events e1, . . . , en . E is

called the mnemonic content.
(S2) At some time t1 , S compositionally1 represents E as an episode of temporally succeeding
events e1, . . . , en S ’s representation of E at t1 is called the mnemonic representation.
(S3) At a time t 0 < t1 , S has a reliable experience of the temporally succeeding events e1′ , . . . , e m′ ,
which make up an episode E ′ = e1′ , . . . , e m′ . E′ is called the experiential base.
(S4) The episode E′ occurs at or before t 0 (factivity).
(S5) The mnemonic content E is ontologically grounded in the experiential base E′ in the follow-
ing sense of counterfactual dependence: Were E′ to occur at or before t 0 , E would also
occur at that time.
(S6) S ’s representation with content E at t1 is causally grounded in S ’s experience of E′ through
a reliable memory trace.
(S7) On the basis of its mnemonic representation with content E , S is capable of generating a
temporally explicit simulation with content E at some time t 2 ≥ t1 . The generated simula-
tion is called a mnemonic simulation.
These conditions can be related to the four major stages of memory processing: perception,
encoding, storage, and retrieval. (S3) and (S4) propose conditions on perception, (S5) and (S6)
on encoding, (S1) and (S2) on storage, and (S7) on retrieval.
According to the Sequence Analysis episodic memory is grounded in experience in a twofold
way: with respect to content and to processing. With regard to content, the experienced occur-
rence of the episode E′ secures the occurrence of the remembered episode E . What one has
experienced, in other words, is a truth-maker of what one remembers. The ontological ground-
edness condition warrants a sufficiently strong, but not too strong dependence relation: it does
not require the identity of the mnemonic content with the experiential base, but allows that the
former be just a part or an abstraction of the latter, sometimes not more than its gist (see also
Chapter 26). With regard to processing, the state of memory is causally grounded in the state of
experience through a reliable memory trace (see also Chapters 6, 7, and 23).
The Sequence Analysis does not content itself with the subject having a mnemonic rep-
resentation, but demands that the subject be capable of generating a mnemonic simulation
(see also Chapter 12). In the mnemonic simulation, the temporal succession of events in the
domain of representational contents is represented itself by a temporal succession of events
in the domain of the representational vehicles—in our case neural processes. This can be
regarded a non-symbolic, emulative way of representation (Grush 2004; Mroczko-Wąsowicz
and Werning 2012; Werning 2012).
15
On the basis of empirical observations, it has been argued that the generative or construc-
tive nature of the episodic memory system might be explained by postulating that information
is added during retrieval (Bernecker 2008; Michaelian 2011; Schacter 2012). Cheng,
Werning, and Suddenorf (2016) have argued that mnemonic simulation should be regarded
as a result of scenario construction where (possibly rather sparse) episodic memory traces are
enriched with semantic information. They discuss three options how one could potentially
account, within this framework, for what Tulving (1985) called autonoetic consciousness
(see also Chapters 14, 15, and 16):
(i) Meta-representation: Autonoetic consciousness might simply consist in the presence of a

meta-representation that one recalls the information about an episode as a result of one’s
own experience of that episode (Redshaw 2014; Suddendorf 1999).
(ii) Viewpoint-dependence: Autonoetic consciousness might arise through the point of view a
person assumes in the constructed scenario that includes the episode (Russell and Hanna
2012; see also Chapter 10).
(iii) Quasi-transparent simulation: The phenomenal transparency of an experience (Harman 1990;
Moore 1903), generally speaking, amounts to the property that having the experience of
a scenario is for the experiencing subject just so (i.e., so and nothing in addition to it) as if
the scenario were present (Werning 2010). Even though experiencing a scenario during
remembering is not as phenomenally transparent as a perception of the scenario would be,
characterizations of remembering as “re-experiencing or reliving the past” suggest that
autonoetic consciousness in certain respects resembles phenomenally transparent mental
states (see also Chapter 11).
Once the Sequence Analysis had been introduced as an explication of episodic memory, it
was now possible to argue for the identification of episodic memory with a natural kind. This
argumentation proceeded along three cornerstones. It was first demonstrated that the Sequence
Analysis is both minimal and maximal with regard to its inductive and explanatory potential.
Regarding the minimality, this means that any violation of one of the conditions corresponds
to a deficiency in episodic memory. Examples for deficient cases of episodic memory are false
memories and memories that are experientially ill-founded because there was no grounding
experience or the experience was not reliable. Moreover, a deficiency might consist in the fact
that the memory is not causally linked to the experience by a reliable memory trace—e.g., as
is the case for retroactive inference due to imagination inflation or the misinformation effect
(Marsh et al. 2008)—or in the fact that the experiential base does not secure the mnemonic
content—as in misattribution (Schacter and Dodson 2001). A deficiency might also occur
because the subject is unable to generate a mnemonic simulation at all, as might be the case with
repressed memories or because the subject generates a mnemonic simulation whose content
does not match with that of the mnemonic representation, e.g., when suggestive questions bias
the report of a subject’s memory (Scoboria et al. 2002). Regarding the maximality, it was argued
that other types of memories, most importantly semantic memories do not fulfill the conditions
of the Sequence Analysis.
In a second step, neuro-anatomical and -physiological evidence was used to support the
view that the principal anatomical substrate of episodic memory is the hippocampus. On
the one hand, all processes hosted by the hippocampus contribute to episodic memory. On the
other hand, even though episodic memory involves interactions with other cognitive processes,
which are supported by a variety of brain regions, processes specific to episodic memory are
hosted by the hippocampus.
16
In a final step, which can only be sketched here, it was argued that neural processes in the
hippocampus provide uniform causal mechanisms for the processing stages proposed by the
Sequence Analysis. Specific neural mechanisms—viz. phase precession and theta sequences—
indicate that the hippocampus provides a uniform causal mechanism that aligns the sequential
representation of mnemonic content with the sequential representation of the experiential
base. According to the Sequence Analysis, to form episodic memory in the hippocampus, a
mnemonic representation of the episode E has to be stored in the hippocampus, where E
has to be ontologically grounded in the experienced episode E′. Many authors have suggested
that the hippocampal circuitry is optimized for storing sequences. There is widespread agree-
ment in neuroscience that mnemonic representations are stored in the weights of the synaptic
connections between neurons. More specifically, it has been suggested that the dense recur-
rent network in a specific subarea of the hippocampus (CA3) is well suited to generate neural
sequences (Azizi et al. 2013; Cheng 2013; Levy 1996; Lisman 1999; Wallenstein et al. 1998).
A specific compression mechanism—viz. theta phase precession (Dragoi and Buzsáki 2006)—
generates a representation of the experienced sequence of events at the shorter timescale
required for synaptic plasticity. In the offline state, i.e., during retrieval, populations of neu-
rons fire in a sequence that correlates with the sequence in which they were active at an earlier
time in the online state (Lee and Wilson 2002). Thus, sequential neural activity in the offline
state is a replay of sequential activity in prior experience. Interventions in the memory trace
warrant that mnemonic representations are causally grounded in experiences as is evidenced
by the disruption of systems consolidation. The claim that episodic memory as explicated by
the Sequence Analysis constitutes a natural kind with a uniform underlying neural mechanism
could thus be corroborated.
5. Conclusions
We have begun our reflections on the unity and taxonomy of memory with the minimal
notion of memory as something that makes information about the past available for present
purposes. We have then seen that grammatical considerations about the verb “remember”
might be valuable in itself, for the purposes of semanticists or for heuristic reasons, but are
not conducive to an adequate taxonomization of the various psychological phenomena so
denoted. For inductive and explanatory purposes, sciences must have appropriate terminology
to refer to the phenomena that they study. In recognition of this necessity, many taxonomies
have been proposed in psychology and neuroscience. As we have discussed in this chapter in
regards to memory, these taxonomies are often based on a mixture of observations and intui-
tions about the most important aspects of the phenomena under consideration. However,
these approaches tend to be ad hoc and generally do not follow a principled agenda. We
therefore propose that for the purposes of sciences such as psychology and neuroscience,
identifying natural kinds would be the most fruitful approach where natural kinds should be
taken as homeostatic property clusters. Uniform causal mechanisms explain why the psycho-
logical properties are shared such that the cluster of those properties subserves inductive and
explanatory purposes.
As explicated by the Sequence Analysis, episodic memory is likely to be a natural kind. It
remains an open question whether episodic memory is the only natural kind of memory. Klein
(2014) appears to argue for this position and goes further by proposing that the term “memory”
be used only to refer to episodic memory. His proposal would bring the scientific use of the
word “memory” more in line with everyday language and ensure that memory, thus under-
stood, is a natural kind. However, it might be premature, since the search for other natural kinds
17
of memory is a fairly young endeavor, and suggestions of other natural kinds (Michaelian 2010)
among the memory phenomena should not be dismissed just yet.
Note
1 Compositionality—the principle that the content of a complex representation is a structure-dependent
function of the contents of its parts—is a widely acknowledged, though not uncontentious, criterion for
the adequacy of representational structures in general, be they linguistic, conceptual or neural (Hodges
2001; Werning 2005; Werning, Hinzen, and Machery 2012).
Further reading
Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press.
Cheng, S. and Werning, M. (2016). What is episodic memory if it is a natural kind? Synthese, 193, 1345–85.
Eichenbaum, H. (2011). The cognitive neuroscience of memory: An introduction. Oxford: Oxford University
Press.
Hasselmo, M. E. (2012). How we remember: Brain mechanisms of episodic memory. Cambridge, MA: MIT Press.
Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge,
MA: MIT Press.
Tulving, E. (1983). Elements of episodic memory. Oxford: Clarendon Press.
References
Atkinson, R. C. and Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes.
In K. W. Spence and J. T. Spence (Eds.), The psychology of learning and motivation (Vol. 2, pp. 89–195).
New York: Academic Press.
Azizi, A. H., Wiskott, L., and Cheng, S. (2013). A computational model for preplay in the hippocampus.
Frontiers in Computational Neuroscience, 7, 161.
Bauer, P. J., Doydum, A. O., Pathman, T., Larkina, M., Güler, O. E., and Burch, M. (2012). It’s all about
location, location, location: Children’s memory for the “where” of personally experienced events.
Journal of Experimental Child Psychology, 113, 510–22.
Bedford, F. L. (1997). False categories in cognition: The Not-the-Liver fallacy. Cognition, 64, 231–48.
Bernecker, S. (2008). The metaphysics of memory. New York: Springer.
Bernecker, S. (2010). Memory: A philosophical study. New York: Oxford University Press.
Boyd, R. (1991). Realism, anti-foundationalism and the enthusiasm for natural kinds. Philosophical Studies,
61, 127–48.
Cheng, S. (2013). The CRISP theory of hippocampal function in episodic memory. Frontiers in Neural
Circuits, 7, 88.
Cheng, S. and Werning, M. (2013). Composition and replay of mnemonic sequences: The contributions
of REM and slow-wave sleep to episodic memory. Behavioral and Brain Sciences, 36, 610–11.
Cheng, S. and Werning, M. (2016). What is episodic memory if it is a natural kind? Synthese, 193, 1345–85.
Cheng, S., Werning, M., and Suddendorf, T. (2016). Dissociating memory traces and scenario construc-
tion in mental time travel. Neuroscience and Biobehavioral Reviews, 60, 82–9.
Clayton, N. S., Bussey, T. J., and Dickinson, A. (2003). Can animals recall the past and plan for the future?
Nature Reviews. Neuroscience, 4, 685–91.
Coltheart, M. (1980). Iconic memory and visible persistence. Perception and Psychophysics, 27, 183–228.
Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage
capacity. Behavioral and Brain Sciences, 24(87), 114–85.
Cowan, N. (2008). What are the differences between long-term, short-term, and working memory?
Progress in Brain Research, 169, 323–38.
Dragoi, G. and Buzsáki, G. (2006). Temporal encoding of place sequences by hippocampal cell assemblies.
Neuron, 50, 145–57.
Friedman, W. J. (1993). Memory for the time of past events. Psychological Bulletin, 113, 44–66.
Grush, R. (2004). The emulation theory of representation: Motor control, imagery, and perception.
Behavioral and Brain Sciences, 27, 377–442.
18
Hannula, D. E. and Ranganath, C. (2009). The eyes have it: Hippocampal activity predicts expression of
memory in eye movements. Neuron, 63, 592–9.
Harman, G. (1990). The intrinsic quality of experience. Philosophical Perspectives, 4, 31–51.
Hassabis, D., Kumaran, D., Vann, S. D., and Maguire, E. A. (2007). Patients with hippocampal amnesia
cannot imagine new experiences. Proceedings of the National Academy of Sciences of the United States of
America, 104, 1726–31.
Hasselmo, M. E. (2012). How we remember: Brain mechanisms of episodic memory. Cambridge, MA: MIT Press.
Henke, K. (2010). A model for memory systems based on processing modes rather than consciousness.
Nature Reviews. Neuroscience, 11, 523–32.
Hodges, W. (2001). Formal features of compositionality. Journal of Logic, Language and Information, 10,
7–28.
Klein, S. B. (2013). Making the case that episodic recollection is attributable to operations occurring at
retrieval rather than to content stored in a dedicated subsystem of long-term memory. Frontiers in
Behavioral Neuroscience, 7, 3.
Klein, S. B. (2014). What memory is. Wiley Interdisciplinary Reviews: Cognitive Science. doi:10.1002/
wcs.1333.
Lee, A. K. and Wilson, M. A. (2002). Memory of sequential experience in the hippocampus during slow
wave sleep. Neuron, 36, 1183–94.
Levy, W. B. W. (1996). A sequence predicting CA3 is a flexible associator that learns and uses context to
solve hippocampal-like tasks. Hippocampus, 6, 579–90.
Liefke, K. and Werning, M. (2017). Evidence for single-type semantics: An alternative to e/t-based dual-
type semantics. Journal of Semantics (in revision).
Lisman, J. E. (1999). Relating hippocampal circuitry to function: Recall of memory sequences by recipro-
cal dentate-CA3 interactions. Neuron, 22, 233–42.
Machery, E. (2009). Doing without concepts. Oxford: Oxford University Press.
Marsh, E. J., Eslick, A. N., and Fazio, L. K. (2008). False memories. In J. H. Byrne (Ed.), Learning and
memory: A comprehensive reference (pp. 221–38). Oxford: Academic Press.
Michaelian, K. (2010). Is memory a natural kind? Memory Studies, 4, 170–89.
Michaelian, K. (2011). Generative memory. Philosophical Psychology, 24, 323–42.
Moore, G. E. (1903). The refutation of idealism. Mind, 12, 433–53.
Mroczko-Wąsowicz, A. and Werning, M. (2012). Synesthesia, sensory-motor contingency, and seman-
tic emulation: How swimming style-color synesthesia challenges the traditional view of synesthesia.
Frontiers in Psychology. doi:10.3389/fpsyg.2012.00279.
Redshaw, J. (2014). Does metarepresentation make human mental time travel unique? Wiley Interdisciplinary
Reviews: Cognitive Science, 5, 519–31.
Russell, J. and Hanna, R. (2012). A minimalist approach to the development of episodic memory. Mind
and Language, 27, 29–54.
Sauerland, U. and Stateva, P. (Eds.). (2007). Presupposition and implicature in compositional semantics.
Basingstoke: Palgrave Macmillan. doi:10.1057/9780230210752.
Schacter, D. L. (2012). Constructive memory: Past and future. Dialogues in Clinical Neuroscience, 14, 7–18.
Schacter, D. L. and Dodson, C. S. (2001). Misattribution, false recognition and the sins of memory.
Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 356, 1385–93.
Scoboria, A., Mazzoni, G., Kirsch, I., and Milling, L. S. (2002). Immediate and persisting effects of
misleading questions and hypnosis on memory reports. Journal of Experimental Psychology: Applied, 8,
26–32.
Squire, L. R. and Zola-Morgan, S. (1988). Memory: Brain systems and behavior. Trends in Neurosciences,
11, 170–75.
Suddendorf, T. (1999). The rise of the metamind. In M. C. Corballis and S. E. G. Lea (Eds.), The descent
of mind: Psychological perspectives on hominid evolution (pp. 218–60). London: Oxford University Press.
Suddendorf, T. and Corballis, M. C. (1997). Mental time travel and the evolution of the human mind.
Genetic, Social, and General Psychology Monographs, 123, 133–67.
Suddendorf, T. and Corballis, M. C. (2008). New evidence for animal foresight? Animal Behaviour, 75,
e1–e3.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving and W. Donaldson (Eds.), Organization
of memory (pp. 381–402). New York: Academic Press.
Tulving, E. (1985). Memory and consciousness. Canadian Journal of Psychology, 26, 1–26.
19
Wallenstein, G. V., Eichenbaum, H., and Hasselmo, M. E. (1998). The hippocampus as an associator of
discontiguous events. Trends in Neurosciences, 21, 317–23.
Weiler, J. A., Suchan, B., and Daum, I. (2010). Foreseeing the future: Occurrence probability of imagined
future events modulates hippocampal activation. Hippocampus, 20, 685–90.
Werning, M. (2005). Right and wrong reasons for compositionality. In M. Werning, E. Machery, and
G. Schurz (Eds.), The compositionality of meaning and content (Vol. I, pp. 285–309). Frankfurt: Ontos Verlag.
Werning, M. (2010). Descartes discarded? Introspective self-awareness and the problems of transparency
and compositionality. Consciousness and Cognition, 19, 751–61.
Werning, M. (2012). Non-symbolic compositional representation and its neuronal foundation: Towards
an emulative semantics. In M. Werning, W. Hinzen, and E. Machery (Eds.), The Oxford handbook of
compositionality (pp. 633–54). Oxford: Oxford University Press.
Werning, M. and Cheng, S. (2014). Is episodic memory a natural kind? – a defense of the Sequence
Analysis. In P. Bello, M. Guarini, M. McShane, and B. Scassellati (Eds.), Proceedings of the 36th Annual
Conference of the Cognitive Science Society (pp. 964–9). Austin, TX: Cognitive Science Society.
Werning, M., Hinzen, W., and Machery, E. (Eds.). (2012). The Oxford handbook of compositionality. Oxford:
Oxford University Press.
20
2
THE PHENOMENOLOGY
OF MEMORY
Fabrice Teroni
The most salient aspect of memory is its role in preserving previously acquired information so
as to make it available for further activities. Anna realizes that something is amiss in a book on
Roman history because she learned and remembers that Caesar was murdered. Max returns to
the party and remembers where he was seated, allowing him to retrieve his lost cell phone. The
fact that information is not gained anew distinguishes memory from perception. The fact that
information is preserved distinguishes memory from imagination. But how do acquisition and
retrieval of information contribute to the phenomenology of memory?
This question cannot receive a simple answer, since memory comes in different varieties.
One may remember that Caesar was murdered (this is often described as semantic memory), a
party one attended last week (episodic or personal memory), or how to play the piano (procedural
memory). These are unlikely to make themselves manifest in the same way. Consider pro-
cedural memory. The phenomenology of procedural memory is that of being engaged in an
activity, viz. that of playing the piano effortlessly. To remember how to play the piano is to
display acquired know-how in the relevant circumstances, and the way it feels to display know-
how differs from the way it feels to fail to do so. In the latter case, attention plays a greater
role and with increased attention comes an increased sense of effort. That being said, the phe-
nomenology of effortless action is distinct from the phenomenology of memory – procedural
memory explains why the action unfolds effortlessly, yet does not make itself manifest to those
who display know-how.
This is why interest in the phenomenology of memory has concentrated on semantic and,
even more intensively, on episodic memory – the guiding idea being that, in these cases at
least, memory contributes to phenomenology. So, what is it like to remember a party one
attended last week or that Caesar was murdered? The exclusive aim of this chapter is to sketch
a map of the phenomenology of memory. Since the types of memory denoted by the labels
“semantic”, “personal” and “episodic” are not always the same and do not align straightfor-
wardly with distinctions at the level of phenomenology, I shall avoid using them in what
follows. What will drive the discussion are rather distinctions and issues that relate directly to
what it is like to remember.1
The chapter is structured as follows. In Section 1, I introduce the contrast between con-
tent (what is remembered) and psychological attitude (remembering). This distinction will be
helpful in disentangling issues in the phenomenology of memory. Section 2 is devoted to the
21
Fabrice Teroni
contribution of memory content to phenomenology, and Section 3 to the contribution of the

attitude of remembering.
1. Attitude vs. content

It is customary to approach mental states with the help of the contrast between content and
attitude. Psychological verbs typically report attitudes, while their complements report the
contents of these attitudes. Consider the following examples: Claire believes that Saturn is a
Roman deity; Fred hopes that Manchester City will win the Premier League; Tess supposes
that the accused is guilty. Believing, hoping and supposing are three distinct attitudes, which
here take different propositions as contents. One may also have different attitudes toward
a given content, e.g. believe, hope, or suppose that Manchester City will win the Premier
League. For present purposes, an attitude can be understood as the way a subject is directed at
something. The content of an attitude is what it is directed at.2
Let me now apply the distinction between content and attitude to memory (Urmson 1967;
Recanati 2007; Matthen 2010). Taken at face value, reports such as “Anna remembers that Caesar
was murdered” and “Max remembers last week’s party” refer to one attitude – remembering –
and the diverse contents that it may take. As regards contents, it seems that memory can have
propositional or nominal contents;3 it is as difficult to provide convincing propositional paraphrases
of “Max remembers last week’s party” than of “Greta sees a tree” or “Spencer loves Kathy”.4
As regards the attitude of remembering, its central features are a relation to previous cognition
(one cognized what one remembers or something closely related to it5), actual judgement (one is
typically disposed to endorse what one remembers) and knowledge.6 Whether this broad-brush
characterization should give way to a more fine-grained specification of distinct types of remem-
bering is a complex issue. For the time being, I shall use the content vs. attitude contrast to
distinguish two groups of questions regarding the phenomenology of memory.
First, one may investigate the impact memory contents have on phenomenology. How does
what one remembers contribute to phenomenology? Central issues here are whether there is a
phenomenology of content exclusive to memory and whether we can explain the phenomeno-
logical differences between perceiving, imagining and remembering via the contents of these
attitudes. These are the topics of Section 2.
Second, one may investigate the impact the attitude of remembering has on phenomenology.
How does remembering itself, as opposed to what is remembered, contribute to phenomenology?
Do the relations to past awareness and belief make themselves manifest in consciousness? These
issues are addressed in Section 3.
2. Phenomenology of content
Two types of content

I shall explore how the content of memory contributes to phenomenology by focusing on two
kinds of memory contents, which I shall respectively call experiential and non-experiential con-
tents.7 Since it is difficult to characterize these contents without begging debated questions,8 the
best procedure given our interest in phenomenology is to give illustrations.
When Anna remembers that Caesar was murdered, it is typically wrong to claim that it is for
her “as if she was seeing or hearing” Caesar’s assassination. From her own perspective, it is not
as if she was watching the ageing general stabbed to death by a group of toga-wearing men and
listening to his pathetic address to Brutus. The same is true when we remember mathematical
22
The phenomenology of memory
and semantic facts. In these cases, I shall speak of memory as having non-experiential content.
The traditional idea of “images” fails to get a grip here (on memory images, see Chapter 12).9
It is more convincing to refer to images in describing other cases of memory, e.g. when
Max remembers last week’s party or the first movement of a symphony.10 From Max’s per-
spective, what is happening resembles what happened when he perceived the party or heard
the symphony. This sanctions “as if” language: it is for the subject as if he was seeing, hearing,
etc. the relevant objects or events again. More specifically, these contents of memory resem-
ble those of perception11 in the following respects: which properties – forms, colours, pitches,
etc. – feature in the content of memory is a function of past perceptual content; both contents
are structured around an origin from which things are presented; they are perspectival, objects
being presented within a structurally similar field, in which they occlude each other in identi-
cal ways as a function of their respective positions.12 I shall call these contents experiential. The
fact that these contents resemble those of perception does not mean that they are the same,
and we can usually tell straight off from the way objects make themselves manifest whether
they are perceived or remembered.13 This raises the question of the relation experiential con-
tents bear to perceptual contents. Before turning to this, let me examine the contribution of
non-experiential contents to phenomenology.
Non-experiential contents
The simplest non-experiential contents of memory include no reference to the subject’s past –
as when Anna remembers that Caesar was murdered or that eight times eight is sixty-four. Do
they contribute in any way to what it feels like to remember? The answer depends on one’s
opinion regarding the existence of cognitive phenomenology.14 Since I cannot go into this
debate here, I shall make only one observation. Even if there was a cognitive phenomenology
to entertaining the contents under discussion, this phenomenology would not be exclusive to
memory. Given that one may as well believe, imagine, or entertain these contents, their phe-
nomenology would actually spread over most mental states. The reason is that these contents
are disconnected from the central aspect of memory, which consists in making previously
acquired information available.
In order to convey the idea that the preserving role of memory is sometimes phenomeno-
logically salient, many scholars use the metaphors of “mental time travel” and “temporal
decentring” (on mental time travel, see Chapter 9). The point of the metaphors is not simply to
emphasize that memory carries information about the past, or even that it preserves previously
acquired information. It is rather to emphasize that, in memory, we are sometimes aware that we
do not acquire information in the way we acquire it in perception. The main challenge posed
by memory phenomenology is to explain in non-metaphorical terms how previously acquired
information makes itself manifest as such. Different attempts will be made in what follows.
At this juncture, let us consider accounts in terms of more complex non-experiential contents
that include, as opposed to the simpler contents discussed above, a reference to the origin of
available information. These accounts differ regarding how non-experiential content should be
enriched so as to capture the phenomenology of previously acquired information. Here are two
representative accounts.
According to the first, enrichment consists in a specific thought about one’s past experience.
Previously acquired information makes itself manifest when memory content includes the
relation between available information and a past experience (Owens 1996; Perner 2000;
Tulving 1985). This is the case when Anna remembers that [Caesar was murdered and this
information is available to me because I learned this at school].
23
Fabrice Teroni
According to the second account, enrichment consists in a specific thought about the kind of
process explaining why information is available (Fernández 2008). There is a phenomenology of
previously acquired information when Anna remembers that [Caesar was murdered and the
representation of this event is available because it stands at the end of a specific causal chain].
Although they are popular, these accounts face substantial problems. First, they do not explain
why the time travel metaphor feels right. One may after all realize that a piece of information is
available because of a past learning event or that it is at the end of a causal chain without feel-
ing as if travelling into the past. Such accounts imply that the preserving role of memory makes
itself manifest in the phenomenology of these cognitive contents. Since this phenomenology is
quite elusive, this makes it a poor candidate to meet the present explanatory need. Second, these
accounts are demanding: they require the subject to deploy a rich theory of mind. Previously
acquired information would make itself manifest only to subjects capable of understanding that
mental states stand in explanatory relations to one another or that memory is underscored by
causal chains. As a result, memory would not constitute a privileged source for understanding
pastness. If previously acquired information makes itself manifest in memory only conditional
on such complex thoughts, this suggests that memory is a by-product of such thoughts rather
than an original source for them. Third, these accounts presuppose a simpler access to previously
acquired information. Representing a relation between available information and a past learning
event, or the fact that this learning event is at the end of a causal chain, depends on access to that
very event, which is often provided by experiential contents. So, to capture the phenomenology
of memory, we should switch focus. We should concentrate on these simpler mnesic phenomena
rather than on complex cognitive elaborations of them (Hoerl 2001).
Experiential contents
It looks indeed more promising to elucidate the metaphors of time travel and decentring in
terms of experiential contents. After all, as Holland puts it, the initial model “is of the mind gaz-
ing into the past and picking out features of the landscape there: looking back across an expanse
of time, analogously with the way we see across an intervening physical space” (1954: 483). So,
it is reasonable to assume that the metaphors get a grip when it is, from the subject’s perspective,
as if she perceived again. But how does the contribution of experiential content to phenom-
enology compare to that of perceptual content? Does experiential content reveal something
exclusive about how it feels to remember? Let me take these issues in this order.
Experiential content appears to occupy “a halfway house”:15 it contrasts with perceptual con-
tent, yet resembles it in many phenomenologically salient respects. Can we elucidate the nature
of experiential content and recruit it to explain the differences between perceiving and remembering
something? Since the relevant accounts have been primarily developed for imagination, I shall
examine memory together with imagination in what follows.
The classical attempt is David Hume’s, according to whom the difference between ideas
and impressions – which more or less corresponds to that between perceptual and experiential
contents – is one of degree.16 Experiential contents are “faint copies” of perceptual experiences
with a lower “degree of force or vivacity”. Much ink has been spilled trying to make sense of
this idea, and the consensus today is against Hume’s account. In particular, it is often rejected
for its failing to do justice to phenomenology. Hume’s proposal that we understand how it feels
to remember an event in terms of how it feels to perceive a copy or image of it indeed raises worries.17
First, the account predicts that remembering an event is more similar to seeing it in dim light
than seeing it in plain sunshine, a claim Byrne rightly finds unconvincing (2010: 18). Second,
remembering an event is not like perceiving an image of it: it is not as if we were aware of
24
something that stands in for the event. In this respect, remembering contrasts with the perception
of images, which make themselves manifest as intermediaries.
For these reasons, contemporary approaches try to avoid reference to images in accounting
for experiential contents. Instead, they appeal to a specific relation between what perceptual and
experiential contents respectively represent. This seems like a way forward. Consider two visual
experiences, one representing a red circle on a dark background, the other a red square on a
light background – these experiences resemble one another insofar as they represent redness.
Similarly, experiential contents strike us as being (dis-)similar as a function of the properties they
represent. Why not generalize so as to encompass additional phenomenological (dis-)similarities,
and in particular those between perceptual and experiential contents?
As a start, observe that accounts of the contrast between perceptual and experiential contents
should confine themselves to properties that can be included in these contents. This means that
they have less room to manoeuvre than the accounts in terms of non-experiential contents
discussed above. Consider the property that a piece of information might bear: the property
of originating in a past event or, more generally, the property of being past. Do experiential
contents include such properties? If properties enter into these contents on the condition that
they were or could be represented in perception, the answer is negative: one cannot perceive
pastness or the causal origin of one’s experience.18 In addition, it is difficult to get one’s mind
around the idea that experiential contents represent these properties: is the idea that they come
with a date stamped on them or with a subtitle stating that they originate in a given experience?
Given this constraint, we can distinguish two contemporary accounts of the contrast
between perceptual and experiential contents. According to the first account, it corresponds
to the contrast between determinable and determinate properties (Byrne 2010). Being red is
a determinable property of which being burgundy, crimson and Indian red are determinates.
Now, observe that a surface cannot be red tout court – it is always of a (more) determinate shade.
By contrast, the representation of something as red (“There is a red towel on the bed”) does not
entail the representation of a determinate shade. The first account champions an understanding
of the contrast under discussion in terms of this feature of representations: it claims that experi-
ential contents are less determinate versions of perceptual contents.19
The main worry here is this. Given that the representational power of peripheral vision is
very limited as compared to that of central vision (e.g. Wassle et al. 1989), the account implies
that the phenomenology of experiential memory (and imagination) is that of peripheral vision.
Memory would make its object manifest in the way it is manifest when indistinctly perceived.
This should be challenged: it goes against the fact that, in memory, it does not seem to us that
information is acquired in the way it is acquired in perception. This fact is not acknowledged if
we assimilate experiential contents to indeterminate perceptual contents.20
The second account explains the contrast between perceptual and experiential content by
claiming that the latter represents the former: while the content of experiential memory includes
a past perceptual experience, the content of imagination includes a possible perceptual experience.
Following Peacocke (1985), Martin (2001, 2002) defends this Dependency View and contends that
the distinctive phenomenology of experiential content is explained by its including a percep-
tual experience. Some of the properties of representations, one may insist, are explained by the
properties of what they represent. For instance, many spatial properties of a map are explained
by the spatial properties of what is mapped. This may suggest that experiential contents occupy
a “halfway house” because they represent perceptual experiences. The phenomenology of a
given perceptual content would explain the phenomenology of the experiential content that
represents it in the same way as the spatial properties of a represented region explain the spatial
properties of a map.
25
Fabrice Teroni
There are two related reasons to doubt that this can be made to work. First, the expla-
nation at hand does not hold for all representations – a detailed linguistic description of a
region possesses none of its spatial properties. In light of this objection, the account might be
toned down to say that experiential contents represent perceptual experiences in a format that
preserves part of the phenomenology. The problem with this move is that it abandons the
explanatory ambitions and makes the account indistinguishable from the main alternative to
the Dependency View, the so-called Simple or Similar Content View (Williams 1966; Noordhof
2002). According to this view, the phenomenological similarities between perceptual and
experiential contents only support the claim that perception, experiential memory and imagi-
nation have similar contents. These similarities do not support the more ambitious claim that
the contents of experiential memory and imagination include past or possible experiences.
The failure to explain these similarities by including such experiences into experiential content
appears to confirm this diagnosis.
This is perhaps too quick. There may still be a difference between experiential contents that
represent a scene from a perspective and more complex experiential contents that represent the perspec-
tive as being occupied.21 If so, we may recruit the Dependency View to account for this difference
within the sphere of experiential content. The second objection is that drawing such a contrast
between two types of experiential contents is not faithful to phenomenology. Experiential con-
tent is characterized by transparency (Harman 1990): when we remember or imagine, no past or
possible experience pops up – none makes itself manifest as mediating awareness. Moreover, it is
difficult to understand how perception, insofar as it is transparent, could make itself – as opposed
to what it is about – manifest in experiential content. Something can feature in the latter, you will
remember, on the condition that it was or could be perceived.22
If this is on the right track, then the Dependency View doesn’t explain the phenomenology
of experiential content. We have reached a deflationary conclusion: while a substantial part of
memory phenomenology is traceable to experiential content, its phenomenological similarities
to perceptual content may have to be taken as basic (McGinn 2006).
Exclusive experiential content?

The final issue regarding content is whether the phenomenology of some experiential con-
tents can be exclusive to memory and not shared by imagination. We already put aside a likely
candidate, pastness, maintaining that it is beyond the ambit of experiential content. Moreover,
in discussing the Dependency View, we have just emphasized how difficult it is to draw the
distinction, within experiential content, between representing a scene from a perspective and
representing this perspective as being occupied. This means that appealing to the phenomenology
of the past perception (as opposed to what it is about) is not an option either.
At this juncture, the only promising option is to maintain that experiential memory is distinct
from experiential imagination because it incorporates a phenomenology of particularity (Martin
2001).23 There is indeed something intuitive in the idea that it is not up to us which object
we remember and that this has to do with the dependency of memory on past perception. As
opposed to this, it seems that it is always up to us to fix the identity and degree of particularity
of what we imagine.
That being said, the idea should be handled carefully. Suppose Philip imagines his mother
travelling to Mars. In such a case, experiential content makes the particularity of his mother
manifest to him – it would be a mistake to insist that particularity must be traceable to other
factors, such as the specific project in which the experiential content is embedded.24 Moreover,
experiential memory encompasses memory of objects (Philip remembers his mother) and
26
event-types (June remembers swimming in the pool); manifest particularity in memory does
not here go beyond the manifest particularity available in imagination.
All in all, the claim that a phenomenology of particularity is distinctive of memory is attrac-
tive only if it is restricted to the memory of particular events. The particularity of events we
imagine appears to be always traceable to the project in which experiential content is embedded,
never to experiential content as such. In other words, “as if perceiving” language is sanctioned
in the case of experiential memory of particular events not only because of the aforementioned
similarities between perceptual and experiential contents (which do not tell apart memory from
imagination), but also because of phenomenological event particularity (which does).
Let me recapitulate the foregoing discussion of memory content. We have seen that memory
can take non-experiential and experiential contents. Non-experiential contents are not exclusive
to memory and unlikely to illuminate its phenomenology. Experiential contents have an “as if
perceiving” phenomenology sometimes possessed by memory. When experiential contents are
about particular events, they have a phenomenology distinctive of memory: its being for the
subject as if she perceived particular events again.
3. Phenomenology of the attitude

Discussions of the attitude of remembering have by and large focused on the existence of a
“memory indicator” (Holland 1954), viz. of a phenomenological signature apt to distinguish
remembering from imagining and other psychological attitudes. It is indeed a striking fact that,
despite far-reaching similarities in the content of memory and imagination, we are almost always
capable of telling whether we are remembering. The self-attribution of attitudes under discussion
is supposed to be non-inferential, since a subject’s capacity to tell whether she is remembering or
imagining is typically not based on reasoning.25 This is not to say that the capacity is infallible –
we are prone to error in claiming that we remember – but rather that it does not rest upon the
assessment of evidence. There would indeed be something odd with a subject who could only
self-ascribe attitudes by, say, observing the behavioural consequences of a given content or its
role in her mental life (e.g. O’Brien 2005).26 The immediacy characteristic of the capacity to self-
ascribe memory is a traditional reason for claiming that there is a memory indicator.
Against the memory indicator

This inference can be challenged, however, and I wish to address two lines of thought before
exploring some accounts of the phenomenology of remembering. According to the first line
of thought, the capacity to self-ascribe memory is groundless in the sense that it does not rest on
cues accessible at the personal level (e.g. Naylor 1985; Bernecker 2010). This amounts to saying
that, from the subject’s perspective, she simply finds herself with a brute inclination to state that
she remembers. This is not very convincing. There is a contrast between finding oneself with a
brute inclination to self-ascribe memory and being in a position to give a first-person explana-
tion of this inclination. Advocates of the memory indicator insist, rightly in my opinion, on this
contrast – in the second sort of situation, we often use the expression “I seem to remember”, in
which “seem” takes its phenomenological sense (Audi 1995; Chisholm 1957).27
The second line of thought attacks the idea of a memory indicator by arguing that it
assumes an inappropriate observational model. Urmson (1967) and, more recently, Hoerl
(2001, 2014) emphasize that the first-personal explanation of the capacity to self-ascribe
memory is not supported by observation of happenings in the stream of consciousness. It
would rather be supported by authorship of the relevant decisions: we are in a privileged
27
Fabrice Teroni
position to discriminate remembering from imagining, say, because these attitudes differ as
regards the constraints to which we intend our mental activity to be answerable.28 Anna can
tell straight off that she remembers that Caesar was murdered because she intends her mental
activity to be answerable to what happened in Rome in 44 bc, as well as to conditions in her
past life. She would be imagining if she left her mental activity free from any such constraint.
We should certainly want to avoid a purely passive view of memory. Remembering is
something we often try to achieve and memory can be accompanied by a phenomenology of
mental effort. By contrast, in imagining, mental activity is spontaneous and not directed to the
attainment of an aim.29 Still, Urmson and Hoerl fail in my opinion to give due weight to the
contrast between trying to remember and remembering. Trying to remember is an activity that differs
from imagining in being governed by the aforementioned intentions. Remembering is the goal
of this activity and need not be preceded by it.30 The present line of thought against a memory
indicator trades on an ambiguity between remembering and trying to remember – it should for
this reason be resisted. The memory indicator is a phenomenological signature of the attitude
of remembering and we have been given no reason to deny from the outset that there is one.
Positive accounts of the indicator

Having addressed two lines of thought against the idea of a memory indicator, we are still a
long way from the conclusion that remembering feels a distinctive way. Which are the options?
Accounts of the attitude of remembering can be divided into two groups, depending on whether
the indicator is elucidated in terms of a relation to the will or a feeling. I shall consider them in
this order.
According to the first account, the phenomenological signature of remembering consists in
the awareness of “not making it up”, which would contrast with the awareness of “making it
up” characteristic of imagining. The main obstacle here is the difficulty in understanding what
this awareness amounts to. Suppose we grant that imagining comes with the phenomenology of
agency. Is remembering accompanied by a negative correlate phenomenology? Should we not
more simply claim that remembering is never accompanied by the phenomenology of agency
characteristic of imagining? The simpler alternative is more convincing.31 If so, there is no
phenomenology of agency characteristic of remembering. We should look elsewhere.
Scholars often distinguish various epistemic feelings that are supposed to monitor our cogni-
tive activities (Koriat 2000). I shall concentrate on three feelings that deserve attention because
of their role in contemporary discussions: feelings of knowing, pastness and familiarity. In contrast
to the phenomenology of agency we discussed just above, these feelings relate to memory’s role
in preserving and making information available. For this reason, accounts based on them hold
the promise of revealing something distinctive about the phenomenology of remembering.
There are two interpretations of the expression “feeling of knowing” and neither leads to an
attractive account. First, one may understand the expression as referring to a feeling that often
drives memory searches and “signals to us that the sought-after information is indeed available in
store and worth searching for” (Koriat 2000: 150). Since we have already insisted on the distinc-
tion between trying to remember and remembering, let me give this interpretation short shrift.
As important as it is in supporting and guiding memory searches, the feeling of knowing cannot
constitute the phenomenological signature of the attitude of remembering – this attitude attaches
to contents that are remembered, something that need not result from a memory search.32
Alternatively, one may understand the expression “feeling of knowing” as referring to a felt
confidence or certainty that a content is true that would accompany the attitude of remem-
bering. As a matter of fact, we frequently feel certain that a content is true when we seem to
28
remember it, and we typically endorse it provided we are aware of no defeater. Still, there are
reasons to doubt that felt certainty is what we are after. First, the feeling is plainly not specific to
memory and is as likely to accompany perception, rational intuition, etc. Felt certainty may help
explain how we discriminate remembering from imagining, which never goes with this feeling.
But it will not help in accounting for the capacity to self-attribute memory as opposed to other
attitudes that are accompanied by this feeling. Second, the relation between felt certainty and
seeming to remember is one of explanation rather than identity: one feels certain because one
seems to remember. Another way to make the same point is to observe that memory content is
seldom self-evident and that there must be an explanation of our felt certainty or confidence that
it is true. On pain of concluding that no such explanation is available at the first-person level, we
should explain it by reference to the memory attitude. In sum, the feeling of knowing cannot
account for the phenomenology of remembering.
Next on the list are feelings of pastness and familiarity. In his seminal discussion of mem-
ory, Russell (1921) appeals to feelings of both types – memory contents would feel past and
familiar.33 His suggestion has not proven popular and most scholars nowadays agree with
Byrne’s observation that “while the ‘feeling of familiarity’ is, well, familiar, surely the ‘feeling of
pastness’ is not” (2010: 23). Since I share these misgivings, I shall focus on feelings of familiarity.
Interestingly, these feelings can merge with experiential and non-experiential contents. So, why
not maintain that the preserving role of memory makes itself manifest in a phenomenology of
familiarity characteristic of the attitude? This would not only allow for a cognitively undemand-
ing and unified account, it would also explain typical mistakes of self-attribution, which are
often due to illusions of familiarity. The final verdict should of course await a clear account of
what feelings of familiarity are. I’ll bring this chapter to a close by exploring whether they are
affective.
There are indeed similarities between emotional experiences and feelings of familiarity. First,
both vary in intensity: one may feel more or less afraid of a dog, as a content may feel more or
less familiar. Second, feelings of familiarity depend on a specific type of appraisal. An influential
theory not only claims that appraisals are key to understanding the emotions, but also that the
typical sequence of appraisals in emotions starts by considering whether the stimulus is novel
(Scherer 2001). We might thus insist that a similar appraisal process underscores emotions and
feelings of familiarity; in the latter case, content is appraised as old.
That being said, these similarities do not add up to a strong case for assimilating feelings of
familiarity to emotional experiences, and there may also be basic dissimilarities between them. A
distinctive feature of emotional experience is its valence, which is often cashed out by saying that
emotions feel either good (admiration, joy) or bad (shame, sadness). Do feelings of familiarity
feel good or bad? Second, emotional experience is often claimed to relate to evaluative proper-
ties: fear relates to danger, amusement to the funny, etc. (Deonna and Teroni 2014). If this is
regarded as a defining trait of emotions, it may constitute another reason to think that feelings of
familiarity do not qualify: to describe something as familiar is not to evaluate it. In light of these
considerations, the likely conclusion is that familiarity has a sui generis type of phenomenology.
This may be too quick, however. One may still try to assimilate feelings of familiarity to
emotional experiences. The claim that the feeling of familiarity is not valenced is disputable.
Titchener, for one, describes it as a “glow of warmth, a sense of ownership, a feeling of inti-
macy” (1910: 410), hardly the hallmarks of a neutral experience. A significant body of empirical
data supports this idea.34 In the same spirit, we may insist that feelings of familiarity are subtended
by evaluations, which may indicate “the availability of appropriate knowledge structures to deal
with a current situation” (Winkielman and Cacioppo 2001: 990). These indeed look like posi-
tive evaluations. We may perhaps go even further and maintain, with Frijda, that familiarity and
29
Fabrice Teroni
unfamiliarity have, in and of themselves, different consequences regarding the capacity for cop-
ing (1986: 350). Familiarity would manifest itself in a positive feeling reflecting one’s capacity to
cope with the relevant content, and unfamiliarity in a negative feeling manifesting one’s diffi-
culty in coping with it. Feelings of familiarity may after all turn out to be emotional experiences.
Conclusion
I distinguished two groups of issues in the phenomenology of memory. As regards the con-
tribution of various memory contents, we saw that non-experiential contents are unlikely to
illuminate the phenomenology of memory. Among experiential contents, contents that are
about particular events include a phenomenology distinctive of memory: its being for the sub-
ject as if she perceived particular events again. As regards the contribution of the attitude of
remembering, we concluded that this attitude makes itself manifest in feelings of familiarity,
which may be specific affective experiences.
Acknowledgement
I am grateful to Sven Bernecker, Margherita Arcangeli, Julien Deonna and Richard Dub for
their helpful comments on a previous version of this chapter.
Notes
1 The phenomenology of memory is the way it makes itself manifest from the first-person perspective,
and it is of the essence of phenomenological issues that non-introspective evidence (e.g. regarding neu-
ronal activity during different memory tasks) should be used with caution. I will thus rarely leave the
armchair in what follows.
2 Contents can be as fine-grained as necessary. “Modes of presentation” correspond in the terminology
adopted here to the level of content.
3 Memory is also attributed with the help of wh-clause constructions (“Anna remembers why Caesar was
murdered”, “Max remembers who attended the party”). I shall not discuss such constructions here. For
a convincing case that they are incomplete propositional attributions, see Bernecker (2010: 20–21).
4 For criticism of the idea that all attitudes are propositional, see Montague (2007) and Crane (2009).
5 The relation between the content of memory and that of past cognition is discussed in Bernecker (2010).
6 The issue of whether one must know or have known that p in order to remember that p is debated
(Naylor 1983, 2015; Bernecker 2010).
7 When I speak of experiential and non-experiential memory, this is only as shorthand for the distinction
between these two types of content.
8 One question is how the contrast between experiential and non-experiential content maps onto the
contrasts between, first, propositional vs. non-propositional content and, second, conceptual vs. non-
conceptual content. Let me simply say that, as many examples below testify, reports of experiential
memory are typically non-propositional. There are also reasons to claim that experiential memory can
be non-conceptual (Martin 1992).
9 These observations are not meant to suggest that memory for historical, mathematical, or semantic facts
is never accompanied by memory images – it may for instance be accompanied by images related to the
learning context.
10 For reasons that will emerge below, images have a bad press. In the meantime, reference to them should
simply be read as a way of emphasizing that remembering something resembles perceiving it.
11 Given the purposes of this chapter, it will do no harm to speak of perceptual content.While the claim
that perception has content is debated (Brogaard 2014), the debate concerns theoretically loaded
conceptions of content and not perceptual aboutness as such.
12 The similarity here concerns spatial structure. For that reason, both field memories (i.e. memories that
preserve the subject’s original spatial perspective) and observer memories (i.e. memories that do not)
have experiential contents.
30
13 It would moreover be unfaithful to how we think of memory to describe it as the perception of past
events – a perception does not turn into a memory simply on account of the fact that it is about an
event (e.g. the implosion of a star) that happened long ago (Martin 2001; Matthen 2010).
14 Friends of cognitive phenomenology appeal to a variety of phenomena (understanding, seeing as, etc.)
to support the claim that there is a phenomenology characteristic of conscious thought. Critics deny
that this is the case and try to describe the phenomena without reference to cognitive phenomenology.
For an introduction to the debate, see Bayne and Montague (2011).
15 I borrow this expression from Noordhof (2002), who uses it in a slightly different context.
16 Regarding Hume’s (1739/1985) exact position, which I shall leave aside here, see Owen (2009).
17 Sartre (1940) is a classic examination of Hume’s account, which is nicely set out in Kriegel (2015). See
also McGinn (2006: 7–41) and Chapter 40 in this volume.
18 The point is emphasized in Matthen (2010). Searle’s account of perceptual experience (1983) goes
against these observations, which is why it has been the target of recurrent criticisms (e.g. Bach 2007).
19 This claim applies to each property featuring in experiential content. A distinct (and compatible) claim is
that experiential content remains silent about many aspects of the scene that would be filled in in perception.
20 In his defence of this account, Nanay (2015) emphasizes that the subject is aware that she has to do dif-
ferent things to make the content more determinate in peripheral vision and in imagination. This may
be right. But it is unlikely to explain how the two types of contents differ, as the phenomenon seems
rather to speak to the level of the attitude.
21 In a movie, one may similarly distinguish the representation of a scene from a point of view from its
representation from a point of view occupied by a character (Noordhof 2002).
22 Alternatively, if perception is opaque, we would need to know more about which of its properties
contribute to phenomenology.
23 Schellenberg (2010: 22–3) helpfully distinguishes phenomenological particularity (the fact that “particularity
is in the scope of how things seem to the subject”) from relational particularity (the fact that one is related
to a particular object).
24 The distinction between experiential content and project is made salient by the fact that the experi-
ential content involved in imagining a suitcase and imagining a cat behind a suitcase may be the same
(Noordhof 2002).
25 Self-attribution of memory does sometimes result from reasoning, as when we consider alternative
explanations and conclude “I must have learned this somewhere”.
26 These observations tell against a variety of accounts of memory self-ascriptions along behaviourist and
functionalist lines, which are discussed in Teroni (2014).
27 Confronted with a comparable issue, Campbell (1984) criticizes the reduction of perception to non-
inferential judgements by drawing attention to the contrast between judging because one sees and
having non-inferential judgements pop up in one’s mind. I wish to insist on a similar contrast regarding
the capacity to self-ascribe memory.
28 Contemporary approaches to self-knowledge along these lines (e.g. Moran 2001) descend from
Anscombe’s (1957) seminal discussion.
29 The contrast cannot be pushed too far, as one may try and fail to imagine what something looks like.
An issue that I shall leave aside is how implication of the will relates to the sense of objectivity that
accompanies memory and these exercises of imagination (see O’Shaughnessy 2000: 352).
30 Moreover, it would be unconvincing to maintain that, when remembering occurs without trying, the
capacity to self-ascribe memory traces back to the awareness of being in a state that the relevant activity
could have aimed at.
31 One may think that Martin and Deutscher’s (1966) case of a person who paints a scene from her own
childhood while being convinced that she makes it up tells even against this idea. I cannot go into this
complex issue here. Suffice it to say that, if it did, this would strengthen the conclusion that remembering
cannot be distinguished from imagining by reference to a phenomenology of agency.
32 Ironically, if Koriat is right, scholars like Urmson and Hoerl who insist on the activity of remembering
are too quick in rejecting the observational model. Feelings of knowing may play a central role within
this activity.
33 For a detailed discussion of Russell’s philosophy of memory, see Chapter 44 in this volume.
34 Zajonc (1968) has given some support to the idea that mere exposure towards an object tends to
elicit positive attitudes towards it. More directly relevant to the present issue are Garcia-Marquez and
Mackie’s (2000) conception of the feeling of familiarity as a positive attitude and Winckielman and
Cacioppo’s (2001) data that support the idea that processing facilitation elicits positive affect.
31
Fabrice Teroni
Further reading
Byrne, Alex (2010). Recollection, Perception, Imagination. Philosophical Studies 148 (1): 15–26.
Hoerl, Christoph (2001). The Phenomenology of Episodic Recall. In C. Hoerl and T. McCormark (eds.),
Time and Memory (pp. 315–38). Oxford: Oxford University Press.
Martin, Michael G. F. (2001). Out of the Past: Episodic Recall as Retained Acquaintance. In C. Hoerl and
T. McCormack (eds.), Time and Memory (pp. 257–84). Oxford: Oxford University Press.
Noordhof, Paul (2002). Imagining Objects and Imagining Experiences. Mind and Language 17 (4): 426–55.
Bibliography
Anscombe, G. E. M. (1957). Intention. Oxford: Blackwell.
Audi, Robert (1995). Memorial Justification. Philosophical Topics 23 (1): 31–45.
Bach, Kent (2007). Searle Against the World: How Can Experiences Find Their Objects? In S. Tsohatzidis
(ed.), John Searle’s Philosophy of Language: Force, Meaning, and Mind (pp. 64–78). Cambridge: Cambridge
University Press.
Bayne, Tim and Montague, Michelle (2011). Introduction. In T. Bayne and M. Montague (eds.), Cognitive
Phenomenology (pp. 1–34). New York: Oxford University Press.
Bernecker, Sven (2010). Memory: A Philosophical Study. New York: Oxford University Press.
Brogaard, Berit (ed.) (2014). Does Perception Have Content? New York: Oxford University Press.
Byrne, Alex (2010). Recollection, Perception, Imagination. Philosophical Studies 148 (1): 15–26.
Campbell, Keith (1984). Body and Mind, 2nd edition. Notre Dame, IN: Notre Dame University Press.
Chisholm, Roderick (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press.
Crane, Tim (2009). Is Perception a Propositional Attitude? Philosophical Quarterly 59 (236): 452–69.
Deonna, Julien and Teroni, Fabrice (2014). In What Sense Are Emotions Evaluations? In C. Todd and
S. Roser (eds.), Emotion and Value (pp. 15–31). Oxford: Oxford University Press.
Fernández, Jordi (2008). Memory and Time. Philosophical Studies 141 (3): 333–56.
Frijda, Nico (1986). The Emotions. Cambridge: Cambridge University Press.
Garcia-Marquez, T. and Mackie, D. (2000). The Positive Feeling of Familiarity: Mood as an Information
Processing Regulation Mechanism. In H. Bless and J. Forgas (eds.), The Message Within: The Role of
Subjective Experience in Social Cognition and Behaviour (pp. 240–61). Philadelphia, PA: Psychology Press.
Harman, Gilbert (1990). The Intrinsic Quality of Experience. Philosophical Perspectives 4: 31–52.
Hoerl, Christoph (2001). The Phenomenology of Episodic Recall. In C. Hoerl and T. McCormark (eds.),
Time and Memory (pp. 315–38). Oxford: Oxford University Press.
Hoerl, Christoph (2014). Remembering Events and Remembering Looks. Review of Philosophy and
Psychology 5 (3): 351–72.
Holland, R. F. (1954). The Empiricist Theory of Memory. Mind 63: 464–86.
Hume, David (1739/1985). A Treatise of Human Nature: Being an Attempt to Introduce the Experimental Method
of Reasoning into Moral Subjects. London: Penguin.
Koriat, Asher (2000). The Feeling of Knowing: Some Metatheoretical Implications for Consciousness and
Control. Consciousness and Cognition 9 (2): 149–71.
Kriegel, Uriah (2015). Perception and Imagination. In S. Miguens, G. Preyer and C. Bravo Morando
(eds.), Prereflective Consciousness: Sartre and Contemporary Philosophy of Mind (pp. 245–76). New York:
Routledge.
Lormand, Eric (1996). Nonphenomenal Consciousness. Noûs 30 (2): 242–61.
McGinn, Colin (2006). Mindsight: Image, Dream, Meaning. Cambridge, MA: Harvard University Press.
Martin, C. B. and Deutscher, Max (1966). Remembering. Philosophical Review 75: 161–96.
Martin, Michael G. F. (1992). Perception, Concepts, and Memory. Philosophical Review 101 (4): 745–63.
Martin, Michael G. F. (2001). Out of the Past: Episodic Recall as Retained Acquaintance. In C. Hoerl and
T. McCormack (eds.), Time and Memory (pp. 257–84). Oxford: Oxford University Press.
Martin, Michael G. F. (2002). The Transparency of Experience. Mind and Language 4 (4): 376–425.
Matthen, Mohan (2010). Is Memory Preservation? Philosophical Studies 148 (1): 3–14.
Montague, Michelle (2007). Against Propositionalism. Noûs 41 (3): 503–18.
Moran, Richard (2001). Authority and Estrangement. Princeton, NJ: Princeton University Press.
Nanay, Bence (2015). Perceptual Content and the Content of Mental Imagery. Philosophical Studies 172:
1723–36.
Naylor, Andrew (1983). Justification in Memory Knowledge. Synthese 55 (2): 269–86.
32
Naylor, Andrew (1985). In Defense of a Nontraditional Theory of Memory. Monist 62: 136–50.
Naylor, Andrew (2015). Inferentially Remembering that P. Logos and Episteme 6 (2): 225–30.
Noordhof, Paul (2002). Imagining Objects and Imagining Experiences. Mind and Language 17 (4): 426–55.
O’Brien, Lucy (2005). Self-knowledge, Agency and Force. Philosophy and Phenomenological Research 71 (3):
580–601.
O’Shaughnessy, Brian (2000). Consciousness and the World. Oxford: Oxford University Press.
Owen, David J. (2009). Hume and the Mechanics of Mind: Impressions, Ideas, and Association. In D. Fate
Norton (ed.), The Cambridge Companion to Hume (pp. 70–104). Cambridge: Cambridge University Press.
Owens, David J. (1996). A Lockean Theory of Memory Experience. Philosophy and Phenomenological
Research 56 (2): 319–32.
Peacocke, Christopher (1985). Imagination, Experience, and Possibility. In J. Foster and H. Robinson
(eds.), Essays on Berkeley: A Tercentennial Celebration (pp. 19–35). Oxford: Oxford University Press.
Perner, Josef (2000). Memory and Theory of Mind. In E. Tulving (ed.), The Oxford Handbook of Memory
(pp. 297–312). Oxford: Oxford University Press.
Recanati, François (2007). Perspectival Thought. New York: Oxford University Press.
Russell, Bertrand (1921). The Analysis of Mind. London: Allen and Unwin.
Sartre, Jean-Paul (1940). L’imaginaire. Paris: Gallimard.
Schellenberg, Susanna (2010). The Particularity and Phenomenology of Perceptual Experience. Philosophical
Studies 149: 19–48.
Scherer, Klaus (2001). Appraisal Considered as a Process of Multilevel Sequential Checking. In K. Scherer,
A. Schorr and T. Johnstone (eds.), Appraisal Processes in Emotion: Theory, Methods, Research (pp. 92–120).
New York: Oxford University Press.
Searle, John (1983). Intentionality. New York: Cambridge University Press.
Teroni, Fabrice (2014). The Epistemological Disunity of Memory. In A. Reboul (ed.), Mind, Values and
Metaphysics (pp. 183–202). Dordrecht: Springer.
Titchener, Edward (1910). A Textbook of Psychology. New York: Macmillan.
Tulving, Endel (1985). Memory and Consciousness. Canadian Psychology 26: 1–12.
Tye, Michael and Wright, Briggs (2011). Is There a Phenomenology of Thought? In T. Bayne and
M. Montague (eds.), Cognitive Phenomenology (pp. 326–43). New York: Oxford University Press.
Urmson, J. O. (1967). Memory and Imagination. Mind 76: 83–91.
Wassle, H., Grunert, U., Rohrenbeck, J. and Boycott, B. (1989). Cortical Magnification Factor and the
Ganglion Cell Density of the Primate Retina. Nature 341: 643–6.
Williams, Bernard (1966/2006). Imagination and the Self. In Problems of the Self (pp. 26–45). Cambridge:
Cambridge University Press.
Winkielman, P. and Cacioppo, J. (2001). Mind at Ease Puts a Smile on the Face: Psychological Evidence
that Processing Facilitation Elicits Positive Affect. Journal of Personality and Social Psychology 81 (6):
989–1000.
Zajonc, Robert (1968). Attitudinal Effects of Mere Exposure. Journal of Personality and Social Psychology
Monograph Supplement 9 (2): 1–27.
33
3
MEMORY AND LEVELS OF
SCIENTIFIC EXPLANATION
John Bickle
Maybe at first your memories of home

Won’t resemble mine at all.
But one thing I know we all have in common
Is the God-given gift to recall . . .
“Sing a Family Song,” comp. Merle Haggard (1979)
1. Memory as favorite case study

For at least thirty years memory has been a central case study in philosophical disputes about levels
of explanation in science. In her groundbreaking work Neurophilosophy, Patricia Churchland pro-
vides “a few samples” of memory research “to illustrate the interdisciplinary nature of the research
program” (1986: 368–73). Her survey begins at the “cellular and molecular level” with work on
invertebrate models (sea slugs, fruit flies) and cumulates eventually in human neuropsychology,
neurology, and cognitive psychology. One reason for the popularity of memory as a case study
of scientific levels is results: memory has long been the best understood interdisciplinary phe-
nomenon across scientific levels, especially at “lower” ones. The best reason why is probably a
serendipitous early guess about brain mechanism: plastic synapses. More than 125 years ago, the
great Spanish neuroanatomist Santiago Ramon y Cajal speculated that the small spaces between
neurons that his staining and microscopy techniques revealed seemed an intuitively promising basis
for memory storage. This hypothesis has paid off handsomely, as we’ll see below.
Widespread acceptance of memory research as a case study of levels of scientific explana-
tion has not led to equally widespread agreement about its correct philosophical lessons. “New
mechanist” Carl Craver interprets memory research as expressing a non-reductive “mosaic
unity” for neuroscience (2007: Chapter 7). William Bechtel (2009) sees it as illustrating “mech-
anistic” reduction, as opposed to “intertheoretic” or “ruthless” varieties. Craver and Bechtel
(2007) together insist that memory research illustrates “nested hierarchies of mechanisms within
mechanisms,” like Russian nesting dolls, and that the mechanistic details must be filled in at
all levels for a complete explanation. Craver also uses multi-level memory research to illustrate
his taxonomy of scientific levels and their inter-relations (2007: Chapter 5). Churchland (1986)
sees the many levels of memory research as an instance of intertheoretic reduction in practice
and a challenge to autonomy-of-psychology arguments from mainstream cognitive science.
34
Memory and levels of scientific explanation
Bickle (2003, 2006) sees it as a demonstration of “ruthless” reductionism, of intervening experi-

mentally into increasingly lower levels of biological organization, tracking the behavioral effects
of those interventions, and then claiming a reduction directly to the lowest biological level
successfully intervened. He (2003) also argues that shared cross-species cellular, and especially
molecular and molecular-genetic, details in memory research counter popular empirically based
claims about the multiple realization of memory. Aizawa (2007) demurs and insists that massive
multiple realization pervades even these lower levels of memory research. And so on.
My plan in this chapter is to update readers on many of the important details now widely
accepted at all scientific levels of memory research. While many philosophers know something
about the higher levels (cognitive psychology, neurology, even some cognitive neuroscience
and neuropsychology), philosophical awareness grows sparser as we move further down. This
is despite the huge increase in discoveries here over the last two decades. Details at these
lower levels have been increasingly filled in since many still much-cited surveys by philoso-
phers were published (e.g., Churchland 1986; Craver and Darden 2001). Many philosophers
agree that arguments about levels—what they are levels of, how the components across levels
relate—should answer to actual scientific details. The recently discovered details thus need
wider dissemination.
2. A useful diagram to guide us

Craver and Darden (2001) provide a useful diagram for my project. Figure 3.1 illustrates four
levels of scientific explanations of rodent spatial memory. The top level is the behavioral, the
level explained by psychology and cognitive science. The next level down is the gross neu-
roanatomical and neurofunctional; below that the cellular neurophysiological; and finally the
molecular-neurobiological. Craver and Darden’s diagram appears plausible. Its main shortcom-
ing is that it does not go very far into what is now known at all the levels illustrated.
3. More scientific details: “mouse navigating Morris water maze”

Start with the diagram’s top-most level. Here philosophers and cognitive scientists tend to have
some familiarity with neuroscientific work, so we can be briefer. On the standard hidden-
platform test of the Morris water maze (Morris 1981), the naïve animal is plunged into a tub of
opaque water deep enough to force it to swim. Being naturally hydrophobic, it seeks escape.
Randomly at first, it discovers a submerged platform it can crawl upon to support itself, mostly
out of the water. A variety of distal visual cues on the walls of the room containing the tub are
available. Trials are repeated numerous times, typically over a multiple-day training period. The
submerged platform is in the same location on every training trial but the mouse is plunged
into the tub at different locations. Over a relatively small number of trials (e.g., 6–12, over
2 to 4 days), rodents decrease their time and shorten their routes to the hidden platform. The
animal seemingly has learned, and can recall and use, the location of the platform relative to the
multiple spatial visual cues. Results from “probe” trials run after standard training strengthen this
interpretation. Remove the platform after training and animals spend significantly more time in
the quadrant where the platform was located during training than in any other. They also cross
over the location of the training platform repeatedly. Trained animals exceed untrained animals
significantly in both of these “probe trial” measures.
The Morris water maze has made a huge contribution to the study of memory. But it is
not the only memory paradigm widely used by behavioral neuroscientists. A variety of forms
of conditioning, both classical and operant, are still prominent. One popular conditioning
35
John Bickle
Figure 3.1 Levels of scientific explanation of spatial memory in a rodent model. Reprinted with
permission from Craver and Darden (2001).
procedure with a “cognitive” feel is contextual conditioning. Rodents are placed in a novel
context, allowed to explore it briefly, and then subjected to an affective stimulus in that new
context, such as a moderate to severe foot shock. The animal is removed from the novel envi-
ronment for a delay period, then introduced back into it. The duration of stereotypic rodent
affective behavior is measured, e.g., “freezing” for fear conditioning: total suppression of all
body movement except breathing, hunched posture. Longer durations of affective response
during testing sessions is assumed to indicate stronger context-shock associative memory.
Widely employed, distinctively non-spatial memory tasks for rodents exploit their natural
curiosity for novelty. In the object recognition task (Berlyne 1950), rodents are exposed to a
novel object and permitted to explore it for a specified time. The object is removed for a speci-
fied delay period and then returned, along with a new object. The amount of time the animal
36
explores the new object during this second exposure, relative to the time it spends exploring
both objects, is taken as a measure of the animal’s memory of the previously exposed object: a
larger ratio means stronger object recognition memory for the novel object of the first exposure.
Social recognition memory is measured similarly (Thor and Holloway 1982). The experimental
rodent is exposed to a juvenile male of the same species, with the amount of time the ani-
mal engages the juvenile in stereotypic rodent exploratory behavior measured. The juvenile is
removed from the environment for a specified delay period, then re-introduced. The reduction
in the experimental animal’s exploratory time on the second exposure contrasted with that of
the first exposure, is taken as a measure of the strength of its social recognition memory for
that juvenile conspecific. All of these behavioral paradigms have well-accepted delay periods
for both short- and long-term memory testing. They are also used in reconsolidation after
re-exposure studies, where the experimental animal is re-exposed to the memory stimulus after
the delay period, and then re-tested after a second delay following re-exposure.
What about forms of memory salient in human cognition: episodic, semantic, conscious
autobiographical? For obvious ethical reasons, experimental interventions at the human neu-
ronal and intraneuronal levels are limited (but more on this in the next section), so specifically
human memory research tends to bottom out quickly at higher levels of brain organization:
neuropsychology, neurology, cognitive neuroscience. (See Tulving and Craik 2000 for a good
treatment of specifically human memory.) Nevertheless, human neuropsychology and neuro-
logy continue to fascinate philosophers. Perhaps the most famous case is global amnesia patient
Henry Molaison (1926–2008), better known as ‘H.M.’ In his mid-twenties, suffering from
increasingly severe and then-untreatable grand mal epilepsy, H.M. underwent a radical sur-
gical procedure that removed numerous structures bilaterally in his medial temporal lobes,
including the full hippocampus formation. Upon recovery, his epilepsy was controlled and his
surgery-induced cognitive deficits appeared limited only to memory, but those effects were pro-
found (Scoville and Milner 1957). H.M. seemingly could no longer retain or retrieve any new
post-surgical memory information, especially pertaining to his own life and experiences. His
retrograde memory for information acquired prior to the surgery was mostly intact (compared
to age- and sex-matched nonamnesic controls), except for about a two-year period immediately
prior to the surgery, for which he was also profoundly amnestic. In short, H.M.’s surgery pro-
duced the paradigmatic “anterograde” amnesic.
Yet neuropsychologist Brenda Milner (1958) quickly discovered that H.M.’s anterograde
amnesia was not universal. He acquired and retained new motor skills and perceptual skills,
and was affected by some priming biases as readily as did or were age- and sex-matched
nonamnesic controls. (See Squire 1992 for a comprehensive review.) More controversially,
H.M. matched control performance on a strictly cognitive memory task, The Tower of
Hanoi, a block-moving game with a few rules and a 31-step optimal solution. In one study
(Cohen et al. 1985), H.M. and other anterograde amnesics improved their performances over
several days of training at a rate comparable to age- and sex-matched nonamnesic controls.
However, Xu and Corkin (2001) famously failed to replicate these findings; the amnesics
failed to learn at non-amnesic rates.
One remarkable feature of these preserved memory capacities is that global amnesics cannot
recall having performed the tasks before, even as their performance improves, matching control
rates, over training trials. Typically, they confabulate common explanations when asked about
their later performances, e.g., “I’m just good at puzzles like this.” Their actions reflect their
specific memory impairments: despite improved performances over time and experience, they
must constantly be reminded of the rules or goals of these tasks. This work spawned neuro-
psychological theories of “multiple memory systems” in the human brain, of a “declarative” or
37
John Bickle
“explicit” system which involves the bilateral hippocampi and so is severely compromised in
medial temporal lobe amnesia, and a “procedural” or “implicit” system which is preserved, and
is responsible for the set of preserved memory capacities. ‘Declarative’ and ‘procedural’ were
popularized by Larry Squire and colleagues, see, e.g., Squire (2004); “explicit” and “implicit”
were popularized in contemporary psychology by Daniel Schacter and colleagues, e.g., Schacter
(1987). Controversially, but based on detailed comparisons of experimental results with human
amnesics, and nonhuman primate and rodent animal models, Squire (1992) equated “decla-
rative memory” with mammalian hippocampus-dependent memory, since analogues of the
declarative memory tasks compromised in human medial temporal lobe anmesics were likewise
compromised in animals undergoing increasingly specific decreased hippocampus activity.
4. More scientific details: “hippocampus generating spatial maps”

Craver and Darden’s next level down has been an area of scientific research for sixty years. It
is still thriving: recently it has garnered three researchers the 2014 Nobel Prize for Physiology
or Medicine, provided key results for widespread scientific interest in a new experimental tool,
and verified single-cell findings previously limited to nonhuman primate models, in humans.
Since the days of Case H.M., the hippocampus-memory connection has been part of
widely accepted science. But what is occuring at the neuronal level to realize hippocampus-
dependent memory? The answer took a huge step forward in the 1970s, initially concerning
spatial memory in rodents, with the discovery of hippocampus “place” neurons. An organ-
ism’s own sense of location in its environment, and its capacity to navigate through a complex
changing environment, are important cognitive functions. The idea of map-like representa-
tions of the environment realized in mammalian brain activity was not new. Psychologist
Edward Tolman (1948) had proposed it, based purely on behavioral data from rats searching
complex mazes for food rewards. But while Tolman’s theoretical suggestion challenged the
chains-of-stimuli-response explanations popular among methodological behaviorists of his
time, his purely behavioral methods could not shed light on these maps’ brain locations or
mechanisms.
It was more than two decades after Tolman’s proposal that single-cell recordings in behav-
ing mammals found the answer. In 1971, John O’Keefe and Jonathan Dostrovsky first reported
cells in the CA1 region of the rat hippocampus whose action potential frequencies increased
significantly when the rat was located at a particular place in a familiar bounded environment
(and sometimes only when its body and head were oriented in a particular direction). Five
years of follow-up experiments reported in O’Keefe (1976) detailed response properties of
these neurons. Borrowing the “activity field” concept from previous single-cell sensory and
motor electrophysiology, O’Keefe’s lab discovered these neurons’ “place fields”: the region
of space in a given environment the animal’s occupation of which generates their maximum
action potential frequency. They further discovered that a given neuron’s place field remained
constant over time and further familiarity with the environment. The collection of the place
fields of all CA1 hippocampus place neurons constituted a full map for any given bounded
environment with which the rat was familiar. Place cell activity did not simply reflect activity
projected from sensory neurons; the place fields of individual place cells remained constant for
any single environment even when significant visually available changes were made in it. Place
cells “remapped” for each new environment encountered, and the specific place field of any
one of these neurons in one environment could not predict the specific place field it would
develop in a new environment; but the collection of the place fields developed by these cells
in each environment captured all “places” within it. Even these “remappings” remained stable
38
over time. For these researchers, place cells were the key to hippocampal memory. As far back
as their (1971), O’Keefe and Dostrovsky speculated that the loss of hippocampus place cell func-
tion potentially could explain the full panoply of results of then-known memory disruptions in
medial temporal lobe amnesia.
Further neurophysiological discoveries followed quickly, including head orientation and
movement-monitoring neurons in nearby regions, but important questions remained unan-
swered about the brain’s spatial coordinating mechanisms. Some have been solved recently by
the discovery of “grid cells” in the entorhinal cortex, a region whose neurons project heavily
into the dorsal hippocampus. May-Brit and Edvard Moser began their research by first inves-
tigating projections onto CA1 place cells from both within and outside of the hippocampus.
Entorhinal cortex is a primary afferent to the hippocampus generally, but most of its projec-
tions go directly to neurons in the dentate gyrus, which in turn project to neurons in the
CA3 region, and those to neurons in CA1. But disconnecting the entorhinal cortex-to-dentate
gyrus-to-CA3-to-CA1 pathways did not disrupt CA1 place cell formation, implicating the
much sparser direct entorhinal cortex-to-CA1 connections in the formation of CA1 neuron
place fields. So the Mosers began looking for place cells in the entorhinal cortex; but what they
found was even more intriguing.
A “place-like” entorhinal neuron increased its action potential frequency to maximum when
the rat was in multiple environment locations, not just one; these multi-place activity loca-
tions formed the nodes of a hexagonal grid. Other place-like neurons in the same region of
entorhinal cortex were maximally active with the same spacing and orientation of places on the
grid, but to different hexagonal-arranged spatial locations. Collectively, these neurons covered
every point in the environment, multiple times. Distances between a given cell’s hexagonally
arranged locations varied across entorhinal cortex. Some had their hexagonally arranged firing
locations spaced closer together, others farther apart. The Mosers named these neurons ‘grid
cells’ (Hafting et al. 2005).
Using theoretical modeling, lesion experiments, and remapping experiments, the Moser lab
explored reciprocal influences between entorhinal grid cells, hippocampus place cells, and head
orientation and movement-monitoring (speed, direction) cells in nearby medial temporal regions.
They showed that collectively these neurons organized to form a spatial coordination system,
whose individual neuronal components were understood at the level of their field properties, and
the system’s activities in turn were explained by how these constituents combined into organ-
ized networks. More recent work using both deep brain recordings in human clinical patients
and functional neuroimaging has confirmed the existence of all of these types of neurons in the
analogue structures in the human brain (Jacobs et al. 2013; Doellar et al. 2010). Recognizing the
importance of these discoveries and their general potential for providing explanations of higher
cognitive functions in terms of lower-level neuronal mechanisms, one-half of the 2014 Nobel
Prize for Physiology or Medicine was awarded to John O’Keefe, and the other one-half jointly
to May-Britt Moser and Edvard Moser “for their discoveries of cells that constitute a positioning
system in the brain” (www.nobelprize.org/nobel_prizes/medicine/laureates/2014/).
Results from two other recent scientific endeavors at the neuronal components level of
memory deserve mention. The first stems from work using optogenetics, an experimental
method in existence for barely one decade, and named Nature Methods “Method of the Year” in
2010. This method affords experimenters unprecedented control over the activation or silenc-
ing of specific selected neurons in behaving mammals: literally at the flip of a light switch. The
key is tricking specific neurons to generate their own light-sensitive membrane-bound ion
channel proteins. DNA for microbial opsins from algae is first engineered to be expressible in
mammalian DNA, then inserted into engineered virus vectors. These viruses are microinjected
39
John Bickle
directly into tiny regions of the mammals’ brains. Typically a promoter region is tagged onto
the engineered opsin gene to limit expression only in neurons with sufficient quantities of the
promoter molecule. The engineered gene complex is inserted into infected neurons’ host DNA
by the usual virus replication cycle. Expression and synthesis of the opsin protein proceeds; the
opsin proteins embed in the neurons’ membranes, and when activated by light stimuli open
direct channels for ions to move into the neurons, either excitatory (e.g., sodium, Na+) or
inhibitory (e.g., chloride, Cl-). These ion channels are activated in the freely moving mammal
by switching on a brain-embedded light source (LED, laser) in their vicinity. The effects of
increased or decreased activity in the selected neurons can be assessed behaviorally (Deisseroth
2015; Goshen 2014).
Results from Susumu Tonegawa’s lab using optogenetics for memory research brought the
technique to wider neuroscientific—and popular—attention. These researchers used optoge-
netics to reactivate engrams for specific memory phenomena, which could then be incorporated
into false memories for events which never occurred. The lab uses contextual fear conditioning
as one of its behavioral protocols. Liu and colleagues (2012) used a transgenic mouse line in
which an excitatory opsin gene (channelrhodopsin 2, or ChR2) got expressed only during fear
conditioning to the novel context, in neurons in the dentate gyrus region of the hippocampus.
A promoter region on the engineered opsin gene further limited its expression only to the most
active infected neurons during the training session: the hypothesized engram for the context-
fear memory trace. Photostimulation from a light source embedded in the mice brains near the
affected neurons causes the ChR2 proteins to open direct channels for Na+ influx, to depolarize
only those neurons and hence to reactivate the hypothesized cellular engram for the context or
context-shock association.
Liu and colleagues’ (2012) ingenious experimental design was to reactivate the infected cells
encoding the context-shock association to the first context while the animal was exploring a
second, different novel context. No actual foot shock was administered in that second context.
However, photostimulated experimental mice later demonstrated significant freezing to the sec-
ond, never-actually-shocked, context. A follow-up study by Ramirez and colleagues (2013) in
Tonegawa’s lab also expressed the opsin only in the most active dentate gyrus neurons, but this
time during non-shocked exploration of the first novel context. They reactivated those specific
neurons by photostimulation when the animals underwent context-fear (foot shock) condition-
ing in a second novel context. The experimental mice demonstrated the fear response when
placed back in the first, neutral context, in which they were never shocked.
These experimenters interpret their results straightforwardly: “we created a false memory
in mice by optogenetically manipulating memory engram-bearing cells in the hippocampus”
(Ramirez et al. 2013: 387). These false memory experiments not only brought optogenetics to
wider scientific notice, but also attracted huge public interest; google ‘false memory optogenet-
ics’ to get a quick feel for the number and range of popular media coverage.
Consider one more recent discovery pertaining to hippocampus memory at the cellular-
physiological level, before we descend further down. Four decades ago, single-cell recordings
in nonhuman primate inferotemporal cortex by Charlie Gross and colleagues revealed visual
neurons with face and hand receptive fields; many of these “face” and “hand” cells didn’t fire
unless the silhouette was in a biologically realistic position in the primate’s visual field (Gross
et al. 1972; Desimone et al. 1984). For obvious ethical reasons, single-cell electrophysiology
is of limited application in human brains. But it can be performed when deep electrode place-
ment is justified on purely clinical grounds, such as in cases of pharmacologically unresponsive
epilepsy. Recently, eight patients with depth electrodes implanted to localize seizure onset
enabled experimenters to record from single neurons and multiple-neuron units in their medial
40
temporal lobes (Quiroga et al. 2005). From a total of 993 units recorded over all sessions,
64 single units and 68 multiple units showed a statistically significant response to at least one
photograph of specific people, landmarks, or objects. These responses were highly selective:
only a small percentage (2–3 percent) of photographs evoked any responses from any neurons
from any patients. But the pattern of responses in the ones activated were remarkable. One
single unit in one subject’s left posterior hippocampus responded exclusively to photographs of
actress Jennifer Aniston, in a variety of orientations, which seems to rule out a straightforward
visual interpretation of its activity. It did not respond to a photograph of Aniston together with
actor Brad Pitt (Quiroga et al. 2005: Figure 1). Another single unit, in the right posterior hip-
pocampus of a different patient, responded exclusively to photographs of actress Halle Berry;
but also to drawings of her, to a photograph of her in Catwoman get-up, and to the letter
string spelling out her name (Quiroga et al. 2005: Figure 2). Other single- and multiple-units
responded exclusively to photographs of actress Julia Roberts, basketball great Kobe Bryant,
the Sydney Opera House, the Tower of Pisa, or specific animals or food items, all from vary-
ing visual orientations. In the paper’s Discussion section the authors note the similarity of their
discovered neurons with the well-known place neurons of spatial memory.
5. More scientific details: “neurons inducing long-term potentiation”

Our next step down takes us to components of neurons, specifically to those whose activi-
ties drive the plastic changes that enable specific neurons to participate in particular memory
processes. One kind of activity that has been central to memory research for more than four
decades is long-term potentiation, or LTP, a form of activity-driven enhancement of excitatory
activity in post-synaptic neurons to subsequent pre-synaptic activity.
Functionally (and greatly oversimplifying!), a neuron is specialized to (1) conduct an electric
current down the length of its axon, or output projection, and (2) transmit that activity to all
other cells, neurons or otherwise, with which its axon is connected. Typical connections are
via chemical synapses, tiny spaces separating the terminal bulbs of the pre-synaptic axon and
processes on the post-synaptic cell. On the pre-synaptic side, vesicles containing tiny portions
of the neurotransmitter substance line up in active regions along the cellular membrane forming
the synapse. The effects of the action potential conducted to the terminal bulb initiate fusion of
vesicles with the cell membrane and release of the neurotransmitter substance into the synaptic
cleft. Passive diffusion results in some of these neurotransmitter molecules binding to protein
receptors embedded in the post-synaptic neuron’s membrane.
Post-synaptic receptors are of two general types. Neurotransmitter binding to ionotropic
(classic) receptors simply change their three-dimensional configuration to open a direct chan-
nel to allow specific ions to pass into the post-synaptic neuron. Activated excitatory receptors
permit positively charged ions to pass (e.g., Na+, calcium, Ca++), generating excitatory post-
synaptic potentials (EPSPs) in the membrane regions of the activated receptors, temporarily
changing the electric charge inside the cell membrane at that region relative to the outside
to less negative. Inhibitory receptors permit negatively charged ions to pass (e.g., chloride,
Cl-), creating inhibitory post-synaptic potentials (IPSPs), further hyperpolarizing that region
by temporarily increasing the negative electric charge inside the cell membrane at that region
relative to outside. The activities of metabotropic (energy-requiring) receptors are a bit more
complicated. Neurotransmitter binding activates a G-protein complex attached to the receptor
protein, which primes adenylyl cyclase molecules inside the post-synaptic neuron to break down
adenosine triphosphate molecules into cyclic adenosine monophosphate molecules (cAMP) and
cellular energy. cAMP is the classic second messenger of molecular biology, whose activation in
41
John Bickle
turn induces effects at other locations in the cell; in this case often at distant receptor proteins,
which in turn permit ion influx, and EPSPs or IPSPs there. For those who learned their basic
neuroscience thirty years ago, the extent and importance of metabotropic receptors are impor-
tant recent updates.
Many synapses are plastic: post-synaptic electrochemical responses change over time to later
pre-synaptic neurotransmitter release. An early systematic study of synaptic plasticity focused
on a laboratory oddity that had been known to electrophysiologists as far back as the 1960s.
The frequency with which a post-synaptic neuron fires to mild electric stimulation to the
pre-synaptic neuron decreases over time and number of stimulations. But a strong electrical
current delivered through the stimulating electrode to the pre-synaptic neuron can return the
post-synaptic neuron to previous firing rates. Working in Per Andersen’s lab, Timothy Bliss and
Terje Lømo made the first systematic study of this phenomenon. Using a variety of stimulus
pulse trains, they stimulated the perforant path from entorhinal axonal projections onto gran-
ule cells in the rabbit hippocampus dentate gyrus, and measured field potential responses from
populations of post-synaptic neurons. Depending on the frequency and strength of stimulus
pulse trains administered, increased excitatory responses in post-synaptic neurons maintained
for up to ten hours—the longest duration they tested (Bliss and Lømo 1973).
Bliss and Lømo’s first publication on “long-lasting potentiation” concludes with some
hesitancy about its significance, especially about whether it occurs under natural conditions,
i.e., without the strong pre-synaptic electric stimuli by which they induced the effect. But
they note explicitly that their results “show that there exists at least one group of synapses in
the hippocampus whose efficiency is influenced by activity which may have occurred several
hours previously—a time scale long enough to be potentially useful for information storage”
(1973: 355). It was quickly noticed that their results connected with the purely theoretical
speculation a quarter-century earlier by neuropsychologist Donald Hebb:
When an axon of cell A is near enough to excite a cell B and repeatedly or persistently
takes part in firing it, some growth process or metabolic change takes place in one or
both cells such that A’s efficiency, as one of the cells firing B, is increased.
(Hebb 1949: 50)
These initial discoveries opened a floodgate of subsequent research. “Long-lasting potentiation”

was soon discovered in all hippocampus circuitries, in amygdala, throughout cortex, in cer-
ebellum, and even in spinal cord. Quickly it was re-named “long-term potentiation,” report-
edly because the acronym ‘LTP’ was easier to pronounce. Research quickly fragmented the
concept into numerous distinct functional subtypes: early- versus late-; associative versus non-
associative. Experiments from the late 1970s and early 1980s suggested that increases in the
number of post-synaptic glutamate receptors was one of the causes of LTP, since the kind of
repetitive stimulation which triggers LTP also increases these receptors (Baudry and Lynch
1980). Evidence for the prominence of glutamate as an excitatory neurotransmitter in the
mammalian central nervous system was increasing at this time, which explained the potentia-
tion effects following LTP induction: more glutamate binding to more available receptors, so
increased EPSPs to subsequent neurotransmitter release. By the mid-1980s, the “memory-like”
physiological features of LTP included:
• selectivity: increased transmission efficacy only at specific synapses on a given neuron;

• cooperativity: larger effects and more stable induction when more afferent fibers are
stimulated;
42
• multiple forms: the distinction between “early-” and “late-” LTP paralleled the short term–
long term memory distinction and the consolidation phase between;
• cumulative nature: successive stimulation to the same afferent fibers produced increased
synaptic potentiation, and
• regional distribution: all neural regions in which LTP had been demonstrated were associated
with some form of long-term memory in mammals (Lynch 1986).
So the LTP-memory hypothesis was thriving. But direct causal evidence for it awaited experiments
down one more level. These were occurring almost concurrently.
6. More scientific detail: “NMDA receptor activating”

That post-synaptic Ca++ influx was involved in LTP had been shown as early as 1983, when
Gary Lynch and colleagues showed that microinjecting the calcium chelator (binding agent)
EGTA into target cells prevented LTP formation (Lynch et al. 1983). This brought a post-
synaptic ionotropic glutamate receptor which opens a channel for Ca++ influx into experimental
focus: the N-methyl-D-aspartate receptor (NMDAR). At resting and slightly depolarized
membrane potentials NMDAR channels, even when bound by glutamate, remain blocked by
a magnesium ion. But when the surrounding membrane is highly depolarized by numerous
nearby activated excitatory receptors, the magnesium block is dislodged and the NMDAR
channel opens to Ca++ influx. Thus NMDARs act as coincidence detectors, for both pre-synaptic
neurotransmitter glutamate release and post-synaptic activity. Notice that these are the two
components of Hebb’s speculative rule, quoted above.
In 1983, Graham Collingridge and colleagues demonstrated a crucial role for NMDARs in
LTP induction in the hippocampus Shaffer collaterals-to-CA1 neurons pathway. A pharma-
cological agent, (2R)-amino-5-phosphonovaleric acid (AP5, sometimes written APV), which
selectively antagonizes NMDARs, blocked LTP induction in CA1 neurons without disrupt-
ing normal cell firing (Collingridge et al. 1983). Building on these electrophysiological results,
Morris (1989) administered AP5 directly into the brains of rats while they learned his water
maze task. AP5-treated rats were deficient on the hidden platform version. Hippocampus slice
electrophysiology in AP5-treated animals revealed deficits in LTP induction. A decade later,
and using a different selective NMDAR antagonist, (±)-3-(2-carboxylpiperazin-4-yl)propyl-
1-phosphonic acid (CPP), Clifford Kentros and colleagues in Eric Kandel’s lab connected
NMDAR activation directly with hippocampus place cell activity (Kentros et al. 1998). They
showed that while blocking NMDARs with CPP did not disrupt already-formed hippocampus
place fields in existing maps, nor prevent re-mapping in a novel environment, nor interfere
with maintenance of remapped place fields in the new environment for up to 90 minutes, it
did disrupt profoundly the long-term stability (24 hours) of the remapped place fields. The
re-mapped place fields in CPP-treated animals changed with each new day’s re-exposure to
the environment.
So NMDAR activation appears not only to influence neurons inducing LTP, but also hip-
pocampi forming spatial maps and animals remembering hidden platform locations in the Morris
water maze. Not all forms of LTP, or learning and memory, turned out to be NMDAR-dependent.
But research at the level of the molecular components of neuronal activity and its direct experi-
mental ties to behaviors indicating learning and memory hardly ended with this receptor—as
the Craver and Darden diagram appears wrongly to suggest. For more than twenty years, the
field of Molecular and Cellular Cognition has developed experimental tools to manipulate
specific components of intra-neuronal signaling pathways, and to tie these interventions directly
43
John Bickle
to behavioral effects. Gene targeting techniques applied to the mammalian genome, borrowed
originally from developmental biology, paved this research path. This research clearly demon-
strates the role of numerous other proteins in the molecular mechanisms of learning and memory
besides activated NMDARs. A word of forewarning: the molecular biology to follow probably
exceeds most philosophers’ and cognitive scientists’ comfort levels. But such is the nature of
cutting-edge memory research over the past 15 years, so I urge the reader to bear with me.
Alcino Silva, in Susumu Tonegawa’s lab, “knocked out” the gene for the -isoform of
the calmodulin kinase II protein (CaMKII). CaMKII is abundant in synapses; it is activated
by calmoduin loaded with calcium. It activates numerous intra-neuronal molecules via phos-
phorylation. It fit the conditions on a then-existing model for a kinase which could connect
NMDAR activity with both synaptic plasticity and learning and memory. Expression and syn-
thesis of -CaMKII was blocked completely in Silva’s mutant mice, but other, closely related
kinases were not affected. Slice-electrophysiology experiments revealed decreased hippocampus
and cortical LTP in mutants, compared to littermate non-mutated controls (Silva et al. 1992b).
Behavioral studies showed deficit learning and memory in hippocampus-dependent tasks,
including the Morris water maze hidden platform task and contextual feat conditioning (Silva
et al. 1992a). Later that year, Seth Green and collaborators in Eric Kandel’ lab “knocked out”
genes for a variety of tyrosine kinases (Grant et al. 1992). One of their mutants, the fyn tyrosine
kinase mutant, showed a similar pattern: deficit LTP in hippocampus and cortical slices, and
deficient learning and memory in a variety of hippocampus-dependent tasks. The general lesson
was clear: disrupting individual molecular components of intra-cellular signaling pathways by
disrupting the genes coding for these proteins could not only affect LTP electrophysiology, but
also learning and memory, even hippocampus-dependent forms.
That beat continues to the present day. A couple of additional examples will round out
this discussion, but they barely scratch the surface of what has been accomplished (see Silva
et al. 2014 for a broader overview). Another “memory molecule” was soon explored in detail.
Cyclic AMP-responsive element-binding-protein (CREB) is a gene-transcriptional effector. In
mammals the - and -isoforms are transcriptional enhancers: when phosphorylated these mole-
cules bind to specific sites on the genome to initiate the transcription of messenger RNA, and
ultimately gene expression and protein synthesis. CREB is part of numerous intracellular signal-
ing pathways across biological tissues. One, in mammalian post-synaptic neurons, begins with
activation of metabotropic dopamine receptors, generating cAMP molecules as its second mes-
senger. These in turn bind to regulatory subunits on protein kinase A (PKA) molecules, freeing
up catalytic PKA subunits to translocate back to the neuron’s nucleus, where they phosphorylate
CREB molecules. Phosphorylated CREB in turn drives gene expression and synthesis of regu-
latory proteins, which keep PKA in an active state after cAMP levels have returned to baseline,
and of effector proteins, which are trafficked back to active synapses and lock “hidden” excita-
tory glutamate receptors into “active” sites in the post-synaptic density. The result is potentiated
post-synaptic synapses, with additional active receptors locked into place for long durations by
the neuron’s own gene expression and protein synthesis machinery: “late-phase” LTP.
Günther Schütz’s developmental biology lab developed a viable homozygous CREB - and
-isoform knock-out mouse. Working with Schütz’s CREB mutants in Silva’s lab, Roussoudan
Bourtchuladze and colleagues (1994) showed they were deficient on long-term memory tasks
involving 24-hour delays between stimulus and test, but intact on learning and short-term memory
tested 30–60 minutes later on these same tasks. Memory tasks included both hippocampus-
dependent (aversive contextual fear conditioning, hidden-platform version of the Morris water
maze) and non-hippocampus-dependent ones (cued Pavlovian aversive conditioning to an audi-
tory tone). These results suggested that CREB plays a key role in the consolidation of memory
44
from short-term to long-term form. Hippocampus slice-physiology work proved consistent with
this interpretation. LTP in mutant slices was intact for a period after the tetanizing stimulus, but
declined back to initial baseline by 90 minutes (Bourtchuladze et al. 1994).
CREB mutants immediately became the target of extensive behavioral neuroscience research,
and the same pattern of results emerged for a large variety of rodent memory tasks: intact short-
term performance but impaired long-term performance on the same tasks. This same pattern
of results was also found for reconsolidation after stimulus re-presentation. Connections with
earlier results with CREB in invertebrate models (fruit flies, sea slugs) led to claims about the
cAMP-PKA-CREB pathway being an evolutionarily conserved mechanism for consolidation
and long-term memory across biological genera (Alberini 2009).
Neuroscientists were also attracted to a different gene engineering technique: the transgene
approach. This approach involves inserting an extra copy or copies of a cloned gene into
the DNA of mammal embryonic stem cells, often attached to a promoter region which
limits expression to specific neurons. Every cell in the mutants’ bodies contains the extra
transgene; but transcription and subsequent protein synthesis only occurs in those tissues or
specific cells possessing the promoter molecule in sufficient quantity. An early influential use
of this approach in memory research was led by Ted Abel in Kandel’s lab (Abel et al. 1997).
They inserted extra copies of the genes for regulatory subunits of PKA into mouse embry-
onic stem cells, with a CaMKII promoter which limited transgene expression to forebrain
regions. Transgene expression thus included hippocampus but excluded significant expression
in amygdala. In Abel and colleagues’ (1997) R transgene mutants, the extra PKA R (regula-
tory) subunits available in neurons in which the transgene is expressed quickly bind up PKA
catalytic subunits freed by increased cAMP, blocking that early step in the cAMP-PKA-
CREB pathway. Hippocampus PKA activity was reduced significantly in R transgenic mice,
as was late-phase LTP. Behaviorally, on the hidden platform version of the Morris water
maze, mutants were deficient in both time-in-target quadrant and number of target crosses
in probe trails administered after training (Abel et al. 1997: Figure 5). Most importantly,
R transgenic mutants were intact on short-term (1-hour delay) contextual fear condition-
ing, but significantly impaired on the long-term (24-hour delay) version, but unimpaired on
both short-term and long-term versions of tone-foot shock (Pavlovian) conditioning com-
pared to control mice. The former task is hippocampus-dependent, where the R transgene
was significantly expressed and synthesized; the latter task is amygdala-dependent, where the
R transgene was not (due to the CaMKII promoter).
Molecular and cellular cognition continues to press on, armed with increasingly sophisti-
cated gene intervention techniques (conditional knock-outs, adult genome manipulations via
virus vectors) and increasingly sophisticated behavioral paradigms indicative of more complex
cognitive functions. On the horizon lies the potential use of optogenetic technologies targeting
specific molecular activities (like CaMKII) and nanotechnological tools to intervene directly
into the three-dimensional configurations of proteins “on the fly,” rather than through the gene
expression and protein synthesis machinery, as the tools of molecular biology currently limit.
The biochemical mechanisms of memory now seem in sight of actual and direct experimental
manipulation and intervention. Philosophers and cognitive scientists ignore these lower levels at
their own peril—of potential irrelevance to ongoing memory research.
Related topics
• Memory causation
• Memory traces
45
John Bickle
Further reading
Bechtel, W. (2009). “Molecules, systems, and behavior: Another view of memory consolidation.” In
J. Bickle (ed.), Oxford Handbook of Philosophy and Neuroscience. New York: Oxford University Press,
13–40.
Bickle, J. (2012). “A brief history of neuroscience’s actual influences on mind–brain reductionism.” In
S. Gozzano and C. Hill (eds.), New Perspectives on Type Identity. Cambridge: Cambridge University
Press, 88–110.
Churchland, P.S. (1986). Neurophilosophy. Cambridge, MA: MIT Press.
Craver, C.F. (2007). Explaining the Brain. New York: Oxford University Press. See especially Chapters 5
and 7.
Silva, A.J., Landreth, A., and Bickle, J. (2014). Engineering the Next Revolution in Neuroscience. New York:
References
Abel, T., Nguyen, P.V., Barad, M., Deuel, T.A., Kandel, E.R., and Bourtchouladze, R. (1997). “Genetic
demonstration of a role for PKA in the late phase of LTP and in hippocampus-based long-term mem-
ory.” Cell 88: 615–26.
Aizawa, K. (2007). “The biochemistry of memory consolidation: A model system for the philosophy of
mind.” Synthese 155: 65–98.
Alberini, C.M. (2009). “Transcription factors in long-term memory and synaptic plasticity.” Physiological
Review 89: 121–45.
Baudry, M. and Lynch, G. (1980). “Regulation of hippocampal glutamate receptors: Evidence for the
involvement of a calcium-activated protease.” Proceedings of the National Academy of Sciences USA 77:
2298–302.
Bechtel, W. (2009). “Molecules, systems, and behavior: Another view of memory consolidation.” In J. Bickle
(ed.), Oxford Handbook of Philosophy and Neuroscience. New York: Oxford University Press, 13–40.
Berlyne, D.E. (1950). “Novelty and curiosity are determinants of exploratory behavior.” British Journal of
Psychology 41: 68–80.
Bickle, J. (2003). Philosophy and Neuroscience: A Ruthlessly Reductive Approach. Dordrecht: Springer.
Bickle, J. (2006). “Reducing mind to molecular pathways: Explicating the reductionism implicit in current
cellular and molecular neuroscience.” Synthese 151: 411–34.
Bliss, T.V. and Lømo, T. (1973). “Long-lasting potentiation of synaptic transmission in the dentate area of
the anaesthetized rabbit following stimulation of the perforant path.” Journal of Physiology 232: 331–56.
Bourtchuladze, R., Frenguelli, B., Blendy, J., Cioffi, D., Schutz, G., and Silva, A.J. (1994). “Deficient
long-term memory in mice with a targeted mutation of the cAMP-responsive element-binding
protein.” Cell 79: 50–68.
Churchland, P.S. (1986). Neurophilosophy. Cambridge, MA: MIT Press.
Cohen, N.J., Eichenbaum, H., DeAcedo, B.S., and Corkin, S. (1985). “Different memory systems under-
lying acquisition of procedural and declarative knowledge.” Annals of the New York Academy of Science 444:
54–71.
Collingridge, G.L., Kehl, S.J., and McLennan, H. (1983). “Excitatory amino acids in synaptic transmis-
sion in the Schaffer collateral-commissural pathway of the rat hippocampus.” Journal of Physiology 334:
33–46.
Craver, C.F. (2007). Explaining the Brain. New York: Oxford University Press.
Craver, C.F. and Bechtel, W. (2007). “Top-down causation without top-down causes.” Biology &
Philosophy 22: 547–63.
Craver, C.F. and Darden, L. (2001). “Discovering mechanisms in neurobiology: The case of spatial mem-
ory.” In P.K. Machamer, R. Grush, and P. McLaughlin (eds.), Theory and Method in Neuroscience.
Pittsburgh, PA: University of Pittsburgh Press, 112–37.
Deisseroth, K.D. (2015). “Optogenetics: 10 years of microbial opsins in neuroscience.” Nature Neuroscience
18: 1213–25.
Desimone, R., Albright, T.D., Gross, C.G., and Bruce, C. (1984). “Stimulus-selective properties of infe-
rior temporal neurons in the macaque.” Journal of Neuroscience 4: 2051–62.
Doeller, C.F., Barry, C., and Burgess, N. (2010). “Evidence for grid cells in a human memory network.”
Nature 463: 657–61.
46
Goshen, I. (2014). “The optogenetic revolution in memory research.” Trends in Neuroscience 37: 511–22.
Grant, S.G., O’Dell, T.J., Karl, K.A., Stein, P.L., Soriano, P., and Kandel, E.R. (1998). “Impaired long-term
potentiation, spatial learning and hippocampal development in fyn mutant mice.” Science 258: 1903–10.
Gross, C.G., Rocha-Miranda, C.E., and Bender, D.B. (1972). “Visual properties of neurons in inferotem-
poral cortex of the macaque.” Journal of Neurophysiology 35: 96–111.
Hafting, T., Fyhn, M., Molden, S., Moser, M.B., and Moser, E.I. (2005). “Microstructure of a spatial map
in the entorhinal cortex.” Nature 436 (7052): 801–6.
Hebb, D.O. (1949). The Organization of Behavior. New York: Wiley.
Jacobs, J., Weidermann, C.T., Miller, J.F., Solway, A., Burke, J.F., Wei, X.-X., Suthana, N., Sperling,
M.R., Sharan, A.D., Fried, I., and Kahana, M.J. (2013). “Direct recordings of grid-like neuronal activ-
ity in human spatial navigation.” Nature Neuroscience 16: 1188–90.
Kentros, C., Hargreaves, E., Hawkins, R.D., Kandel, E.R., Shapiro, M., and Muller, R.V. (1998).
“Abolition of long-term stability of new hippocampal place cell maps by NMDA receptor blockade.”
Science 280: 2121–6.
Liu, X., Ramirez, S., Pang, P.T., Puryear, C.B., Deisseroth, K., and Tonegawa, S. (2012). “Optogenetic
stimulation of a hippocampal engram activates fear memory recall.” Nature 484 (7394): 381–5.
Lynch, G. (1986). Synapses, Circuits, and the Beginnings of Memory. Cambridge, MA: MIT Press.
Lynch, G., Larson, J., Kelso, S., Barrioneuvo, G., and Schottler, F. (1983). “Intracellular injections of
EGTA block induction of hippocampal long-term potentiation.” Nature 305: 719–21.
Milner, B. (1958). “Psychological defects produced by temporal lobe excision.” Research Publications—
Association for Research in Nervous and Mental Diseases 36: 241–57.
Morris, R.G.M. (1981). “Spatial localization does not require the presence of local cues.” Learning and
Motivation 12: 239–60.
Morris, R.G.M. (1989). “Synaptic plasticity and learning: Selective impairment of learning in rats and
blockage of long-term potentiation in vivo by the N-methyl-D-aspartate receptor antagonist AP5.”
Journal of Neuroscience 9: 3040–57.
O’Keefe, J. (1976). “Place units in the hippocampus of the freely-moving rat.” Experimental Neurology 51:
78–109.
O’Keefe, J. and Dostrovsky, J. (1971). “The hippocampus as a spatial map: Preliminary evidence from unit
activity in the freely-moving rat.” Brain Research 34: 171–5.
Quiroga, R.Q., Reddy, L., Kreiman, G., Koch, C., and Fried, I. (2005). “Invariant visual representation
by single neurons in the human brain.” Nature 435: 1102–7.
Ramirez, S., Liu, X., Lin, P.-A., Suh, J., Pignatelli, N., Redondo, R.I., and Tonegawa, S. (2013).
“Creating a false memory in the hippocampus.” Science 341 (6144): 387–91.
Schacter, D.L. (1987). “Implicit memory: History and current status.” Journal of Experimental Psychology:
Learning, Memory and Cognition 13: 501–18.
Scoville, W.B. and Milner, B. (1957). “Loss of recent memory after bilateral hippocampal lesions.” Journal
of Neurology, Neurosurgery, and Psychiatry 20: 11–21.
Silva, A.J., Landreth, A., and Bickle, J. (2014). Engineering the Next Revolution in Neuroscience. New York:
Silva, A.J., Paylor, R., Wehner, I.M., and Tonegawa, S. (1992a). “Impaired spatial learning in alpha-
calcium-calmodulin kinase II mutant mice.” Science 257 (5067): 206–11.
Silva, A.J., Stevens, C.F., Tonegawa, S., and Wang, Y. (1992b). “Deficient hippocampal long-term
potentiation in alpha-calcium-calmodulin kinase II mutant mice.” Science 257 (5067): 201–6.
Squire, L.R. (1992). “Memory and the hippocampus: A synthesis of findings with rats, monkeys, and
humans.” Psychological Review 99: 195–231.
Squire, L.R. (2004). “Memory systems of the brain: A brief history and current perspective.” Neurobiology
of Learning and Memory 82: 171–7.
Thor, D.H. and Holloway, W.R. (1982). “Social memory of the male laboratory rat.” Journal of Comparative
and Physiological Psychology 96: 1000–1006.
Tolman, E.C. (1948). “Cognitive maps in rats and men.” Psychological Review 55: 189–208.
Tulving, E. and Craick, F. (2000). The Oxford Handbook of Memory. New York: Oxford University Press.
Xu, Y. and Corkin, S. (2001). “HM revisits the Tower of Hanoi puzzle.” Neuropsychology 15: 69–79.
47
PART II
The metaphysics of memory

4
MEMORY AND TRUTH
Sven Bernecker
Broadly speaking, there are two kinds of philosophical theories of memory: traditional archival
views and contemporary constructive views.1 The archival view claims that memory is a purely
passive device for registering, storing and reproducing representations of particular past experi-
ences. On this picture, a subject misremembers whenever the content of her state of seeming
to remember differs from the content of the corresponding original representation. Given
that memory aims at preservation of content, any discrepancy between the encoded and the
retrieved content is taken to be a mistake. Hume, for instance, declares that “memory preserves
the original form, in which its objects were presented, and that wherever we depart from it in
recollecting anything, it proceeds from some defect or imperfection in that faculty” (2000: 12).
The archival view is still very much with us today; it is a tacit assumption behind the widespread
storehouse metaphors of memory.
Proponents of constructivism argue that the archival view is at odds with what science tells us
about the workings of memory. Retrieval is said to be almost always more a process of con-
struction than one of simple retrieval. The fact that our everyday memory frequently changes
the encoded information should not be regarded as an abnormal lapse of an otherwise reliable
cognitive faculty, but as part of the very function of memory. According to constructivism,
all attempts at remembering—whether or not the encoded content is preserved—stem from
a single, adaptive process whose function it is to construct ‘accurate’ representations at the
time of recall. The ‘accuracy’ criteria have to do with the role that memory is said to play in
regulating expectations, steering future planning and action, and establishing social cohesion
by generating self-narratives. Among the philosophers who promote constructivism about
memory are De Brigard (2014), Michaelian (2016), and Sutton (2007).
Whether memory is said to imply truth and what is meant by the notion of ‘truth in
memory’ crucially depends on which of these camps one belongs to. In what follows, I will
steer a middle course between the archival view, on the one hand, and constructivism, on the
other. I do not go as far as some constructivists in claiming that remembering never involves
attempting to get at, and establish the truth about what happened in the past faithfully. But
I also distance myself from the archival theorist’s claim that it is the function of human
memory to produce exact copies of past representations. According to the position sketched
in this chapter, memory is supposed to provide us with knowledge of the past but it also has
the function of editing the encoded information.
51
Sven Bernecker
The first section explains and motivates the truth condition on memory. The second section
discusses positions whereupon memory reports need not be completely true but only true to a
degree. The third section explores the authenticity condition on memory. A memory report
is authentic if it correctly represents the subject’s initial representation of reality, regardless of
whether the initial representation was veridical. The fourth and final section deals with the
question of how we assess whether someone’s recollection of an event accurately depicts the
event or his initial representation of the event.
1. The truth condition

Though talk of ‘false memory’ is familiar enough, it is an oxymoron. ‘To remember’ is fac-
tive in the sense that an utterance of “S remembers that p” (where ‘S’ stands for a subject
and ‘p’ stands for a proposition) is true only if p is the case. If not-p, then S may think she
remembers that p (she may be in a state of seeming to remember that p), but she doesn’t actu-
ally remember that p. And it is not only propositional memory that implies truth. Memory
of persons, objects, events and properties is also factive. I cannot remember a person, say,
John F. Kennedy or an event, say, his being assassinated by Lee Harvey Oswald unless there
was a person called ‘John F. Kennedy’ who was assassinated by a person called ‘Lee Harvey
Oswald.’ And habit memory (remembering how), though it does not imply truth, implies
something similar, namely success under certain counterfactual circumstances (Hawley 2003).
For instance, remembering how to play Mozart’s Piano Concerto No. 21 requires that I would
succeed in playing the concerto if I were to try under certain circumstances.
Hazlett (2010) has recently challenged the orthodox view among philosophers that the verb
‘to know’ is factive. He cites a few cases in which ‘know’ is used non-factively but which do
not strike most people as deviant, improper, unacceptable, or necessarily false. One such case
is “Everyone knew that stress caused ulcers, before two Australian doctors in the early [19]80s
proved that ulcers are actually caused by bacterial infection” (Hazlett 2010: 501). An analogous
case for the verb ‘to remember’ is “Thales remembered that the earth was flat when he set sail
towards Sicily.” Hazlett argues that the best explanation for why non-factive usages of allegedly
factive verbs seem acceptable is that these verbs are non-factive.
It is debatable whether Hazlett’s argument for the non-factivity of ‘know’ and ‘remember’
is in fact the best available explanation. There are alternative explanations for why non-factive
usages of ‘know’ (‘remember’) do not strike us as odd, explanations that do not challenge the fac-
tivity of ‘know’ (‘remember’). Consider again the statement “Everyone knew that stress caused
ulcers, before two Australian doctors in the early [19]80s proved that ulcers are actually caused by
bacterial infection.” If it literally means everyone, then the statement is clearly false. What is
meant is that some people knew. But these people knew, at best, that stress caused some ulcers,
not all ulcers. Next consider “Two Australian doctors proved that ulcers are caused by bacterial
infection.” Again, these doctors only proved that bacterial infection causes some ulcers. What the
statement means is therefore something like this: “Some people knew that stress caused some
ulcers, before two Australian doctors proved that bacterial infection causes some ulcers.” But
note that we can understand this statement as true without challenging the factivity of ‘know.’
It could be that stress and bacterial infection cause different kinds of ulcers. Another possibility is
that stress causes bacterial infections which, in turn, cause ulcers. The upshot is that non-factive
usages of ‘know’ that strike us as appropriate occur in loose talk and overstatements.2
The truth condition of memory can be motivated in two ways. One argument makes use
of the fact that statements of the form “I remember that p; but p is false” and “I remember
such-and-such; but such-and-such never happened” have a paradoxical ring to them. These
52
Memory and truth
statements are in the same way paradoxical as G.E. Moore’s famous statement “It is raining; but
I don’t believe that it is raining.” Though not literally contradictory, none of these statements
can be used to make a coherent assertion.
The crux with using Moore’s paradox to motivate the truth condition of memory is that the
paradoxical nature of the statement “I remember that p; but p is false” can be explained without
challenging the thesis that memory implies truth. Consider the following explanation: When
I claim to remember that p, I am convinced that p is the case. This is what the first part of the
statement “I remember that p; but p is false” expresses. Yet the second part of the statement
denies that p is the case. Thus the reason “I remember that p; but p is false” need not be that
one cannot remember that p without p being the case. Instead, the incoherence of the statement
may be due to the fact that one cannot claim to remember that p while claiming that p is false.
And given that the conditions for claiming to remember that p are distinct from the conditions
for remembering that p, it does not follow that memory implies truth just because claiming to
remember that p implies the truth of p.
Another argument for the truth condition of memory is based on syntactical considera-
tions. Both factive and non-factive propositional verbs can take that-clause complements (e.g.,
“S remembers/believes that Kennedy was shot by Oswald”), but the that-clauses following
factive verbs are different from those following non-factive verbs. Only the that-clauses fol-
lowing factive verbs can be transformed into wh-nominals, i.e., clauses beginning with ‘who,’
‘whom,’ ‘what,’ ‘where,’ ‘when,’ and ‘why’ (Vendler 1972: 93–9; 1980: 280–2). We can say “S
remembers who shot Kennedy,” “S remembers who Oswald shot,” “S remembers why Oswald
shot Kennedy,” “S remembers where Oswald shot Kennedy,” “S remembers what Oswald did
to Kennedy,” and “S remembers when Oswald shot Kennedy.” But we cannot say “S believes
who shot Kennedy,” “S believes who Oswald shot,” “S believes why Oswald shot Kennedy,”
“S believes where Oswald shot Kennedy,” “S believes when Oswald shot Kennedy.”3
Given that memories are necessarily true, they are not transparent from a first-person per-
spective. I cannot tell, on the basis of reflection, whether the proposition I ostensibly remember
is in fact true. But if I cannot know this, then I cannot tell, by reflection alone, whether the
activity I am currently engaged in qualifies as remembering or whether it is an instance of,
say, confabulating. The same applies to other factive verbs such as ‘know’ and ‘see.’ I may be
mistaken in thinking that I know something which is in fact false and cannot be known. Or I
may be mistaken in thinking that I see something which isn’t there. I will return to this topic
in Section 4.
Since errors regarding the self-attribution of memories may be irremediable by introspec-
tion, there are four possible reasons for why a claim to remember a past factive attitude may
be incorrect. Suppose I claim to remember having seen that the cat is on the mat. First, I may
have taken, say, a dog for a cat. Here it is the past perception, not the memory which is to
blame. Second, I may misremember what it is that I saw in the past. Here, the fault lies with the
memory, not the past perception. Third, both kinds of mistakes can be combined. Suppose I
ostensibly remember having seen a cat on the mat, but what I took myself to be seeing, at that
time, was a dog and the perception was false for there was no dog but, say, a squirrel. Here,
the fault lies both with the memory and the perception. Fourth, the perceptual mistake and the
memory mistake can balance each other out. Suppose I claim to remember having seen a cat
on the mat; but what I took myself to be seeing, at that time, was a dog, and the perception
was false for it was in fact a cat that I saw. In this case, though the memory report is veridical,
it doesn’t qualify as a genuine memory because the causal condition of memory is not satisfied.
The upshot is that one can misremember something not only because one’s memory malfunc-
tions but also because the representation fed into the memory process is false.
53
Sven Bernecker
While most philosophers maintain that the content emerging from the memory process
must be veridical, there is no consensus as to whether the content fed into the memory process
must be veridical as well. Anscombe (1981: 105–6) claims that memory requires present and
past truth. Von Leyden (1961: 62), on the other hand, seems to hold that present truth is suf-
ficient. Memory, he claims, allows for “the inheritance of a mistake” but is incompatible with
“a mistake of inheritance.”
In my view, for a representation to qualify as memory, it must track what one took to be
true. A memory must authentically report a subject’s past representation, but the past repre-
sentation need not have been true at the time it was entertained. Memory demands a present
truth-condition but not a past truth-condition. Elsewhere (2010: 38–9, 74) I have argued that
the authentic reproduction of a proposition that was false at the time it was initially entertained
but which, in the meantime, has become true due to good fortune may qualify as memory.
Even if the content of one’s past representation, p*, was false, one’s present representation that
p qualifies as a memory provided the following conditions hold: (i) p is true, and (ii) one would
not represent that p in the present unless one had represented that p* in the past.
2. Standards of truth
As was mentioned above, there are two schools of thought in philosophy regarding the impor-
tance of representational fidelity for remembering. According to the traditional archival view,
remembering demands that the content of the present representation be the same as that of the
past representation. Constructivists, on the other hand, maintain that memory is not primarily
about reproducing the contents of previous experiences. That is why remembering is said to
allow for (minor) distortions and errors. Bernstein and Loftus express the constructivist position
when they write:
All memory is false to some degree. Memory is inherently a reconstructive process,

whereby we piece together the past to form a coherent narrative that becomes our
autobiography. In the process of reconstructing the past, we color and shape our life’s
experiences based on what we know about the world.4
(Berstein and Loftus 2009: 373)
Other scientists go one step further by claiming that truth is irrelevant for memory. Klein, for
example, writes:
there is no principled reason for episodic recollection to adhere to any particu-

lar degree of fidelity to the past; all that matters, from a functional perspective, is
that the information supplied is beneficial to the adaptive challenges faced by the
organism . . . Environmental regularities and the demands of reality place limits on
which anticipatory behaviors will work, how well they will work, and which will fail.
Nonetheless, within the (sometimes fairly broad) constraints imposed by reality, the
memory content served up to consciousness need not entail ‘precision of match’ to
past events as a criterion of success.
(Klein 2014: 438–9)
The views discussed so far represent two poles on a continuum of philosophical positions.
Instead of claiming that truth is either indispensable or irrelevant for memory, some argue for a
position in the middle whereby memories must be true to a degree. Hamilton (1998: 283),
54
Memory and truth
for instance, distinguishes between memories that are false in detail and those that are completely
false: in the former, we transpose or condense experiences or elements of past experiences,
whereas the latter fail to match past experience entirely. According to Hamilton, memories
may be false in detail but they may not be completely false. A related distinction is that between
memory for gist and memory for details (see Koriat et al. 2000: 291–3). The idea is, once again,
that memories can be true without being perfectly accurate or verbatim records of past events.
A real-life example of memory for gist are John Dean’s recollections of conversations with
U.S. President Nixon in the context of the Watergate scandal. When John Dean, Counsel
to Nixon, testified against Nixon before the Senate’s Watergate Committee, he was unaware
that the conversations he had with Nixon in September 1972 had been tape-recorded. In his
testimony, Dean provided an incredibly detailed account of these conversations. When this tes-
timony was compared with the tape recordings, members of the hearing committee concluded
that Dean had told the truth despite the fact that almost all of his memories for details were false.
In his study of Dean’s memories, the psychologist Neisser (1981) identified three standards of
truth which can be imposed on memories: (i) accurately reproducing the details of a conversa-
tion, (ii) distorting the details but retaining the gist or overall meaning, and (iii) distorting both
details and gist, but remaining faithful to the overall theme or ‘narrative truth’ of the events.
Temporally displaced memories (the reported event really happened, but not when the sub-
ject claims it happened) and source monitoring errors are further examples of memories that are
partially false. Source monitoring errors occur when a subject confuses what she experienced
first-hand with what she has learned via testimony from another source. A real-life example
of this type of partially correct memory is provided by Crombag and colleagues’ (1996) study.
Crombag studied the memories of one hundred Amsterdam residents of El Al Flight 1862, a
Boeing 747 cargo aircraft, crashing into an apartment building in Amsterdam in 1992. Even
though no one had filmed the crash, the researchers found that 66 percent of the witnesses said
they saw the plane crash on TV. It would appear that the Amsterdam residents pieced together
what they had heard about the crash from different sources to construct an image of the crash,
and then accepted the suggestion that they had watched it on TV.
Sometimes our memory reports are basically accurate as to observable facts about the world
while they are mistaken because of their bizarre interpretation of those facts. Schechtman (1996:
126) provides the following example: someone who constantly sees in clearly innocent actions
and gestures evidence of a sinister conspiracy directed against him gives the following memory
report: “Yesterday, while shopping, there was a group of men in black suits watching me and
taking notes. This is no coincidence; the CIA is once again after me.” Provided the paranoiac is
not suffering from hallucinations, the observable facts mentioned in his memory report—men
in black suits watched him and took notes—are correct. It is just the interpretation of these facts
which is absurd. The men were not CIA agents but market researchers counting the number
of shoppers. This seems to be yet another example of a memory report that is partially correct.5
Memory errors can be grouped into two categories: errors from omission and errors from
commission. Omissions are forgetting errors. Commissions involve distorted or unwanted rec-
ollections and are commonly labeled false memories.6 Some memory disorders (e.g., Alzheimer’s
disease) mainly affect the amount of correct information (or percentage of the input) repro-
duced while other memory disorders (e.g., Korsakoff syndrome) primarily diminish the
accuracy of the reported information by inserting fictitious elements into the recall. Whether
a report is classified as a genuine memory depends on the underlying assessment of accuracy:
whether one counts the absolute number of errors in the reported information or whether one
looks at the proportion of erroneous and confabulated statements that is indexed. Presumably
the choice between these standards of truth is not arbitrary but depends on the social context.
55
Sven Bernecker
In a forensic setting, for instance, it is usually the absolute number of errors reported rather
than the accuracy rate (regardless of the absolute number of errors and confabulations) that is
crucial. Yet in a clinical setting. it may be the accuracy rate that is more important than the
absolute number of errors.
3. The authenticity constraint

So far the accuracy of memory has been understood as consisting in the correspondence of
the memory report with the objective reality. This is not the only conception of memorial
accuracy. An alternative approach is to say that a memory is accurate if it accords with the
subject’s initial perception of reality, whether or not the initial perception was veridical. On
the former (external) conception, memorial accuracy has a mind-to-world direction of fit;
on the latter (internal) conception, it has a mind-in-the-present-to-mind-in-the-past. The
idea is that we cannot ask more out of memory than that recollections reflect the person’s
original perspective; otherwise we confuse errors in perception with errors in memory. Odell
(1971) illustrates the internal conception of memorial accuracy by imagining that he teaches a
child the false statement that Columbus discovered America in 1392. When asked to remem-
ber when Columbus discovered America, the child answers ‘1392.’ “What does [the child]
remember? He remembers what I told him. What did I tell him? That Columbus discovered
America in 1392. So it follows that he remembers that p.”7 In Odell’s mind, cases like this one
suggest that one can remember a falsehood.
I use the term authenticity to refer to the accuracy of the present rendition of a past repre-
sentation (true and false) by means of a memory judgment. The truth of a memory report, on
the other hand, has to do with the memory content correctly representing the objective real-
ity. Sometimes these two accuracy standards are played off against one another. In my view, a
memory report must accurately represent the objective reality and resemble the subject’s initial
perception. Just as the faithful rendering of a false proposition does not qualify as memory,
neither does the distorted rendering of a true proposition. Memories must be both veridical
and authentic.
What are the conditions on the authentic rendering of a past representation? According to
the archival view, the mark of memorial authenticity is content identity: for a present representa-
tion to be memory-related to a past representation, the contents of both representational states
must be type-identical.8 Constructivists, on the other hand, hold that a memory report need not
amount to an exact reproduction of some previously encoded content. The contents of the past
representation and of the present rendition need to be only sufficiently similar. But what is the
permissible range of aberration between the content of a past representation and the memory
content thereof? What is the margin of error regarding content reproduction? What, in other
words, are the bounds of authenticity with respect to remembering?
In (2010: 222–9), I develop an account of memorial authenticity for propositional memory.
The basic idea is that the content of a propositional attitude retrieved from non-inferential
memory may be informationally impoverished vis-à-vis the content of the propositional atti-
tude fed into the memory system. Non-inferential memory allows for the decrease, but not
the increase or enrichment of information. The content of a non-inferential memory must be
a relevant entailment of the original content.9 The entailment thesis can be illustrated by this
example. On Monday morning you have scrambled eggs for breakfast. On Tuesday all you can
remember is that you had eggs for breakfast; you have forgotten how the eggs were prepared.
Notwithstanding the fact that I had eggs for breakfast and I had scrambled eggs for breakfast are dif-
ferent propositions, it is natural to suppose that the former belief is memory-related to the latter
56
Memory and truth
one—provided, of course, the other memory conditions are met. The reason, I reckon, the
discrepancy between the two content tokens does not and should not prevent us from granting
propositional memory is that the proposition I had eggs for breakfast is entailed by the proposition
I had scrambled eggs for breakfast.
While non-inferential memory allows only for the decrease of information, inferential
memory also allows for the increase or enrichment of information. Consider the follow-
ing example. On Monday morning you have scrambled eggs for breakfast. On Tuesday you
remember that the breakfast you had the previous day was not vegan. I have scrambled eggs for
breakfast relevantly entails that my breakfast wasn’t vegan. The belief that my breakfast wasn’t
vegan does not qualify as a non-inferential memory of the belief that I have scrambled eggs for
breakfast, but it does meet the conditions for inferential memory, provided, of course, I know
what ‘vegan’ means.
Michaelian (2011, 2016) develops an opposing view, according to which non-inferential
memory allows for the enrichment of informational content. Appealing to research on con-
structive memory, he argues that remembering has a simulational character, in the sense that
it routinely involves the generation of new information and the incorporation of information
originating in sources other than experience of the remembered event. For example, in the phe-
nomenon of boundary extension (Intraub and Richardson 1989), the subject ‘remembers’ more
of a scene than he actually saw, i.e., he ‘remembers’ parts of the scene that were beyond his field
of view at the time of the experience. Since boundary extension and similar phenomena are
ordinary and frequent occurrences, Michaelian argues that they are compatible with the proper
functioning of the memory system and hence with the occurrence of genuine remembering.
On his view, then, it is literally possible for one to remember more than one experienced.
This is a counterintuitive view, and several more conservative responses to the research
on constructive memory are available. First, we might insist that cases in which remembering
involves the generation or incorporation of new content are cases of merely apparent remem-
bering. Second, we might distinguish between the components of the content which originate
in the relevant experience and those which do not, treating the former as genuinely remem-
bered but the latter as merely apparently remembered. Finally, we might argue that the relevant
cases are best understood as cases of inferential memory.
So far we have concerned ourselves with the authenticity of propositional memory. But what
are the bounds of authenticity with respect to visual remembering? The answer to this question
crucially depends on whether the internal representations of memory images are said to repre-
sent in the manner of (physical) pictures or in the manner of language. Proponents of pictorialism
(like Kosslyn and Fodor) hold that the mental representations we experience as imagery are
like pictures with intrinsically spatial representational properties of the sort that pictures have.
Proponents of descriptionalism (like Pylyshyn and Dennett), on the other hand, hold that the
mental representations that we experience as imagery are more like linguistic descriptions of
visual scenes.
If memory images are linguistic descriptions of visual scenes, as descriptionalism claims, the
entailment thesis which was developed for propositional memory carries over to visual memory.
But what is the rule of memory abstraction, if the content of visual memories is imagistic?
Elsewhere I have argued that non-inferential visual memory allows for the omission of content
(2015: 458–61). The omission of content can take different forms. One way for content to be
omitted in the process of remembering is that the memory image is indeterminate with respect
to particular features. You may, for instance, visually remember your friend sitting across from
you sipping coffee but your memory may be non-committal regarding the color of your friend’s
pullover. Another way for content to be lost in the process of remembering is that the mental
57
Sven Bernecker
image retrieved from memory is a cropped version of a previously perceived scene. You may,
for instance, visually remember your friend sipping coffee but not remember the scenery around
her—the color of the wall, the people at the adjacent table, the decor of the room, etc.
4. Assessing accuracy
How do we assess whether someone’s recollection of an event accurately depicts the event
or his initial representation of the event? We do not have direct access to the subject’s initial
representation of the event. And we cannot bring back the past to directly compare the event
with the subject’s recollection of the past. Instead, we have to use indirect means when try-
ing to validate ostensible memories. We have to rely on diaries, photographs, hearsay, etc.
But this kind of validation is circular. Suppose you seem to remember having put your key in
the drawer. You open the drawer, and there is the key. Does finding the key in the drawer
confirm your ostensible memory of the key being in the drawer? Does it prove that what you
took to be a memory was indeed a memory? No, for the key might have been in the drawer
for reasons completely unknown to you. Next suppose you try to confirm your ostensible
memory of having put the key in the drawer by asking a friend whether he saw you put the
key there. The friend answers in the affirmative. Does this validate your ostensible memory?
No, for your friend’s ostensible memory might be just as unreliable as your own. And to that
your friend’s memory has been reliable in the past you would have to rely, once again, on your
own memory. The problem of verifying ostensible memories is only pushed from you to your
friend. Price nicely summarizes this problem:
no one memory can be validated or invalidated without relying on other memo-

ries . . . It is often supposed that we can validate or invalidate a memory-judgement by
means of a present perception, for example by consulting documents or records. Again,
it is supposed that we can do it by appealing to the established laws of nature . . . But
in both cases we are using memory over again, because we are relying on inductive
generalizations . . . however great the probability of an inductive generalization may
be, its probability is derived from past observations. We have only memory to assure us
that those past observations existed, or what sort of observations they were.
(Price 1969: 78–9)
What Price points out is that any inductive argument for the trustworthiness of memory
experiences is open to the charge of vicious circularity. We cannot validate our ostensible
rememberings without already assuming the reliability of ostensible rememberings. And since
a circular justification is no justification at all, the problem of validating ostensible memory
seems to be insoluble.
The problem of validating our ostensible remembering has striking similarities to Hume’s
problem of induction, that is, the problem of justifying our tacit belief in the principle of the
uniformity of nature. Both problems result from attempts to justify epistemic methods which
apparently can be justified only by appeal to their own principles. It is not surprising then that
the same approaches that are used to deal with the problem of induction are also brought to
bear on the problem of validating memory experiences. The three main putative solutions to the
Humean problem of induction are the postulational, the pragmatic, and the analytic approach.
Russell (1948: 288) and Saunders (1963: 486) solve the problem of validating our ostensible
memories by maintaining that the reliability of ostensible memory is a postulate that can-
not be proven or disproven but must be assumed. Yet the postulation of an unjustified and
58
Memory and truth
unjustifiable principle makes our reliance on memory experiences a matter of faith. If memory
knowledge is basically a matter of faith, then this faith exists on a par with other faiths. Reliance
on memory experiences has no ground on which to maintain its cognitive superiority to any
other form of irrationalism.
According to the pragmatic justification of our reliance on ostensible memories, trusting
one’s ostensible memories, though ultimately unjustified, is better suited to the goal of uncover-
ing the past than any other method that might be adopted. This approach has been defended
by Brandt who writes:
the only acceptable theory [of our ostensible memories] is one which asserts that a
large proportion of our memory beliefs are veridical. No alternative to such a theory
has been proposed; nor can one imagine what one would be like.
(Brandt 1955: 92–3)
As was noted by Locke (1971: 113–14), the crux with Brandt’s justification of ostensible mem-
ory is that such an alternative theory has indeed been proposed as a logical possibility that has not
yet been ruled out. The alternative is Russell’s (1921: 159–60; 1948: 228) suggestion that there
has been no past at all for us to remember since the world sprang into being five minutes ago.
Until Russell’s proposal has been ruled out, Brandt does not seem to be entitled to conclude
that ostensible memory is reliable.
Malcolm (1963: 196) and Shoemaker are proponents of the analytic approach to the problem
of validating ostensible memories. Shoemaker, for instance, wants to show “that it is a necessary
(logical or conceptual) truth, not a contingent one, that when perceptual and memory state-
ments are sincerely and confidently asserted, i.e., express confident beliefs, they are generally
true” (1963: 229).10 As far as I can see, there are two main arguments to the effect that ostensible
memories are generally true. This is Shoemaker’s presentation of the first argument:
A primary criterion for determining whether a person understands the meaning of such
terms as ‘see’ and ‘remember’ is whether under optimum conditions the confident
claims that he makes by the use of these words are generally true. If most of a person’s
apparent perceptual and memory claims turned out to be false, this would show, not
that the person had exceptionally poor eyesight or an exceptionally bad memory, but
that he did not understand, had not correctly grasped, the meanings of the words he
was uttering, or was not using them with their established meanings, i.e., was not using
them to express the perceptual and memory claims they appear to express.
(Shoemaker 1963: 231; cf. Malcolm 1963: 193–4)
Shoemaker claims that if someone were to consistently make wildly inaccurate claims about the
past, and seemed to remember things that never happened, we would have to say not that he
was misremembering, but that he had lost his understanding of ‘to remember.’
It seems to me that Shoemaker is right in that habitual mistakes regarding one’s memory
claims can be due to a mistake of terminology. The question is, however, whether someone
whose memory claims are habitually wrong necessarily misunderstands ‘to remember’ and the
past tense. Couldn’t the habitual mistakes in question be due to mistakes of fact rather than
meaning? I am inclined to believe that knowing what a term means and how to use it need
not go together with using the term to form true statements. Whether someone understands an
expression is determined by what he thinks he is saying when he uses that expression to form
a statement and what he thinks would make the statement true—not by the truth value of the
59
Sven Bernecker
statement. Even if someone’s memory claims were consistently wrong, he could still have a
correct understanding of the verb ‘to remember.’ That he correctly understands ‘to remember’
could be established by the fact that he uses the term only to talk about things that, he believes,
did happen and not about things that, he believes, he imagined.
The second argument to the effect that ostensible memories are generally true rests on the
claim that one cannot question one’s confident perceptual and memory beliefs. Shoemaker writes:
It is precisely one’s confident beliefs, and especially one’s confident perceptual and
memory beliefs, that one expresses by saying ‘I know . . . ;’ it is not a psychological
fact, but rather a logical fact, that one cannot help regarding one’s confident perceptual
and memory beliefs as constituting knowledge.
(Shoemaker 1963: 234)
Shoemaker goes on to argue that what is true of me is also true of other people. Each one
of us has to claim that his confident perceptual beliefs and memory beliefs are generally true.
Shoemaker concludes that it is necessarily true of confident perceptual and memory beliefs in
general that they are generally true. But this conclusion is problematic. Just because I cannot
question my own confident memory beliefs doesn’t mean that I cannot question someone else’s
claim concerning his confident memory beliefs. Recognizing that the other person cannot
question his own confident memory beliefs is not the same as accepting them as true, or even
generally true (O’Connor and Carr 1982: 140). A further worry about the second argument
for the analytic approach is that even if it were incoherent to question one’s confident memory
beliefs, this does not mean that one’s memory beliefs could not be consistently false. The skepti-
cal problem doesn’t get any better but instead gets worse, for not only is it possible that one’s
memory beliefs are consistently false but also one is not even in a position to coherently enter-
tain this possibility.
In my view, the most promising solution to the problem of verifying our ostensible memo-
ries is to reject epistemic internalism, that is, the view that all of the factors required for a belief
to be justified must be cognitively accessible to the subject and thus internal to his mind.11 The
skeptic about memory knowledge asks how one can know (or justifiably believe) that the mem-
ory experience on which one bases one’s memory report is in fact an accurate, a reliable, guide
to the past. This question is committed to internalism about justification. For unless it is deemed
reasonable to expect an epistemic subject to have insight into his justifying reasons, the skeptic’s
question is irrelevant. Epistemic externalism, by contrast, holds that some of the justifying factors
may be external to the subject’s cognitive perspective. Given externalism, we need not be able
to respond to the skeptical query. The fact that we don’t know (or justifiably believe) that our
memory reports amount to knowledge does not mean that they do not amount to knowledge.
A person who knows something does not have to know that what he has in his evidential base
amounts to knowledge. As long as he in fact satisfies the conditions of knowing something,
there is nothing more he has to do in order to know. No skeptical worry gets started.12
Notes
1 The labels for the two views are borrowed from Robins (2016: 432). In previous writings, I have
referred to the archival view as the xerox model of memory (2008: 144–6) and the identity theory of memory
(2010: 217–21). See note 8.
2 See Turri (2011) and Hazlett (2012). For an empirical investigation of Hazlett’s claims, see Buckwalter (2014).
3 Some non-factive verbs seem to allow for co-occurrences with wh-nominal complements. ‘What,’ for
instance, can introduce the verb-object of ‘believe’ as in “S may believe what I said as he may remember
60
Memory and truth
what I said.” Vendler writes: “The possibility of ‘believing what’ (= ‘that which’ or ‘the thing which’) is
restricted to things that can be objects of belief. For this reason, such sentences as ‘I believe what he lost’
are ruled out: the relevant co-occurrence set of ‘believe’ and ‘lose’, unlike those of ‘believe’ and ‘say’, do
not overlap. Roughly speaking,‘believe’ demands that-clauses, but ‘lose’ requires object nouns” (1972: 98).
4 Bernstein and Loftus (2009: 373). Similarly, Conway and Loveday (2015: 580) declare: “All memories
are to some degree false in the sense that they do not represent past experience literally . . . One of the
main functions of memories is to generate meanings, personal meanings, that allow us to make sense
of the world and operate on it adaptively. Memories are, perhaps, most important in supporting a wide
range social interactions where coherence is predominant and correspondence often less central.”
5 Schechtman (1994) argues that the weaving together, summarizing, and re-editing of parts of our own
past is a key means by which we produce and maintain continuity of identity over time.
6 Schacter (2001) distinguishes three kinds of omission errors—transience, absent-mindedness, blocking—
and four kinds of commission errors—misattribution, suggestibility, bias, and persistence.
7 Odell (1971: 593). See also Newby and Ross (1996: 205).
8 I have labeled this position the identity theory of memory (2010: 217–21).
9 Anything follows from a false antecedent and any conditional with a true consequent is true. What is
unsettling about these paradoxes of material implication is that in each of them the antecedent is the-
matically irrelevant to the consequent. The reason I analyze memorial authenticity in terms of relevant
entailment is so as to rule out some far-fetched entailments of one’s past thoughts as instances of mem-
ory. The notion of relevant entailment ensures that the content of the present propositional attitude is
not on a completely different topic than the content of the past propositional attitude. See Lewis (1998).
10 Shoemaker (1963: 229).The expression ‘perceptual and memory statements,’ as Shoemaker uses it, refers
not only to statements explicitly referring to memories but also to statements that are “directly based
on . . . memory, i.e., are putative reports of what the speaker . . . remembers.”
11 See Chapter 22 of this volume.
12 For comments on a previous draft, I am grateful to Kirk Michaelian.
References
Anscombe, G.E.M. (1981). The Reality of the Past. In G.E.M. Anscombe (ed.), Collected Philosophical
Papers 2: Metaphysics and the Philosophy of Mind (pp. 103–19). Oxford: Oxford University Press.
Bernecker, S. (2008). The Metaphysics of Memory. Dordrecht: Springer.
Bernecker, S. (2010). Memory: A Philosophical Study. Oxford: Oxford University Press.
Bernecker, S. (2015). Visual Memory and the Bounds of Authenticity. In D. Moyal-Sharrock, V. Munz,
A. Coliva (eds.), Mind, Language and Action (pp. 445–63). Berlin/New York: de Gruyter.
Bernstein, D.M. and Loftus, E.F. (2009). How to Tell if a Particular Memory is True or False. Perspectives
on Psychological Science 4: 370–74.
Brandt, R. (1955). The Epistemological Status of Memory Beliefs. Philosophical Review 64: 78–95.
Buckwalter, W. (2014). Factive Verbs and Protagonist Projection. Episteme 11: 391–409.
Conway, M.C. and Loveday, C. (2015). Remembering, Imagining, False Memories and Personal
Meanings. Consciousness and Cognition 33: 574–81.
Crombag, H.F.M., Wagenaar, W.A., and Van Koppen, P.J. (1996). Crashing Memories and the Problem
of ‘Source Monitoring.’ Applied Cognitive Psychology 10: 95–104.
De Brigard, F. (2014). Is Memory for Remembering? Recollection as a Form of Episodic Hypothetical
Thinking. Synthese 191: 1–31.
Hamilton, A. (1998). False Memory Syndrome and the Authority of Personal Memory-Claims: A
Philosophical Perspective. Philosophy, Psychiatry, and Psychology 5: 283–97.
Hawley, K. (2003). Success and Knowledge-How. American Philosophical Quarterly 40: 19–31.
Hazlett, A. (2010). The Myth of Factive Verbs. Philosophy and Phenomenological Research 80: 497–522.
Hazlett, A. (2012). Factive Presuppositions and the Truth Condition on Knowledge. Acta Analytica 27:
461–78.
Hume, D. (2000). A Treatise of Human Nature. Ed. D.F. Norton and M.J. Norton. Oxford: Oxford
University Press.
Intraub, H. and Richardson, M. (1989). Wide-Angle Memories of Close-Up Scenes. Journal of Experimental
Psychology: Learning, Memory, and Cognition 15: 179–87.
Klein, S.B. (2014). Autonoesis and Belief in a Personal Past: An Evolutionary Theory of Episodic Memory
Indices. Review of Philosophy and Psychology 5: 427–47.
61
Sven Bernecker
Koriat, A., Goldsmith, M., and Pansky, A. (2000). Towards a Psychology of Memory Accuracy. Annual
Review of Psychology 51: 481–537.
Lewis, D. (1998). Relevant Implication. In D. Lewis, Papers on Philosophical Logic (pp. 111–24). Cambridge:
Leyden, von W. (1961). Remembering: A Philosophical Problem. London: Duckworth.
Locke, D. (1971). Memory. London: Macmillan.
Malcolm, N. (1963). Knowledge and Certainty. Ithaca, NY: Cornell University Press.
Michaelian, K. (2011). Generative Memory. Philosophical Psychology 24: 323–42.
Michaelian, K. (2016). Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past.
Cambridge, MA: MIT Press.
Neisser, U. (1981). John Dean’s Memory: A Case Study. Cognition 9: 1–22.
Newby, I.R. and Ross, M. (1996). Beyond the Correspondence Metaphor: When Accuracy Cannot Be
Assessed. Behavioral and Brain Sciences 19: 205–6.
O’Connor, D.J. and Carr, B. (1982). Introduction to the Theory of Knowledge. Minneapolis, MN: University
of Minnesota Press.
Odell, S.J. (1971). Malcolm on ‘Remembering that.’ Mind 80: 593.
Price, H.H. (1969). Thinking and Experience. 2nd edition, London: Hutchinson.
Robins, S.K. (2016). Misremembering. Philosophical Psychology 29: 432–47.
Russell, B. (1921). The Analysis of Mind. London: George Allen & Unwin.
Russell, B. (1948). Human Knowledge: Its Scope and Limits. New York: Simon and Schuster.
Saunders, J.T. (1963). Skepticism and Memory. Philosophical Review 72: 477–86.
Schacter, D.L. (2001). The Seven Sins of Memory. Boston, MA: Houghton Mifflin.
Schechtman, M. (1994). The Truth about Memory. Philosophical Psychology 7: 3–18.
Schechtman, M. (1996). The Constitution of Selves. Ithaca, NY: Cornell University Press.
Shoemaker, S. (1963). Self-Knowledge and Self-Identity. Ithaca, NY: Cornell University Press.
Sutton, J. (2007). Philosophy and Memory Traces: Descartes to Connectionism. Cambridge: Cambridge
University Press.
Turri, J. (2011). Mythology of the Factive. Logos und Episteme 2: 143–52.
Vendler, Z. (1972). Res Cogitans: An Essay in Rational Psychology. Ithaca, NY: Cornell University Press.
Vendler, Z. (1980). Telling the Facts. In J.R. Searle, F. Kiefer, and M. Bierwich (eds.), Speech Act Theory
and Pragmatics (pp. 173–290). Dordrecht: Reidel.
62
5
MEMORY CAUSATION
Dorothea Debus
1. Introduction
The topic of “memory causation” is a core topic in the philosophy of memory and has attracted
a good amount of philosophical attention. At its most general, the claim which some philoso-
phers endorse in this context, and which other philosophers very emphatically and passionately
think to be false, is the claim that a philosophical account of memory will need to make some
causal claims; that is, that in order to offer a complete philosophical account of memory, we will
need to refer to causal relations of some kind. Accordingly, the debate on “memory causation” is
centered around the following question:
(Causal Question) Which role should reference to causation and to causal relations
play in an attempt to understand memory from a philosophical perspective?1
In order to address this question, we first have to determine what forms of memory are under
consideration here; indeed, we should distinguish between propositional, or “factual,” memo-
ries on the one hand, and experiential, or “recollective,” memories on the other.2 Experiential,
or “recollective” memories are those cases of memory which characteristically “correspond to
our use of the distinct senses” (Martin 2002: 403), whereas factual memories are those cases of
memory which do not have any experiential characteristics, but are beliefs or judgments of a
specific kind.
Both factual and recollective memories occur regularly in our everyday mental lives. For
example, Bob remembers that his mother was born on January 3, 1967; his memory takes the
form of a belief, and therefore should count as a factual memory. By contrast, when Emma
remembers meeting Sam for the first time at a party three years ago, Emma has an experience
as if seeing Sam again, smiling broadly, wearing a bright blue top; Emma’s memory of her first
encounter with Sam is a vivid, visual memory; it has experiential characteristics, and should
therefore count as a recollective memory. Everyday cases like the ones just sketched suggest
that we can successfully identify, and distinguish between, factual and recollective memories.3
Factual and recollective memories are different kinds of memory, but they have various
features in common. Indeed, all factual and all recollective memories are mental states or events;
that is, they are all elements of our mental lives. Both factual and recollective memories provide
63
Dorothea Debus
the remembering subject herself with some information, usually (but not necessarily) information
about the past, and in cases of both factual and recollective memory, the relevant information
which the memory provides the remembering subject with was acquired in an appropriate way at
some earlier time.4
In the case of recollective memories, the appropriate way of earlier information acquisition is for
the subject to have witnessed the event which is now recollectively remembered. For example,
Emma’s recollective memory of meeting Sam provides her with information (e.g., information
about what Sam looked like) which Emma acquired in an appropriate way (namely by means of
perception) at an earlier time (namely on the night when they did meet three years ago).
In the case of factual memories, having witnessed a relevant event in the past might count
as an appropriate way of information acquisition, but at other times having been told about
a certain fact by others might also be appropriate. For example, Bob’s factual memory of his
mother’s date of birth provides him with information (namely, about the date on which his
mother was born) which he was told about by his parents at some point in the past. More gen-
erally, therefore, it is plausible to accept that both factual and recollective memories provide the
remembering subject with information which the subject acquired in an appropriate way at an
earlier time.5
Furthermore, it also seems plausible to accept that in order for a subject to be said to remem-
ber something factually or recollectively, it is necessary that the information which the relevant
memory provides her with is (more or less) accurate. For example, assume that as a student, Bob’s
mother studied Biology, but Bob believes that his mother studied History; it seems that Bob’s
belief could not possibly count as a memory, because it is false. Similarly, if Emma has gone to
see a hypnotist who has brought it about that Emma now has a vivid experience as of having
encountered a pink elephant at a party three years ago, but Emma has never in her entire life
encountered a pink elephant at all (nor did Emma have a hallucinatory experience as of a pink
elephant during the party, nor did a pink elephant feature in any other way whatsoever at the
relevant party), then even if Emma herself mistakenly thinks that her current visual experience
as of encountering a pink elephant at the party presents her with how things were in the past,
the fact that Emma’s present experience provides her with inaccurate information about the rel-
evant past event gives us good reason to hold that Emma’s experience of a pink elephant should
not count as a recollective memory. Once more, this is so because in order for an experience
to count as a memory, it is necessary that the information which the experience provides the
subject with is (more or less) accurate, and Emma’s experience does not meet this condition.
More generally, we find that it is a part of our concept of “memory,” a part of how we use
the expression “to remember,” that in order for a certain mental state or event to count as a
memory, it is necessary that the information which the relevant mental state or event provides
the subject with is (more or less) accurate.
So far, then, we have found that both factual and recollective memories provide the remem-
bering subject with information which the subject has acquired in an appropriate way at an
earlier time, and in order for a subject to be said to remember something either factually or rec-
ollectively, it is necessary that the information which the relevant memory provides the subject
with is (more or less) accurate.
Furthermore, we also have reason to accept that in order for a mental state or event to
count as a factual or recollective memory, it is necessary that the information which the mental
state or event presently provides the subject with was acquired in an appropriate way at some
earlier time. For example, in order for Bob to be said to remember that his mother was born on
January 3, 1967, it is necessary that he acquired the relevant information at some earlier time.
It seems part of our concept of “memory,” part of how we use the expression “to remember”
64
Memory causation
that, had Bob not acquired the relevant information in an appropriate way in the past, his
current belief should not now count as a memory either. Similarly, in order for Emma to be
said to have a recollective memory of her first encounter with Sam, it is necessary that Emma
acquired the information which her current recollective memory now provides her with in an
appropriate way, namely by perceiving the relevant event, at the time when the relevant event
occurred. If Emma had not seen Sam at the party three years ago, her current visual experience
as of Sam smiling at her at the relevant party could not count as a memory of an encounter with
Sam at the time either.
More generally, we have found that in order for a current mental state or event to count as a
factual or a recollective memory, it is necessary that the information which the relevant mental
state or event provides the subject with now is (more or less) accurate, and that it was acquired
in an appropriate way at some earlier time in the past.
The question at the heart of the debate about “memory causation,” in turn, is the question
as to how the information which is provided to a subject when she has a factual or recollective
memory is linked to the relevant past event of information acquisition. Two different kinds of
answers are standardly offered in the philosophical literature: some say the relevant link is a causal
link, others say the relevant link is a link of (non-causal) information retention. C.B. Martin and
Max Deutscher are prominent defenders of the former view, and we will begin our considera-
tion of the topic of “memory causation” by considering some aspects of Martin and Deutscher’s
“causal theory of memory” in detail.
2. The causal condition

When contemporary philosophers talk about the topic of “memory causation,” sooner or
later they will refer to a seminal paper by Martin and Deutscher, published under the title
“Remembering” in 1966. At the time of writing, Martin and Deutscher’s paper is now entering
the fiftieth year of its publication, and their paper is arguably still the main point of reference for
any work on memory causation in contemporary philosophy. It should thus certainly count as
a “classic” point of reference in the philosophy of memory.
Martin and Deutscher’s paper offers a wide range of subtle distinctions, but its main point is
very straightforward. Using our own terminology as introduced in the previous section, we can
summarize their main claim as follows:
In order for a mental state or event to count as a factual or a recollective memory, it is
necessary that
(1) the information which the mental state or event provides the subject with is (more or less)
accurate;
(2) the information which the mental state or event provides the subject with was acquired by
the subject herself in an appropriate way sometime in the past; and
(3) the relevant past event of information acquisition “was operative in producing a state or
successive states in [the subject] finally operative in producing” the mental state or event in
question (Martin and Deutscher 1966: 166).
Following Martin and Deutscher, a defender of a causal theory of memory might hold that each
of these three conditions is necessary, and that taken together, they are also sufficient for a mental
state or event to count as a factual or recollective memory.
And indeed, as we saw in the introductory section of this chapter, we have good reason to
accept that conditions (1) and (2), as just set out, are necessary in order for a certain mental state
65
Dorothea Debus
or event to count as a factual or recollective memory. As we saw, it seems eminently plausible to

accept that in order for a certain mental state or event to count as a factual or recollective mem-
ory, it is necessary that (1) the information which the relevant mental state or event provides the
subject with is (more or less) accurate, and that (2) the information which the relevant mental
state or event provides the subject with was acquired by the subject in an appropriate way some-
time in the past. Thus, conditions (1) and (2) of the account are bound to be uncontentious.
The more contentious, and thus also the most important condition in Martin and Deutscher’s
account, and the reason why their account can appropriately be described as a “causal” theory
of memory, is condition (3), which says that in order for a present mental state or event to
count as a memory, it is necessary that an appropriate past event of information acquisition “was
operative in producing a state or successive states in [the subject] finally operative in producing”
(Martin and Deutscher 1966: 166) the present mental state or event in question. To rephrase
(and simplify) Martin and Deutscher’s point slightly, we might reformulate this as the suggestion
that in order for a present mental state or event to count as a memory, it is necessary that it be
(in some way) caused by an appropriate past event of information acquisition.
Why should we accept this? What speaks in favor of this causal condition? Martin and
Deutscher suggest we consider the following case:
A man whom we shall call Kent is in a car accident and sees particular details of
it, because of his special position. Later on, Kent is involved in another accident in
which he gets a severe blow on the head as a result of which he forgets . . . the first
accident . . . Some time after this second accident, a popular and rather irresponsible
hypnotist gives a show. He hypnotizes a large number of people, and suggests to them
that they will believe that they have been in a car accident at a certain time and place.
The hypnotist has never heard a thing about Kent nor the details of Kent’s accident,
and it is by sheer coincidence that the time, place, and details which he provides are
just as they were in Kent’s first accident. Kent is one of the group which is hypnotized.
The suggestion works and so, after the act is over, Kent . . . believes firmly that he
has been in an accident. The accident as he believes it to be is just like the first one in
which he was really involved.
(Martin and Deutscher 1966: 174)
It seems plausible to accept that after the hypnotist’s intervention, Kent meets conditions
(1) and (2) as set out earlier: Kent has various beliefs about an accident he was involved in, the
information which those beliefs provide him with is more or less accurate, and as it happens,
Kent did himself acquire the relevant information in an appropriate way (namely, by witnessing
it) sometime in the past. Nevertheless, it seems that Kent should not be said to remember the
accident—after all, he lost all relevant beliefs about the first accident during the second accident,
and he now endorses relevant beliefs only because of the hypnotist’s random intervention.
Further consideration of Kent’s case also shows that Kent does not, after the hypnotist’s
intervention, meet condition (3) of the account as set out earlier. For Kent believes what he
does “only because of the hypnotist’s suggestion” (Martin and Deutscher 1966: 174). The fact
that Kent witnessed the event, and acquired relevant information about the accident when it
took place, is not operative in producing Kent’s present beliefs about the accident; Kent’s present
belief is not due to, or caused by, any relevant past event of information acquisition; and “it is,”
so Martin and Deutscher rather plausibly suggest, “for this reason that [Kent] cannot be said to
be remembering the accident, despite the fact that he correctly recounts what he saw” (Martin
and Deutscher 1966: 174). If we accept this, we are also, more positively, committed to holding
66
Memory causation
that in order for a mental state or event to count as a memory, it is necessary that condition
(3) as listed earlier be met. Thus, we have reason to hold that, as condition (3) says, in order for
a certain mental state or event to count as a factual or recollective memory, it is necessary that
a relevant past event of information acquisition “was operative in producing a state or succes-
sive states in [the subject] finally operative in producing” (Martin and Deutscher 1966: 166) the
mental state or event in question.
We therefore find that the case of Kent, as described by Martin and Deutscher, might give us
good reason to accept a causal condition as part of a philosophical account of memory. Others
will strongly contest this, and we will consider their objections in due course. However, given
our everyday intuitions about Kent’s case it does, at least at first sight, seem rather plausible to
accept that condition (3) as formulated above is a condition which needs to be met in order
for a mental state or event to count as a memory, and for the time being I suggest we grant the
Causal Theorist this much for the sake of the argument.
3. Deviant causal chains

However, the account as it stands is not complete yet. Indeed, as Martin and Deutscher them-
selves point out, the account’s condition (3), the causal condition, will have to be developed
further. This becomes clear when we join Martin and Deutscher in considering another story
about Kent:
Kent has an accident, and tells his friend Gray about it. Then Kent has a second acci-
dent which destroys all his memories of the first accident, but Gray tells Kent all about
the first accident again, and on the basis of Gray’s testimony, Kent forms new, true
beliefs about the first accident. Kent eventually forgets that Gray told him about the
first accident; Kent still has true beliefs about the first accident, but it seems “clear that
[Kent] does not remember the [first] accident itself.”
This in turn presents us with a problem, because it seems that Kent does meet the causal theo-
rist’s three conditions for memory which we listed earlier; accordingly, if we were to hold that
the three conditions as stated so far are sufficient for someone to remember something, we would
have to say that Kent remembers the first accident now; but it seems eminently plausible to hold
that in the scenario as described, Kent does not remember the first accident now, and accord-
ingly, we need to modify the causal account of memory as stated earlier.
In order to do so, we should more carefully consider Kent and Gray’s story. Something seems
to be odd with respect to the causal chain which obtains between Kent’s acquisition of relevant
information at the time when the first accident occurred, and Kent’s present beliefs about the
relevant accident. Kent’s telling Gray about the first accident, Gray’s acquisition of the relevant
information, and Gray’s then retelling the story to Kent are essential parts of the relevant causal
chain in the story as told, but something seems “not quite right” with this causal chain when
we consider the question whether the beliefs which Kent presently has about the first accident
should count as memories. Indeed, the intuition that the beliefs which Kent presently has about
the first accident should not count as memories seems based on the observation that the causal
link which obtains between the original event of Kent’s acquiring relevant information during
the first accident on the one hand, and the beliefs which Kent now has about that accident on
the other, is “not quite right” for memory, but is rather “deviant” from what we would expect
relevant causal chains to be like.
67
Dorothea Debus
More generally, a causal theory of memory should therefore demand that the causal link
which, in order for a certain mental state or event to count as a case of memory, needs to obtain
between a relevant past event of information acquisition and the relevant present mental state or
event, must be a causal link of the right kind, and must not be a deviant causal link.
However, in order for this addition to the causal condition to be informative, we need to be
able to distinguish between causal links which are to count as being “of the right kind” and those
which are to count as “deviant.” Intuitively, it seems rather plausible to accept that something
is “not quite right” with the causal link which obtains in the case of Kent and Gray as sketched
earlier, that is, that the relevant causal link is somehow deviant; but then, which feature of this
particular causal link gives us reason to hold that the relevant causal link should count as devi-
ant; and more generally, how can we distinguish between deviant and non-deviant causal links?
One might respond that the characteristic feature of the causal link in Kent and Gray’s case,
as well as in other cases of deviant causal links, is the fact that the relevant causal link does
not “continue without interruption within the body of the person concerned” (Martin and
Deutscher 1966: 181). Accordingly, one might suggest that we can “fix” our earlier causal
account of memory by adding a clause which says that the relevant causal chain “should con-
tinue without interruption within the body of the person concerned” (Martin and Deutscher
1966: 181). However, as Martin and Deutscher (1966: 181) show, this clause is too strong.
Instead, Martin and Deutscher suggest that we can explain the difference between deviant
and non-deviant causal chains with the help of “the idea of a memory trace.”6 In order for
someone to remember something, so runs the suggestion, it is necessary that a relevant event
of information acquisition has left some “trace,” a trace which has been preserved and is caus-
ally relevant for the occurrence of the mental state or event which should count as a memory.
Indeed, Martin and Deutscher suggest that we need the idea of a “memory trace” in order to
develop a causal account of memory; for reference to some such “trace” provides us with the
best possible way of spelling out the claim that a present mental state or event is causally depend-
ent on a relevant past event of information acquisition. What is more, according to Martin and
Deutscher we can draw the distinction between deviant and non-deviant causal chains with
reference to “memory traces”7 by demanding that in order for a present mental state or event
to count as a case of memory, the causal link which obtains between a relevant past event of
information acquisition and the relevant present mental state or event must be dependent on a
continuous memory trace, and by pointing out that in cases of deviant causal chains, such as the
case of Kent and Gray, relevant memory traces are not continuous. Thus, according to the pre-
sent suggestion, a deviant causal chain is characterized by a discontinuous memory trace, whereas
a non-deviant causal chain is characterized by a continuous memory trace between the relevant
past event of information acquisition and a certain present mental state or event whose status as
a memory is under consideration. Martin and Deutscher (1966: 190f.) seem convinced that this
suggestion will solve the problem of deviant causal chains (although they highlight some prob-
lems which they do not discuss in their own paper), and Bernecker (2010: Chapter 5) develops
and defends this suggestion further.
However, in order to be able to find a successful way to distinguish between deviant and
non-deviant causal chains with the help of the idea of a “memory trace,” we have to determine
what should count as a continuous memory trace without in turn having to rely on an intuitive
understanding of the distinction between deviant and non-deviant causal chains. It remains
unclear to me whether this is possible, because it seems difficult to see how we could or should
decide in difficult cases whether or not a “continuous memory trace” obtains without relying on
some intuitive distinction between deviant and non-deviant causal chains, which in turn means
that there might be some reason to doubt whether the present suggestion is ultimately viable.
68
Memory causation
But be this as it may, in order to offer a causal account of memory we will, as we have seen,
have to hold that in order for a present mental state or event to count as a memory, it is necessary
that the causal link which needs to obtain between a relevant past event of information acquisi-
tion and the present mental state or event be of the right kind, that is, that the relevant causal link
not be deviant, and in order to formulate this condition fully, we have to be able to identify the
feature which differentiates deviant from non-deviant causal chains.8
4. The sufficiency claim

Whether or not it is possible to identify the feature which differentiates deviant from non-
deviant causal chains will in turn be of utmost importance for the causal theorist’s claim that the
causal theory of memory can formulate conditions which, taken together, are sufficient in order
for a present mental state or event to count as a memory. For if it turned out to be impossible to
identify a relevant differentiating feature, and if it thus turned out to be impossible to determine
what it takes for a causal link to be “of the right kind,” it would be impossible for a defender
of a causal account to offer sufficient conditions for a certain mental state or event to count as a
case of memory.
Of course, in the face of the present challenge, a defender of a causal account might well
retreat and give up on the sufficiency claim, but insist that even if they are not sufficient, the
conditions which we have discussed so far are all necessary conditions for memory, and that hav-
ing stated such necessary conditions is valuable in its own right because it certainly does further
our understanding of the phenomena at hand; but in its most ambitious version, a theory of
memory should not only offer some necessary conditions, but it should also offer a set of condi-
tions which jointly are sufficient conditions for memory, and this is what Martin and Deutscher
quite explicitly aim to do.
However, consideration of the problem of deviant causal chains might provide us with some
initial doubts as to whether the causal theorist will be able to offer a list of conditions which
taken together are sufficient for a certain mental state or event to count as a memory. In order
to assess the prospects of the causal theory’s sufficiency claim in greater detail, we should next
join Martin and Deutscher in considering the following exemplary case:
Suppose that someone asks a painter to paint an imaginary scene. The painter agrees
to do this and, taking himself to be painting some purely imaginary scene, paints a
detailed picture of a farmyard, including a certain colored and shaped house, various
people with detailed features, particular items of clothing, and so on. His parents then
recognize the picture as a very accurate representation of a scene which the painter
saw just once in his childhood. The figures and colors are as the painter saw them only
once on the farm which he now depicts.
(Martin and Deutscher 1966: 167f.)
We might grant that the painter “did his work by no mere accident” (Martin and Deutscher 1966:
168); but what exactly is happening here? How should we describe the case of the painter? Martin
and Deutscher hold that “Although the painter sincerely believes that his work is purely imagi-
nary, and represents no real scene, the amazed observers have all the evidence needed to establish
that in fact he is remembering a scene from childhood” (Martin and Deutscher 1966: 168).
However, this assessment of the situation might be contentious.9 It seems clear that the
painter witnessed the relevant scene in the past, he now paints the scene accurately, and it
also seems plausible to assume that there is some sort of causal connection between the painter
69
Dorothea Debus
witnessing the relevant scene in his childhood, and his presently painting the scene; but “why
say that the painter remembers the scene at all?” (Lewis 1983: 25). It is not obvious that we
should say that he does remember the scene—unless, of course, we already presuppose that the
causal theory’s three conditions are sufficient for an event to count as an instance of remem-
bering; but as this is the claim that is under consideration here, we should not presuppose it in
assessing the case of the painter. Quite on the contrary, as long as we do not already presup-
pose the truth of the causal theorist’s sufficiency claim, it seems rather questionable whether
the painter should be said to remember anything at all, which in turn means that the present
case might throw some further doubt on the claim that the three conditions offered by the
causal theory are sufficient for memory.
Indeed, as I show in Debus (2010), in order for an experience to count as an instance of
remembering, it is necessary that the relevant experience have “epistemic relevance” for the
subject; that is, that the experiencing subject be disposed to take the relevant experience into
account when thinking about the past, and when making judgments about what the past was
like. It seems odd to say that someone who has an experience as of a certain event which is
in fact a particular past event which they once witnessed, but who, asked about the relevant
past event, will not in any way rely on the relevant experience, should nevertheless be said to
remember the relevant past event. Quite on the contrary, consideration of such rather odd cases
gives us reason to hold that in order for an experience to count as a memory, it is necessary
that the experiencing subject be disposed to take the relevant experience into account when
thinking about the past, and when making judgments about the past. The painter as described
certainly does not give his experience any epistemic relevance, which in turn indicates that
his experience should not count as a memory either. Thus, we have reason to conclude that
the case of the painter as described by Martin and Deutscher highlights a problem with their
sufficiency claim.10
More generally, we find that our earlier considerations of the problem of deviant causal
chains, as well as our present considerations related to the epistemic role of recollective memo-
ries, might both give us very good reason to be at least skeptical with respect to the causal
theory’s claim to be offering a list of sufficient conditions for recollective and factual memories.
As we said earlier, in the face of such skepticism, a defender of the causal theory of memory
might retreat, give up on the sufficiency claim but continue to hold, somewhat more modestly,
that the conditions offered by the causal theory of memory are all necessary conditions for a cer-
tain mental state or event to count as a memory. However, as we will see in the next section,
even this claim has been vigorously attacked in the literature.
5. The necessity claim

A substantial amount of the discussion of the causal theory of memory has centered around
the question whether the causal condition, which is a constitutive part of any causal theory of
memory, is even necessary for a present mental state or event to count as a memory. Some fierce
opposition has been mounted to this claim, and opponents usually suggest that the link which
obtains between a relevant past event of information acquisition and a present mental state or
event which should count as a memory is not a causal link, but is rather a link of non-causal infor-
mation retention. A prominent defender of this view is Roger Squires, whose paper “Memory
Unchained” (1969) responds directly to Martin and Deutscher’s (1966) paper.
Squires aims to show first “that in describing something as having retained a quality, as
having stayed the same in a certain respect, there is no forced reference to causal connec-
tions” (Squires 1969: 178). For example, so he points out, “If some indigo curtains have kept
70
Memory causation
their color, [it] does not follow . . . that there should be any causal connection between their
previous and their present state” (Squires 1969: 178). He then moves on to show “that memory
is essentially the retention of knowledge” (Squires 1969: 185). Under the assumption that
the retention of a certain quality (e.g., the quality of having knowledge) is not necessarily
dependent on causal connections, we can, so Squires seems to suggest, also conclude that
memory does not necessarily depend on causal connections; thus, a relevant causal link is not
necessary in order for a mental state or event to count as a memory, which in turn means that
the causal theory’s causal condition is false.
However, while it might well be true that not all cases in which a quality is retained are
cases in which a causal connection obtains between earlier and later states of the object which
has retained the relevant quality, the specific retention of information which is characteristic
of memory might nevertheless be a retention relation which does depend on a causal relation
between a relevant past event of information acquisition and a present mental state or event.
Thus, the argument as sketched above is misleading (if not invalid). Accordingly, the defender
of the non-causal retention theory of memory has not so far shown convincingly that we should
think of memory as of non-causal retention.
Putting this problem to one side, we might, somewhat more positively, ask how precisely a
relation of “non-causal retention” between a past event of information acquisition and a present
mental state or event that should count as a case of memory could be further described. In cases
of memory, what sort of relation could a relation of non-causal retention between an appropri-
ate past event of information acquisition and a relevant present mental state or event possibly be?
It seems difficult to answer this question. Indeed, as Bernecker puts it, the claim that the relevant
relation is a relation of non-causal retention
appear[s] informative only as long as one refrains from asking what is involved in the
process of retaining [information]. When this question is raised, proponents of this
approach must concede that they don’t have a positive story to tell. The only thing
they do say about the retention process is that it is not of a causal kind.
(Bernecker 2010: 112)
Thus, the non-causal retention theory is not particularly appealing, because it lacks explanatory
power: It seems rather difficult to see how the non-causal retention theory might be developed
convincingly and in detail.
6. Why oppose the causal condition?

However, it might be useful to try to understand, more generally, why precisely some philoso-
phers seem so adamant in their view that our account of memory should not rely on anything
like the causal theory’s causal condition. Indeed, contemporary experimental psychologists usu-
ally take it for granted that memory depends on a causal link which obtains between a relevant
past event of information acquisition and the present mental state or event which is to count as
a memory; so why are some philosophers so fiercely opposed to this claim?
One possible response might be found (if only somewhat indirectly) in Squires’s paper itself.
Describing the main tenet of the causal theory of memory, Squires talks of the causal theorist’s
“belief that there is a causal chain, perhaps discoverable by scientists, which mysteriously links
the states of the [remembering person] over time,” but, so he contends, “There is no need for
such science fiction” (Squires 1969: 182). This comment suggests that Squires is opposed to the
causal theory of memory at least in part because he suspects the causal theorist of wanting to
71
Dorothea Debus
imitate the sciences, of somehow “giving in to” a scientific worldview, and then ending up with
developing some account that might be classified as “science fiction,” that is, as bad philosophy.
Now, it is certainly true that a philosophical account of memory which endorses the causal
condition will be easily compatible with a scientific exploration of memory; for as we said earlier,
experimental psychologists generally assume that when someone remembers something, there
is a causal link between relevant past events and the remembering subject. Thus, while philoso-
phers might offer arguments as to why a relevant causal link is necessary in order for a mental
state or event to count as a memory, empirical scientists might explore the question as to how
the relevant causal link is realized (e.g., on a neurophysiological level). Thus, Squires is certainly
right in intimating that a causal theory of memory will have certain affinities with a scientific
exploration of the same phenomenon. However, at least some people will think that such com-
patibility counts in favor of a relevant philosophical account; for it might seem rather appealing to
find an area of research where philosophers and experimental psychologists can work together in
understanding a phenomenon which is of joint interest to both groups. Thus, rather than being
a disadvantage, it might well be an advantage to develop a causal account of memory which is
compatible with a scientific account of the same phenomenon.
Another, more specific reason which might prompt some philosophers to oppose a causal
theory of memory might be based on the following train of thought:
(i) If we accept the causal condition put forward by the causal account of memory, we also have
to accept what we might call a “Cartesian” account of relevant mental states and events.
(ii) But we have good reason to reject any Cartesian account of mental states and events.
Hence,
(C) we have good reason to reject the causal condition put forward by the causal theory
of memory.
In order to get clear about this objection to the causal account of memory, we should first
get clear about what the opponent might mean by “Cartesianism” here. Following Child, we
might say that “Cartesianism” is “one way of trying to make sense of the causal elements in
psychology” (Child 1994: 2); Cartesianism
invoke[s] an ontology of mental states and entities, modelled on ordinary physical states
and entities; the internal mental phenomena [are] causally related to one another, and to
external phenomena, in just the way that physical phenomena are causally related to one
another . . . Beliefs, desires, [perceptual] experiences, [memories] and so on are internal
states and entities. Perception consists in an object’s appropriately causing an inner expe-
rience; [memory consists in a past event’s appropriately causing a current mental state or
event]; and so on. And the causal behaviour of those mental entities is regular and law-
governed, in just the way that the causal behaviour of physical things is governed by laws.
(Child 1994: 2)
What is more, so Child points out, “it is often assumed that this view of mental phenomena
as internal, causally interacting entities is the only way of making sense of causalism about the
mind” (Child 1994: 2), that is, it is often assumed that, just as the opponent’s premise (i) says, if
we accept a causal account of memory, we also have to accept what we might call a “Cartesian”
account of relevant mental states and events. However, this claim is false, which becomes clear
once we think more carefully about the relata between which, according to the causal theory of
memory, a causal relation obtains.11
72
Memory causation
On one possible view, the relevant relata are relevant past events of information acquisition,
and present mental states or events. This particular view might well entail “Cartesianism” about
the mind, just as the opponent’s above objection does suggest, because on this view, relevant
current mental states and events might well be conceived of as internal entities which themselves
stand in causal relations to past events and other current events.
However, this is not the only way in which we might identify the relata of the causal rela-
tion in memory. For alternatively, we might hold that in cases of memory, a causal relation
obtains between relevant past events of information acquisition and current “sub-personal”
states and events (typically neurophysiological states and events which obtain and occur in the
remembering subject’s brain). Those “sub-personal” states and events (i.e., typically neuro-
physiological states and events), so one might hold, are not identical with relevant mental states
and events which obtain and occur in the subject’s present mental life, that is, with relevant
memories which the subject presently has. Nevertheless, a subject’s memories will depend on
those sub-personal (usually neurophysiological) states and events in some way—for example,
a subject’s memories might be said to “supervene” on relevant sub-personal (usually neuro-
physiological) states and events. And even though relevant mental states and events themselves
are not the effects of relevant past events of information acquisition, so runs the present sug-
gestion, a subject who remembers something does stand in an appropriate causal relation to
relevant past events of information acquisition, because the relevant mental states or events
which are to count as memories depend on relevant sub-personal (i.e., usually neurophysi-
ological) states and events which are in turn caused in the appropriate way by relevant past
events of information acquisition.
There is no time to develop this suggestion in greater detail here, but I think it can be suc-
cessfully developed (and I have tried to do so for the case of recollective memory in Debus
(2008)). But then, under the assumption that the present suggestion can be successfully devel-
oped, we can also conclude that the opponent’s above objection to the causal theory of memory
is unsuccessful, because the first premise of the objection is false; it is not the case that if we
accept a causal account of memory, we also have to accept what we might call a “Cartesian”
account of relevant mental states and events, as the opponent claims. Quite on the contrary: we
can accept a causal account of memory while firmly rejecting a “Cartesian” account of relevant
mental states and events. Thus, we can certainly continue to endorse a causal theory of memory
in the face of the opponent’s objection.
More positively, our present considerations indicate that the causal theory of memory might
be developed in two different ways. Traditionally, a causal theorist of memory holds that when
a subject remembers something, a causal relation obtains between a relevant past event of
information acquisition and the present mental state or event which is to count as a memory.
However, so we have now seen, a causal theorist of memory might alternatively hold that in
order for a subject to be said to remember something, a causal relation needs to obtain between
a relevant past event of information acquisition and current sub-personal (i.e., usually neuro-
physiological) states or events (usually occurring in the brain of the subject who is said to
remember) which in turn somehow ground relevant mental states or events; those sub-personal
(usually neurophysiological) states or events are not identical with relevant memories, but the
subject’s memories somehow depend on the relevant sub-personal states or events, and it is in
virtue of their dependence on such appropriately causally linked sub-personal states or events
that the subject’s memories are causally linked with relevant past events of information acquisi-
tion. This alternative account of the relata between which the causal relation obtains in cases of
memory will leave plenty of room for various different accounts of the nature of mental states
and events quite generally, and of mental states and events which should count as memories
73
Dorothea Debus
more specifically. It most definitely does not commit us to a “Cartesian” view of mental states
and events, and it should therefore be perfectly acceptable by those who would find such a
“Cartesian” view of mental states and events unappealing.
7. Conclusion
As we said at the outset, the question which stands at the heart of recent philosophical discussions
of “memory causation” is the “Causal Question,” which asks:
(Causal Question) Which role should reference to causation and to causal relations
play in an attempt to understand memory from a philosophical perspective?
In an attempt to answer this question, we have meanwhile seen that we have good reason to
hold that reference to causation and to causal relations should play a central role in our attempts
at understanding memory from a philosophical perspective. Indeed, Martin and Deutscher offer
us good reason to hold that in order for a subject to be said to remember something, it is neces-
sary that a causal relation obtain between a relevant past event of information acquisition and a
relevant present state or event. According to Martin and Deutscher, the relevant present state
or event will be the relevant mental state or event which is to count as a memory itself; alterna-
tively, as we have just seen, a “less traditional” defender of the causal theory might hold that the
relevant present state or event be a relevant sub-personal (usually neurophysiological) state or
event upon which the relevant mental state or event depends in some way. Either way, it seems
plausible to accept that Martin and Deutscher have shown convincingly that a causal relation is
necessary in order for a current mental state or event to count as a memory, the causal account
of memory has much greater explanatory power than the alternative “non-causal retention
account” of memory, and a defender of the claim that a causal relation is necessary for memory
can successfully respond to various possible objections.
We therefore do have good reason to answer the “Causal Question” by holding that in order
for someone to be said to remember something, it is necessary that an appropriate causal relation
obtain between a relevant past event of information acquisition and the presently remembering
subject.
Notes
1 Analogous questions have been asked with respect to some other (mental) phenomena, prominently
perception, action and knowledge (see Child (1994: Chapter 3)).
2 See also Malcolm (1963) and Chapter 1 of this handbook.
3 There are other kinds of memory—e.g., cases of “habit memory,” such as remembering how to ride a
bike. “Habit memories” differ from factual and recollective memories in so far as they usually manifest
themselves in physical actions, such as someone’s actually riding a bike, and are not usually elements of a
subject’s mental life (see Russell 1997: Lecture IX, 166–8).
4 Many philosophers hold that memories “represent” facts, or past events, to the subject. I do not follow
this convention here, but instead express a similar point by saying that “memories provide a subject with
information,” because I argue in Debus (2008) that we should not think of recollective memories as of
representations at all. There is no need to argue for this here, but the terminology used should help us
to remain neutral on this point.
5 Which ways of information acquisition count as “appropriate” will vary across different kinds of
memories—e.g., as we just saw, in order for a subject to recollectively remember an event it is necessary
that the subject witnessed the relevant event in the past, whereas this is not necessary in the case of
factual memory.
74
Memory causation
6 “Memory traces” are considered in their own right in Chapter 6 of the present handbook.
7 See also Bernecker (2010), especially Chapter 5.
8 A similar challenge arises for other “causal theories,” such as the causal theory of action—see Davidson
(1980: 79ff.), and for a good summary of the problem see O’Brien (2015: 15–21).
9 The following draws on Debus (2010).
10 Martin and Deutscher’s own explanation for why their three conditions should count as sufficient is
disarmingly simple: “We can,” so they say, “offer no argument for the sufficiency of our list of crite-
ria other than the failure, after examination of cases, to find the need for a longer list” (Martin and
Deutscher 1966: 172). However, as we have just seen, at least for recollective memories an “epistemic
relevance condition” might well have to be added to a relevant list.
11 See also Child (1994), especially Chapters 3 and 4.
Further reading
Classic points of reference in any philosophical discussion of the problem of memory causation are Martin
and Deutscher (1966) and Squires (1969). Detailed contemporary assessments of the question which role
causation should play in an account of memory are offered by Bernecker (2010: Chapters 4 and 5), and
Michaelian (2016: Chapter 5).
References
Child, W. (1994). Causality, Interpretation and the Mind. Oxford: Clarendon Press.
Davidson, D. (1980). “Freedom to Act,” in his Essays on Actions and Events. Oxford: Clarendon Press, 63–81.
Debus, D. (2008). “Experiencing the Past: A Relational Account of Recollective Memory,” Dialectica 62:
405–32.
Debus, D. (2010). “Accounting for Epistemic Relevance: A New Problem for the Causal Theory of
Memory,” American Philosophical Quarterly 47: 17–29.
Lewis, D. (1983). “Dualism and the Causal Theory of Memory,” Philosophy and Phenomenological Research
44: 21–30.
Malcolm, N. (1963). “Three Lectures on Memory,” in his: Knowledge and Certainty. Englewood Cliffs, NJ
and London: Prentice Hall, 187–240.
Martin, C.B. and Deutscher, M. (1966). “Remembering,” Philosophical Review 75: 161–96.
Martin, M.G.F. (2002). “The Transparency of Experience,” Mind and Language 17: 376–425.
Michaelian, K. (2016). Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past. Cambridge,
MA: MIT Press.
O’Brien, L. (2015). Philosophy of Action. Basingstoke: Palgrave.
Russell, B. (1997). The Analysis of Mind. London: Routledge.
Squires, R. (1969). “Memory Unchained,” The Philosophical Review 78: 178–96.
75
6
MEMORY TRACES
Sarah K. Robins
1. Introduction
Memory traces feature in nearly every account of memory. They appear as birds in Plato’s
aviaries and images in Locke’s storeroom of ideas, as well as grooves in phonographic records,
pictures in a gallery, and textual and digital archives in the vast library of the mind.1 The per-
sistence of such metaphors reveals a long-standing commitment to the existence of memory
traces as well as a lack of clarity about their nature. What are memory traces, and why does
remembering appear to require them? I use these questions to guide the discussion below.
Section 2 provides a survey of views about the nature and features of memory traces. Section 3
introduces four distinct arguments for the necessity of memory traces and the challenges that
each account faces.
2. What are memory traces?

Memory is generally understood to be a diachronic capacity—the acquisition of information,
ideas, or experiences at one time, which is then available for recall at a subsequent time. It is
by preserving information that memory serves as a psychological constant, ideal for connecting
beliefs, experiences, intentions, obligations, and the like across time. The preservative charac-
terization of memory is what earns memory traces their pride of place in so many theories of
memory: traces are the vehicles through which preservation is made possible. Memory traces
provide a connection to the past by storing or somehow making available information about
and from one’s previous encounters. For this reason, traces are widely considered to be mental
representations: mental states whose content reflects the facts, faces, and feelings previously
encountered.2 They go by many names: memory trace, engram, memory image, representation,
etc.3 Which term one favors differs largely as a function of how one answers a set of questions
about memory traces, questions which I use to structure the remainder of this section.
Which types of memory involve memory traces?

Remembering comes in many forms—I can remember how to knit, the smell of my grandfa-
ther’s cologne, the capital of Argentina, and my ninth birthday party. Do all of these memories
76
Memory traces
involve traces, and is there a particular kind of trace for each kind of memory? Taxonomies
of memory often begin from a distinction between procedural memory and declarative memory—
the retention of motor skills (how to knit or ride a bike) versus the retention of information
(the capital of Argentina or that I received a bike on my ninth birthday).4 For some, the repre-
sentational nature of memory traces means that only declarative memories are trace-involving
(e.g., Tulving 2007). Others argue that all forms of retention, even basic learned behaviors,
require memory traces (e.g., Thompson 2005).
The most emblematic memory traces are those that correspond to a specific form of declara-
tive memory: memory for particular past events or experiences. These are often called episodic
memories, in contrast to semantic memories for facts and other pieces of general knowledge.
Episodic memories are the quintessential memories—when memory is likened to a photo
album or movie reel, it is episodic memories that supply the images. They include remem-
bering a visit to the Grand Canyon, the birth of your first child, or learning to drive a car.
Episodic memory traces are often thought to come with distinguishing features. Unlike seman-
tic memories (e.g., remembering that Buenos Aires is the capital of Argentina) a fact or facts
one may have encountered many times in many contexts, episodic memories are for one-off
experiences. And despite the limited window for retention, episodic memories often involve
exquisite detail, as well as perspective, imagery, and emotion. For this reason, many argue that
episodic memory traces have unique features—e.g., that they are stored in de se form (Burge
2003), require an image (Martin 2001), involve mental time travel (Tulving 2002), or are
cognitively integrated in epistemically relevant situations (Debus 2010). Others reject the idea
that any particular imagery or phenomenology, over and above retained information about the
event, is required for episodic traces (e.g., Martin and Deutscher 1966).
Discussion of this question has so far has focused on the role of traces in the kinds of
memory available to individual persons, but there may be other forms of memory that also
involve memory traces. Forms of remembering in non-human animals may be supported by
memory traces, depending on which kinds of memory one considers to be trace-involving
and which kinds of memory one thinks non-human animals possess. Such memory traces are
often termed engrams, reflecting the focus on the physical properties of information storage
(Thompson 2005).5 Similarly, there may be memory traces, or something analogous, that sup-
ports collective remembering: memories shared across a country, culture, community, or perhaps
between a lifelong couple (Wertsch 2002). For a community, a monument or other cultural
touchstone might serve as the memory trace for a particular tragedy or achievement. For a
couple, external mementos may provide the trace, or exogram, that allows them to complete,
together, the recollection of a shared encounter (Sutton and Windhorst 2009).
Are memory traces personal or subpersonal?

We can ask, further, about the cognitive level at which the appeal to memory traces is made.
Does the memory trace refer to the mental image I have while remembering (personal level)
or does the memory trace refer to the underlying cognitive or neural mechanism by which
remembering is made possible (subpersonal level)? There are accounts of memory traces as per-
sonal and as subpersonal. Each way of answering this question comes with its own challenges
for distinguishing memory traces—either from other mental states or from other mechanisms.
For personal level accounts, the distinction is to be made from the first-person perspective. For
subpersonal accounts, the distinction comes instead from the third-person perspective.
Those who favor a personal level characterization of memory traces require a way to distin-
guish between the mental images that are memory traces and other imagistic mental states—most
77
Sarah K. Robins
especially those of perception and imagination. Some achieve the distinction by appeal to con-
tent: memory traces are representations of the past (Aristotle), or representations whose content
is in the past tense (Locke 1971). Others choose to distinguish memory traces by their concomi-
tant feeling, like familiarity (Broad 1925) or intimacy (James 1890). Hume (1739), for example,
distinguished memory from imagination in terms of the vivacity of the former’s images. Still
others appeal to an accompanying belief or judgment, as Locke does when he claims that
memory is the revival of an image along with “the additional perception that it has had them
before” (Locke 1690/1975: 150).
Subpersonal accounts require a way of distinguishing memory traces from other cognitive
or neural mechanisms. According to Martin and Deutscher (1966), traces can be distinguished,
at least in part, by their causal history—i.e., when the trace was acquired relative to the content
it represents. Similar accounts distinguish memory traces by their causal contribution to the
re-represented information, requiring the trace to be an Insufficient but Necessary part of an
Unnecessary but Sufficient condition (INUS) condition for the later state (Bernecker 2010).
Others proceed functionally, aligning the trace with whatever neural changes occur during
learning and persist until remembering (Rosen 1975; Tulving 2007). Some go further and
identify memory traces with the brain location or neural process identified to play this role,
citing molecular processes as the basis for memory retention (Craver 2003; Bickle 2005).
How do memory traces represent the past?

When inquiring about memory traces, one may be interested in knowing not only that traces
represent one’s past experiences, but how they do so. This could be understood as a general
question about the semantic properties of memory traces as mental representations or a particular
question about the specific requirements of representing memories. Whether one’s interests are
general or specific, approaches to the question appeal to one or more of a series of related dis-
tinctions: traces are viewed as either static or dynamic, analogues or patterns, symbolic or connectionist,
representations or dispositions, localized or distributed. Accounts of memory traces as static stem
largely from archival models of memory, where memory is seen as a warehouse or repository
of well-preserved representations of past encounters. Dynamic accounts of memory traces, in
contrast, often accompany network or associationist views of memory—views according to
which information from the past is retained in a general and flexible way that can be adapted
to one’s ever-changing interests.6 The static and dynamic approaches to memory traces crosscut
the personal/subpersonal distinction introduced above. There are both personal and subpersonal
accounts of static traces. The static trace could be a mental image held in the mind’s eye during
remembering, a cognitive state with fixed semantic properties, or a dedicated neural pathway.
And the same is true for dynamic traces, which can be understood as dispositions to remember,
patterns in a semantic or connectionist network, or potential held in neurons (or, historically,
animal spirits—see Sutton 1998).
The aim of this section has been to survey the range of answers offered to a set of standard
questions about memory traces. How such questions are answered depends, of course, on why one
thinks memory traces are required in the first place. In the next section, we turn to that question.
3. Why does remembering require memory traces?

Although the appeal to memory traces is nearly universal in theorizing about memory, there
are many distinct reasons why the appeal is made. In this section, I introduce four of the most
prominent argumentative strategies used to establish the necessity of memory traces. The four
78
Memory traces
are not mutually exclusive; theorists may appeal to more than one. Each account of why
remembering requires traces is accompanied by further details of the features memory traces
must have in order to play the role that offers their legitimation. For each of the accounts sur-
veyed below, I sketch the general argument for memory traces and then identify a few of its
characteristic proponents. A brief discussion of objections and challenges follows at the end of
each section.
Representing the past

The first account of why remembering requires memory traces stems from consideration of
the following question: how can a person have a thought in the present moment about a past
event, idea, or experience, which—given that it is in the past—no longer exists?7 The reason-
ing from this question to the necessity of memory traces proceeds as follows. Memory is a
form of thought or judgment and so it must have an object. There is, moreover, an additional
requirement that the object of the thought be contemporaneous with the thought itself. For
perception, this works smoothly; but for memory, whose objects are in the past, the require-
ment presents a problem. The objects of memory no longer exist; they are not available to serve
as the object of occurrent thought. There must be something else, something that does exist in
the present moment, which can stand in place of the past event. This something is the memory
trace. In short, memory traces are required to serve as the object of the remembering experi-
ence, a surrogate for the past event that is no longer available.
This view of memory, and memory traces—sometimes referred to as representationalism
(De Brigard 2014b)—is found amongst those for whom the mind comprises mental images,
serving as a storehouse for previous sensory impressions and ideas. Origin of this account is often
attributed to Aristotle, and is perhaps most familiar from empiricist views of the mind during the
early modern period, especially the writings of Hobbes, Locke, Hume, and Mill. In a later articu-
lation of the view, Russell expresses the central idea succinctly: “memory demands an image”
(1921: 186). Representationalism requires memory traces to be direct replicas of the past events
or ideas for which they stand. It is only by this resemblance that they can fulfill their surrogacy.
For Aristotle, the memory trace is not simply an image, but an eikôn: “a likeness or copy of the
thing remembered” (Sorabji 2004: 2). Similarly, Locke characterizes memory as the revival of a
previous idea (1694/1979) and Hume questions, “for what is the memory but a faculty, by which
we raise up the images of past perceptions?” (1739/1978: 260).8
The representational account of memory traces is not without its difficulties. Proponents of
the view must find a way to distinguish memory traces from other mental images (as discussed in
the earlier section “Are memory traces personal or subpersonal?”) and no candidate criterion is
foolproof.9 There are also two broader objections to this account of memory traces: (1) immediacy
and (2) explanatory relevance.
The immediacy objection is offered by those who take issue with the appeal to mental states
(images or ideas) as the objects of thought. Immediacy—a direct connection between one’s
mental capacities and the world—is required, either to avoid skepticism or to avoid implausi-
ble ideas about remembering and the world’s structure. The first form of the objection is most
familiar in Reid (1785/1951), who challenges both Locke and Hume’s views of memory. There
is no sense to be made of memory traces as the same ideas from perception, Reid argues, and
more importantly, such traces would provide no guarantee of their origin by which knowledge
about the world could be secured, making skepticism inevitable.10 The alternative is a form
of commonsense realism about memory, according to which it is a capacity that allows direct
acquaintance with the past. Memory is not an act of re-perceiving or re-experiencing; no mental
79
Sarah K. Robins
image is required to serve as intermediary.11 Continuation of this commonsense approach can

be seen in Wittgenstein’s asking, incredulously, “why must a trace have been left behind?”
(1981: no. 610). Malcolm (1970) extends the worry further, arguing that there is no need for
a memory image in remembering. Reflecting on uses of memory reveals that it often proceeds
(successfully, even) without any such intermediary. What’s more, even in cases that do involve
imagery, the image has no fixed content or structure. Remembering the previous night’s dinner
party, for example, could involve an image of the set table, or of the friends as they arrived, or an
auditory echo of the lively conversation; no one candidate has any claim to being the structure
of the past event, much less the recurrence of previous perception (Malcolm 1970: 67). Instead,
memory is unmediated access to things previously learnt (and not forgotten).
The explanatory irrelevance objection challenges the explanatory role that memory traces
play in representationalism. The claim is that the appeal to memory traces presupposes what
is to be explained, and so leads to either circularity or infinite regress, depending on how the
challenge is fleshed out. The problem is a form of Meno’s paradox of inquiry turned inward.
Representationalists claim that memory traces must be replicas of the events for which they
stand, but this alone does not explain how an event is remembered. Remembering requires,
further, the ability to identify the desired memory trace from amongst all the other traces
stored in memory. How does one know which memory trace is to be selected? If one already
knows which trace corresponds to the event that she wants to remember, then it is this
knowledge—and not the trace—that explains her success in remembering. Some stop here
and proclaim representationalism to be circular: memory traces are used to explain remem-
bering; but appealing to traces presupposes remembering (i.e., remembering which trace is
which: Heil 1978; Bennett and Hacker 2003). Others illustrate the objection as an infinite
regress: if remembering always requires a trace, then one’s ability to remember which trace
one wants to select must itself have a trace, one that explains when and how to go in search
of the former trace. Explaining how this trace is retrieved requires the introduction of yet
another memory, and in turn, another memory trace. And so on. If, at any point, the repre-
sentationalist attempts to halt the regress, by stipulating that some forms of remembering do
not involve traces, then the initial motivation for appealing to traces to explain remembering
is lost (Malcolm 1970).
Preventing causal gaps

A second, distinct approach to memory traces develops out of a general commitment to mem-
ory as a causal process—that is, to the idea that remembering requires that one’s representation
of a past event be because of or due to that event in some causal sense. As Anscombe puts the
point, “The original witnessing of an event is a cause of any present memory (even a false
memory) of that event” (1981: 130).12 This line of thought is the basis for a view known as
the Causal Theory of Memory (Martin and Deutscher 1966; Bernecker 2010; Debus 2010).
Motivation for the causal theory comes from recognition that memory requires more than sim-
ply an accurate representation of one’s past experience. This condition alone is not enough to
guard against the possibility of veridical confabulation: serendipitously re-encountering infor-
mation that was either never fully learned or had been forgotten. Remembering requires, in
addition, a causal connection.
Memory traces enter into the causal theory via the assumption that action at a distance is
impossible. The past event and its subsequent remembering are separated by a temporal gap.
Since causation requires contiguity, the only way for the past event to be the cause of the
80
Memory traces
remembering is for there to be a memory trace—or series of memory traces—that form an

uninterrupted causal chain between the two events. Causation must be gapless; hence memory
causation requires memory traces.
Amongst supporters of the causal theory of memory, there are differences in the particulars of
how this general line of argument is used to secure the existence of memory traces. For some,
contiguity (i.e., gapless causation) is a constraint on all causal processes (Martin and Deutscher
1966); others restrict their contiguity requirement to memory causation alone (Bernecker
2010). There are further differences between those who view the contiguity requirement as an
a priori demand on causal relations (Shoemaker 1970) and those who view it as an empirical
discovery (Rosen 1975). Proponents of the causal theory agree that memory traces are necessary
for remembering, but not all believe they are sufficient (see, e.g., Debus 2010). The focus on the
causal role of memory traces emphasizes their physical nature. Memory traces must be the kind
of physical states that could prevent gaps in the causal chain: for many, these are brain states,
but others caution against this assumption and allow that various prosthetics or external objects
could serve as memory traces (Martin and Deutscher 1966; Bernecker 2010).
The causal justification of memory traces faces its own set of objections and challenges.
First, there are some who support a broadly anti-causal understanding of the mind, and use
this to argue against the view of memory as causal and the further requirement that memory
traces are needed to facilitate this causal process. Squires (1969), for example, claims that
there are many instances in which a non-causal interpretation of retention is most apt, as
when a garment retains its hue or an object acquires and loses an ability at different points
in the year, depending on the weather. Second, there are some who support the claim that
memory is a causal process, but deny that this entails the existence of memory traces. Russell
(1921) advocated for the possibility of mnemic causation—the view that a past event could
cause a later remembering without any intermediary. Russell did not deny that intermediary
memory traces could be discovered; he argued only that their discovery was not guaranteed
by the nature of causation:
There is . . . no a priori objection to a causal law in which part of the cause has ceased
to exist. To argue against such a law on the ground that what is past cannot operate
now, is to introduce the old metaphysical notion of cause, for which science can find
no place.
(Russell 1921: 89)
Additional difficulties can be found within the causal theory of memory. Stipulating that mem-
ory is a causal process, sustained by memory traces, is only the first step. Much of the remaining
work lies ahead, in clarifying the particular relationship between the past event and its subse-
quent remembering that is constitutive of memory causation. As with causal analyses of other
concepts (like perception, action, and knowledge), the causal theory of memory must address
worries that arise from the possibility of deviant causal chains—causal relations between the past
event and its representation that are not constitutive of remembering. There are concerns about
prompts that pre-empt the act of remembering (Martin and Deutscher 1966), contents that
change over time (Michaelian 2012), and influences of other non-memorial cognitive processes
(Bernecker 2010), to name only a few of the deviant cases troubling the causal theory. One par-
ticular type of wayward causation, relearning, is so critical that its consideration is often thought
to provide its own justification for the existence of memory traces. The relearning argument for
memory traces is considered in the next section.
81
Sarah K. Robins
Distinguishing remembering from relearning

Another route to the need for memory traces comes from consideration of what remem-
bering requires. On this account, memory traces emerge as a requirement for distinguishing
remembering from a similar mental state: relearning. Relearning happens when one acquires
information, forgets it, and then re-acquires it from another source. Relearning can occur eas-
ily, as forgetting is common and there are many other ways for information from past events
to be retained—diaries, digital servers, photos, videos, etc. Relearning is a particular challenge
because it is not an error in the traditional sense. If you have forgotten something, then it is good
to re-acquire that information from another source. However, relearning is not remembering.
Distinguishing between the two is thought to require appeal to memory traces. And since the
appeal to memory traces is essential to excluding cases of relearning from the analysis of remem-
bering, remembering requires a memory trace.
Relearning is often introduced by example. Suppose a person has a particular experience—
say, witnessing a car accident—and afterward develops retrograde amnesia (erasing memories
of the accident and other past events). Then suppose further that information about the acci-
dent is made available to them by other means (e.g., from a diary entry or from a friend who
also witnessed the event). Both Martin and Deutscher (1966) and Bernecker (2010) introduce
hypothetical cases of this type. Relearning can also be illustrated by documented cases of post-
traumatic amnesia. For example, physicians describe the case of an individual, ML, who has no
memory for life events prior to his accident. He is, however, able to acquire and retain new
information, including the names of his doctors, and to relearn facts about the past, including
events from his own life (Kapur 1993).
Distinguishing remembering from relearning is difficult because the difference is not often
detectible to the rememberer herself. One can retain information and remain uncertain of its
source, as recollections from early childhood illustrate:
A person has an apparent recollection of something from early childhood, and wonders
whether he really remembers it. His parents may tell him that what he describes did hap-
pen, and that he witnessed it, but the discussion of whether he remembers it still goes on.
They wonder whether his witnessing of the event has any connection with his now giving
the story or whether his description can be completely explained by what he heard later.
The distinction between remembering and relearning is often cast as a difference between two
ways that a past event could cause a later representation of that event, and so this argument
for the existence of memory traces is often made from within the causal theory of memory.
The general need for the distinction has, however, been recognized at least since Aristotle
(see Sorabji 2004).13 Both remembering and relearning involve a causal connection between
the past event and its subsequent representation. The general causal condition on memory must
be modified so as to require that the causal connection stay within the person, in some important
(if difficult to articulate) sense. That is, the difference between remembering and relearning is
that, in the case of the former, the causal connection is sustained by a memory trace.
There is an important worry for the relearning approach to memory traces—namely, that
it is difficult to explain how memory traces support this distinction. Simply stating that remem-
bering involves memory traces while relearning does not fails to provide an independent sense
of what memory traces are. The issue is further complicated because many who support this
approach to the distinction recognize how few stipulations can be placed on what counts as a
82
Memory traces
memory trace. For instance, memory traces might often be neural states, but many hesitate to
make this a defining feature, as it is plausible that traces may someday be sustained by neural
prosthetics (Bernecker 2010). Being in contact with the body will also not do as a requirement,
since some forms of relearning involve sustained contact with the body as well (Martin and
Deutscher 1966).14 What’s more, contact with the body of the rememberer may not be neces-
sary for remembering either. It is conceivable that members of another species could store their
memories externally, or that other technological advances will make possible external storage.
Martin and Deutscher urge this consideration via a hypothetical case where we discover a class
of beings that remember through the use of a metal box that they carry around. In such a case,
“they do not ask the box questions about the past, but when they are connected with the box
they remember as we do” (1966: 181).15
The distinction between remembering and relearning matters. It is the distinction between
eyewitness testimony and hearsay, between a perfect exam that results from studying and one
that relies on a notecard tucked into a shirtsleeve. This does not mean, however, that the
difference between the two is easily—or ever—detectible. Relearning is, after all, a form of
retention, just not the right kind. Remembering and relearning may differ in terms of when
the information was acquired, but there needn’t be any telling mark of this difference in their
causal histories. Both forms of retention could be supported by the same neural or cognitive
mechanism and could involve the same mental imagery and feeling. Additional worries of this
sort come from consideration of how memory traces represent past events. If memory traces are
distributed and blended together, then this would preclude any way of sorting between traces
in terms of when they were acquired (Robins 2016).
Making sense of memory science

The final account of why remembering requires memory traces involves appeal to scientific
practice. The claim that remembering is sustained by a memory trace is a guiding assumption
of empirical inquiry into the memory process. And, most importantly, it is an assumption that
appears well confirmed by these inquiries.
Across memory science, there is a widespread idea that remembering involves three pro-
cesses: encoding, storage, and retrieval. Encoding refers to the process by which information
makes its way into memory. One must actually attend to a piece of information in order for
it to be a candidate for remembering. Storage refers to the process by which information is
maintained in memory, and retrieval is the process by which information is recovered from
memory. It is rare to find an introduction to memory that does not begin with a description
of these three processes. The Encoding-Storage-Retrieval, or E-S-R, model of memory has
been described as the “pretheoretical orienting framework within which a systematic analy-
sis of memory can be pursued” (Schacter 2001: 138).16 Traces are at the heart of the E-S-R
model, which provides a functional characterization of memory traces: traces are the cognitive
and neural changes that result from encoding and are stored until retrieval. Consider, for example,
Tulving’s characterization of remembering:
An event happens, a person experiences it, memory traces are laid down representing
the event, the past vanishes and is replaced by the present. The memory traces of the
event continue to exist in the present, they are retrieved, and the person remembers
the event. This, in a nutshell, has been the understanding of how memory works. It is
simple and straightforward; there is no need or room for magic, or marvel.
(Tulving 2002: 19)
83
Sarah K. Robins
The widespread commitment to the trace theory in memory science is perhaps best illustrated
by the fact that skeptics about the existence of memory traces continue to endorse the trace as
a requirement for remembering. Karl Lashley once argued that the engram did not exist, after
a series of experiments failed to identify any area of the rat or monkey brain that was required
for retention of the ability to run mazes or perform other similar tasks. Even as he reached this
skeptical conclusion, Lashley continued to assume that remembering requires traces, writing
“I sometimes feel, in reviewing the evidence on the localization of the memory trace, that the
necessary conclusion is that learning just is not possible” (Lashley 1950: 477–8).
Memory scientists continue to assume that the functional description of memory traces,
as provided by the E-S-R model, will serve as a guide to the cognitive and neural mecha-
nisms that constitute the trace, or engram. Lashley’s inability to locate the trace is attributed
to imprecision, remediated via clear specification of the encoding and retrieval conditions
and systematic interventions into the various components of the underlying neural cir-
cuit (Thompson 2005). Such regimentation has resulted in the discovery of Long-Term
Potentiation, or LTP.17 LTP is a change to the synaptic connections between two neurons
that occurs as a result of their previous interactions. Many memory scientists believe that LTP
is an important component of memory, offering a neurological complement to the experi-
ence of remembering.18 In remembering, one undergoes an experience and is changed by it;
this change is then reflected in a modification to the connections between the neurons in her
brain that were active during the experience. These changes persist, but are not evident until
these neurons are re-activated in a subsequent experience of remembering. Studies of animal
memory lend further credence to this idea. One can, for example, predict a rat’s path through
a maze by recording from the neurally encoded “spatial map” in its hippocampus (Wilson and
McNaughton 1993). The rat’s hippocampus engages the same firing pattern repeatedly after
the rat runs the maze, and the hippocampus generates similar maps when the rat is returned
to the same maze again (Diba and Buzsaki 2007). Further studies of the molecular processes
supporting memory allow for electrophysiological and pharmacological interventions. One
can also stimulate the underlying neural mechanism to produce the remembered behavior
(Chapman et al. 1988), as well as intervene to inhibit the engram, which degrades, erases, or
distorts the memory (Ramirez et al. 2014).
Memory scientists often take the discovery of the neural mechanisms involved in remem-
bering as vindication of the assumed trace requirement. As more evidence regarding the neural
mechanisms of remembering is accumulated, the question of whether there are memory traces
recedes further into the distance. Within memory science, little attention is given to the justi-
fication of this commitment to memory traces. The question is not whether there are memory
traces, but rather how they work (Dudai 2002).
For the view of memory traces that is compelled by scientific practice, the biggest challenge
is internal—from discoveries within memory science. Evidence of the persistence and pervasive-
ness of memory errors makes trouble for the idea that memory is sustained by the storage and
retrieval of memory traces. Memory errors can be sorted into two general types: errors of omis-
sion and errors of commission. Omissions are forgetting errors. While perhaps unfortunate when
they occur (but see Michaelian 2011), they do not pose a threat to the central role of traces in a
theory of memory. Forgetting occurs when a particular trace has been misplaced, or has become
degraded, or has been lost altogether. The existence of such errors is compatible with successful
remembering being the result of memory traces (i.e., those that are not missing or malformed).
Errors of commission—the production of false memories—are more troubling. Producing
representations of past events that are inaccurate, either in part or in total, suggests more
systematic disruption of or possible disorganization within the memory system. Evidence
84
Memory traces
from memory science shows that the representations produced in remembering often contain
information from many sources, even if they feel to the rememberer as if they depict a
particular past event (Brainerd and Reyna 2005). What’s more, the amount of detail con-
tained in any alleged memory is a poor guide to its veracity. In remembering, confidence and
accuracy are orthogonal. People are often confident in their reports of details that were not
part of their previous experience, and may even retain this confidence in light of contraven-
ing evidence (Loftus 2003). These results threaten the belief that memory traces come with
characteristic features that signal when remembering is occurring. And they suggest, at a
minimum, that the content of the memory trace is malleable.19
False memories are most troubling, however, not because of the type of error that they are,
but because of the frequency with which they occur. These errors are readily and repeatedly
observed in both laboratory and naturalistic settings. They occur so often, in fact, that some
have suggested that preservation of a particular memory trace for a specific idea or event is
either rare or impossible. It would be strange to characterize memory as a capacity that aims
for storage and retrieval if it rarely manages to keep memories of particular past events intact.
For this reason, views of memory traces on the dynamic end of the static–dynamic dichotomy
may be better suited to accommodating this evidence (Michaelian 2012). Or it may be that this
evidence offers a reason to reject memory traces—or the distinctness of memory as a capacity—
altogether (De Brigard 2014a). The need to reconcile views of memory traces with evidence
of the form and frequency of memory errors is one of the most interesting and difficult issues
facing contemporary theories of memory.
Notes
1 See Draaisma (2000) for a comprehensive overview of the history of memory metaphors.
2 See Rosen (1975) for a non-representational, “logical” account of memory traces.
3 Tulving (2007) provides an overview of these terms and the slight variations in connotation.
4 See Squire (2004) for a more complete taxonomy.
5 The term “engram” was first coined by Semon (1904).
6 See Sutton (1998) for an overview of these two approaches to memory traces and a comprehensive
history of the dynamic view.
7 Aristotle’s view of memory, for example, is guided by a question of this sort (450a25, see Sorabji 2004
for discussion).
8 Neisser (1967) labels this theory of memory, especially as seen in Locke and Hume, the “Reappearance
Hypothesis.”
9 See Bernecker (2008: Chapter 6) for a review of why various memory markers fail.
10 Reid asks, of the memory trace,“What mark does it bear of the date of its archetype?” (1785/1951: 476).
11 See Copenhaver (2006).
12 See also Ayer (1956), Shoemaker (1970), and Armstrong (1987).
13 It is possible that one could appeal to memory traces as the difference between remembering and
relearning without also endorsing the causal interpretation of memory. Being supported by memory
traces could be the difference between these two mental states without that difference being a causal one.
14 Such a premise is at work in the movie Memento, where the protagonist Leonard compensates for his
inability to form memories by tattooing facts about his life onto his body.
15 For further discussion of the complications that arise from such external memories, see Sutton and
Windhorst (2009).
16 Some trace the E-S-R model back to the work of Köhler (1947), others to Melton (1963).
17 Kandel was awarded the Nobel Prize in 2000 for his work on long-term potentiation (LTP) in the gill-
withdrawal reflex of the snail Aplysia. For comprehensive historical overview of the discovery of LTP,
see Craver (2003).
18 But see Sullivan (2010).
19 Some go even further and suggest that the mechanisms of memory are flexible too, and may reconsoli-
date each time the trace is activated (Nadel and Moscovitch 1997).
85
Sarah K. Robins
References
Anscombe, G. E. M. (1981). Memory, experience, and causation. In G. E. M. Anscombe (ed.), Collected
Philosophical Papers, Vol. II: Metaphysics and the Philosophy of Mind (pp. 120–30). Oxford: Oxford
University Press.
Aristotle. De Memoria et Reminiscentia (W. D. Ross, trans., 1955). Oxford: Clarendon Press.
Armstrong, D. M. (1987). Mental concepts: The causal analysis. In R. L. Gregory (ed.), The Oxford
Companion to the Mind (pp. 464–5). Oxford: Oxford University Press.
Ayer, A. J. (1956). The Problem of Knowledge. Harmondsworth: Penguin.
Bennett, M. and Hacker, P. M. S. (2003). Philosophical Foundations of Neuroscience. Oxford: Basil Blackwell.
Bernecker, S. (2008). The Metaphysics of Memory. New York: Springer.
Bickle, J. (2005). Molecular neuroscience to my rescue (again): Reply to Looren de Jong and Schouten.
Philosophical Psychology, 18: 487–94.
Brainerd, C. J. and Reyna, V. F. (2005). The Science of False Memory. Oxford: Oxford University Press.
Broad, C. D. (1925). The Mind and Its Place in Nature. London: Routledge and Kegan Paul.
Burge, T. (2003). Memory and persons. Philosophical Review, 112: 289–337.
Chapman, P. F., Steinmetz, J. E., and Thompson, R. F. (1988). Classical conditioning does not occur
when direct stimulation of the red nucleus or cerebellar nuclei is the unconditioned stimulus. Brain
Research, 442: 97–104.
Copenhaver, R. (2006). Thomas Reid’s theory of memory. History of Philosophy Quarterly, 23: 171–89.
Craver, C. F. (2003). Interlevel experiments and multilevel mechanisms in the neuroscience of memory.
Philosophy of Science, 69: 83–97.
De Brigard, F. (2014a). Is memory for remembering? Recollection as a form of episodic hypothetical
thinking. Synthese, 191: 155–85.
De Brigard, F. (2014b). The nature of memory traces. Philosophy Compass, 9: 402–14.
Debus, D. (2010). Accounting for epistemic relevance: A new problem for the causal theory of memory.
American Philosophical Quarterly, 47: 17–29.
Diba, K. and Buzsaki, G. (2007). Forward and reverse hippocampal place-cell sequences during ripples.
Nature Neuroscience, 10: 1241–2.
Draaisma, D. (2000). Metaphors of Memory: A History of Ideas about the Mind. Cambridge: Cambridge
University Press.
Dudai, Y. (2002). Memory from A to Z: Keywords, Concepts and Beyond. Oxford: Oxford University Press.
Heil, J. (1978). Traces of things past. Philosophy of Science, 45: 60–72.
Hume, D. (1739/1978). A Treatise of Human Nature, L. A. Selby-Bigge (ed.). Oxford: Clarendon.
James, W. (1890). The Principles of Psychology. London: Macmillan.
Kapur, N. (1993). Focal retrograde amnesia in neurological disease: A critical review. Cortex, 29: 217–34.
Köhler, W. (1947). Gestalt Psychology. New York: Liveright.
Lashley, K. (1950). In search of the engram. Symposium on Explorations of Biology, No. 4. (pp. 454–82).
Cambridge: Cambridge University Press.
Locke, J. (1694/1979). Essay Concerning Human Understanding, P. H. Nidditch (ed.). Oxford: Clarendon.
Loftus, E. F. (2003). Our changeable memories: Legal and practical implications. Nature Reviews:
Neuroscience, 4: 231–4.
Malcolm, N. (1970). Memory and representation. Nous, 4: 59–70.
Martin, M. G. F. (2001). Out of the past: Episodic recall as retained acquaintance. In C. Hoerl and
T. McCormack (eds.), Time and Memory: Issues in Philosophy and Psychology (pp. 257–84). Oxford:
Clarendon Press.
Martin, C. B. and Deutscher, M. (1966). Remembering. Philosophical Review, 75: 161–96.
Melton, A. W. (1963). Implications of short-term memory for a general theory of memory. Journal of
Verbal Learning and Verbal Behavior, 2: 1–21.
Michaelian, K. (2011). The epistemology of forgetting. Erkenntnis, 74: 399–424.
Michaelian, K. (2012). Generative memory. Philosophical Psychology, 24: 323–42.
Nadel, L. and Moscovitch, M. (1997). Memory consolidation, retrograde amnesia, and the hippocampal
complex, Current Opinion in Neurobiology, 7: 217–27.
Neisser, U. (1967). Cognitive Psychology. New York: Appleton-Century-Crofts.
Ramirez, S., Tonegawa, S., and Liu, X. (2014). Identification and optogenetic manipulation of memory
engrams in the hippocampus. Frontiers of Behavioral Neuroscience, 7: 226.
86
Memory traces
Reid, T. (1785/1951). Essays on the Intellectual Powers of Man, A. D. Woozley (ed.). London: Macmillan
and Co.
Robins, S. K. (2016). Representing the past: Memory traces and the causal theory of memory. Philosophical
Studies, 173(11): 2993–3013. doi:10.1007/s11098-016-0647-x.
Rosen, D. A. (1975). An argument for the logical notion of a memory trace. Philosophy of Science, 42: 1–10.
Russell, B. (1921). The Analysis of Mind. London: Allen and Unwin Limited.
Schacter, D. (2001). The Seven Sins of Memory. New York: Houghton Mifflin.
Semon, R. (1904). Die Mneme. Leipzig: W. Engelmann.
Shoemaker, S. (1970). Persons and their pasts. American Philosophical Quarterly, 7: 269–85.
Sorabji, R. (2004). Aristotle on Memory (2nd edition). London: Duckworth.
Squire, L. R. (2004). Memory systems of the brain: A brief history and current perspective. Neurobiology of
Learning and Memory, 82: 171–7.
Squires, R. (1969). Memory unchained. The Philosophical Review, 78: 178–96.
Sullivan, J. (2010). Reconsidering spatial memory and the Morris Water Maze. Synthese, 177: 261–83.
Sutton, J. (1998). Philosophy and Memory Traces: Descartes to Connectionism. Cambridge: Cambridge
University Press.
Sutton, J. and Windhorst, C. (2009). Extended and constructive remembering: Two notes on Martin and
Deutscher. Crossroads: An Interdisciplinary Journal for the Study of History, Philosophy, Religion, and Classics,
4: 79–91.
Thompson, R. F. (2005). In search of memory traces. Annual Review of Psychology, 56: 1–23.
Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53: 1–25.
Tulving, E. (2007). Coding and representation: Searching for a home in the brain. In H. L. Roediger,
Y. Dudai, and S. M. Fitzpatrick (eds.), Science of Memory: Concepts (pp. 65–8). Oxford: Oxford
University Press.
Wertsch, J. (2002). Voices of Collective Remembering. Cambridge: Cambridge University Press.
Wilson, M. A. and McNaughton, B. L. (1993). Dynamics of the hippocampal ensemble code for space.
Science, 261: 1055–8.
Wittgenstein, L. (1967). Zettel. G. E. M. Anscombe and G. H. von Wright (eds.), G. E. M. Anscombe
(trans.). Berkeley, CA: University of California Press.
87
7
THE INTENTIONAL
OBJECTS OF MEMORY
Jordi Fernández
1. Introduction
Memories are mental states with a number of interesting features. One of those features seems
to be their having an intentional object. After all, we commonly say that memories are about
things, and that a subject represents the world in a certain way by virtue of remembering
something. It is unclear, however, what sorts of entities constitute the intentional objects of
memory. In particular, it is not clear whether those are mind-independent entities in the world
or whether they are mental entities of some kind. The purpose of this chapter is to map the
different positions on this issue, and to highlight the virtues and difficulties for each of the
options. In Section 2, I will specify the question of what the intentional objects of memory
are by clarifying the relevant notions of memory and intentional object. In Section 3, I will
motivate the significance of identifying the intentional objects of memory by exploring the
relations between, on the one hand, the intentionality of memory and, on the other hand, the
phenomenology and the epistemology of memory. In Section 4, I will consider two natural
candidates for being the intentional objects of memory, namely, worldly entities and mental
entities, and I will raise some concerns for each of the two candidates. A promising alternative
will emerge, in Section 5, as preserving the virtues of the two original candidates while avoid-
ing their difficulties. The alternative will concern a certain combination of worldly and mental
entities; a combination that involves both causal and truth-making relations. I will conclude
by sketching how the alternative candidate can shed some light on the phenomenological and
epistemological issues raised in the third section.
2. Memory and intentional object

It is hard to decide which approach to take towards the question of what the intentional objects
of memory are without knowing, first, what type of memory we are enquiring about and, sec-
ond, what counts as an intentional object for the purposes of our inquiry. It is worth specifying
the two issues since, for all we know, it may turn out that different types of memory happen to
have different types of entities as their intentional objects. Let us start, therefore, by distinguish-
ing some varieties of memory.
A useful distinction in the psychological literature is that between ‘episodic’ and ‘semantic’
memory.1 Remembering something semantically is a matter of having a belief with a certain
88
The intentional objects of memory
causal history. If a subject semantically remembers something, then they believe it, and they
believe it because they formed their belief at some point in the past, and their belief has been
preserved until now. With the term ‘semantic memory’, we refer to the faculty that allows
a subject to preserve their beliefs over time. (Derivatively, we may also refer to those beliefs
which have been preserved by the faculty of semantic memory as ‘semantic memories’.) If I
remember that Columbus arrived to America in 1492, for example, then I remember it seman-
tically: I believe that he did, and I believe that he did because I formed that belief during some
history lesson in the past, and because my belief has been preserved until now. Remembering
something episodically, on the other hand, consists in having an experience which involves
some kind of imagery and originates in a past perceptual, introspective, or agentive experience
of the remembering subject. If I episodically remember that I left my house keys in the door,
for example, then I undergo an experience wherein my keys are presented to me as having
been in the door, and I have that experience because, at some point in the past, I seemed
to perceive that the keys were in the door. With the term ‘episodic memory’, we can refer
both to the faculty that allows a subject to generate such experiences, and to the experiences
delivered by the faculty.
The key characteristic of episodic and semantic memory, as far as their intentionality is
concerned, is that the two faculties deliver states that have the capacity to be true, accurate, or
correct. We can think of the intentional object of a memory, then, as what it takes for the mem-
ory to be correct. It will be that object, event, action, or state of affairs whose presence makes
the memory correct and whose absence makes it incorrect. We specify the intentional object
of a memory, then, by specifying its truth conditions. With this notion of intentional object in
mind, the question regarding the intentional objects of semantic memory is relatively easy to
answer: If you remember something semantically, then the object of your memory is the object
of your belief, that is, whatever you believe. Since beliefs are propositional attitudes, it seems
that those things which you remember semantically will always be propositions, that is, states
of affairs of some kind.2 The precise kind will simply depend on the subject matter of your past
belief in each case. Suppose, for example, that on Monday you believe that the painful sensation
in your stomach is intense and, on Thursday, you remember that you felt an intense painful
sensation in your stomach because, on Monday, you formed that belief and your belief has been
preserved until Thursday. Then, the remembered state of affairs is not mind-independent, since
it involves a sensation: It is the fact that you felt an intense painful sensation in your stomach.
By contrast, in the Columbus case, the remembered state of affairs is clearly mind-independent.3
The question about the intentional objects of episodic memory is the hard and interesting
question to answer. We speak of episodic memories as if they were directed at either states
of affairs, actions, events, or objects. Thus, we express episodic memories with expressions
of the form ‘I remember that p’ (where ‘p’ stands for a state of affairs, as in ‘I remember that
the house keys were in the door’), ‘I remember -ing’ (where ‘ ’ stands for an action, as in
‘I remember inserting the keys in the door’), and ‘I remember x’ (where ‘x’ stands for an
object or an event, as in ‘I remember the house keys’ and ‘I remember the sale of my house’
respectively). I take episodically remembering to be a propositional attitude. Thus, I take the
case where episodically remembering is an attitude towards a state of affairs to be the basic
case. In what follows, I will assume, accordingly, that episodically remembering objects, epi-
sodically remembering events and episodically remembering actions should be construed as
remembering certain states of affairs that involve those objects, events and actions to have
obtained. Thus, I will take expressions of the form ‘S remembers -ing’ and ‘S remembers x’
to abbreviate, respectively, that S remembers that she -ed, and that S remembers that S was
in some (contextually salient) relation to x.
89
Jordi Fernández
There are alternative approaches, to be sure. One might take episodically remembering to
be, in the first instance, an attitude towards an event and, derivatively, an attitude towards a
proposition; the proposition that the event at issue had this or that property. The advantage
of the propositional approach over the events-based approach is that, on the propositional
approach, it is easy to accommodate the intuition that episodic memories have truth conditions;
conditions under which they may be right or wrong. For those conditions can be captured
by a proposition straightforwardly. On the events-based approach, by contrast, it is not easy
to accommodate that intuition, since events are not bearers of either truth or falsity. On the
events-based approach, it may make more sense to claim that episodic memories do not have
truth conditions and, strictly speaking, there is no such thing as a false, or incorrect, episodic
memory. I, however, will be assuming that misremembering episodically is still remembering;
hence my preference for the propositional approach.4
If the propositional approach is correct, then it seems that, similarly to the case of semantic
memory, the objects of our episodic memories are states of affairs of some kind. One might
wonder, though, what kind that is exactly. Are episodically remembered states of affairs always
worldly states of affairs? Are they always mental states of affairs? We have seen that, in the case
of semantic memory, remembered states of affairs do not need to belong to a single kind. Each
time we remember something semantically, one of our past beliefs plays the role of determining
which state of affairs we are remembering, and that belief may sometimes be about a mental
state of affairs, and sometimes be about a worldly state of affairs. Analogously, each time we
remember something episodically, our past (typically perceptual) experiences seem to play the
parallel role of determining which state of affairs we are remembering. But, as we are about
to see, the role that past experiences play in episodic memory is harder to specify than that of
beliefs in semantic memory. And, for that reason, it is not straightforward whether episodically
remembered states of affairs need to belong to a single kind or not.
There is a certain dis-analogy between the role of past belief in determining the intentional
object of a corresponding semantic memory and the role of past perceptual experience in deter-
mining the intentional object of a corresponding episodic memory. In the case of episodic
memory, there is room for drawing a distinction between a direct and an indirect contribution
that the past perceptual experience is making. One might argue that, directly, the past percep-
tual experience determines the intentional object of the episodic memory in which it originates
in that the fact that the subject had that perceptual experience in the past is the immediate inten-
tional object of the memory. But, one might add, the past perceptual experience also determines
the intentional object of the episodic memory indirectly: in virtue of the fact that the subject’s
past perceptual experience was about some state of affairs, the seemingly perceived past state of
affairs is also the intentional object of the memory, although in a mediated way. The distinction
between the two types of contribution that past perceptual experience makes in determining
the intentional object of episodic memories opens up a certain position about the question of
whether episodically remembered states of affairs are worldly states of affairs or mental states of
affairs. It allows us to take the position that the question at issue is a false dichotomy: The option
that both types of states of affairs are the intentional objects of episodic memories at the same
time, one might claim, has been neglected.
In the case of semantic memory, by contrast, there seems to be no room for a distinction
between two types of contribution that past belief makes in determining the intentional object
of a semantic memory. It does not seem that, when we learn some fact in the past and we now
remember it semantically, there is some sense in which our semantic memory is about the belief
that we formed in the past during the learning episode. If, for example, at some point in the past
I learnt that Columbus arrived in America in 1492, and I now remember that fact, it does not
90
seem that my current belief is about the belief that I formed in the past, when I learnt that fact.
What I believe, when I semantically remember, seems to be something about Columbus. There
seems to be only one type of contribution that the belief I formed in the past, when I learnt
about Columbus, is making in determining the intentional object of my semantic memory:
whatever I learnt about Columbus constitutes the intentional object of my semantic memory.
Now, if things are not as straightforward in the episodic memory case as they are in the
semantic memory case, then should we not take the possibility that both worldly states of affairs
and mental states of affairs may be the intentional objects of episodic memories at the same time
seriously? In what follows, I will remain neutral on whether episodic memories can indeed have
intentional objects in the mediated way sketched above. For the purposes of our discussion here,
the intentional objects of episodic memory will be conceived as their immediate intentional
objects. (Those may be their only intentional objects or not, depending on whether an account
of what it is for a state of affairs to constitute the intentional object of an episodic memory in a
mediated way can be worked out in detail.) Let us consider, next, the significance of specifying
this type of intentional objects for episodic memory.
3. The phenomenology and epistemology of episodic memory

In addition to being intentional states, our episodic memories are mental states that enjoy dis-
tinctive features of two kinds. Episodic memories have, first, a characteristic phenomenology.
There are, in other words, some phenomenal features that episodic memories typically enjoy
or, equivalently, some feelings that they normally elicit in the remembering subject. Second,
episodic memories afford a special knowledge of some domains. That is, we can acquire, in
virtue of having episodic memories, knowledge of certain areas of reality that is unlike the
knowledge of those areas which we can acquire through other means. Let us consider these two
sets of features of episodic memories in order.
In order to highlight two characteristic phenomenal features of episodic memory, it may
be useful to contrast episodic memory with some of our other faculties. One way in which
episodic memory differs, for example, from perception and introspection is that remembered
states of affairs appear to us to be in the past. By contrast, perceived states of affairs do not
appear to us to be in the past, and introspected states of affairs do not appear to us to be in the
past either. When we have a perceptual experience, it feels to us as if the perceived state of
affairs obtains in the present.5 Likewise, when we undergo an episode of introspection, it feels
to us as if the introspected state of affairs is obtaining in the present. However, in virtue of
having an episodic memory, the state of affairs that we are remembering appears to us to have
been the case in the past. Let us call this feature of episodic memory a ‘feeling of pastness’.
One way in which episodic memory is, however, similar to introspection concerns our own
past experience. Suppose that we introspect a mental state with some phenomenal features,
such as a sensation, an emotion, or a perceptual experience. Then, in virtue of undergoing
that episode of introspection, we are aware of what it is like to occupy the introspected mental
state. Likewise, when we have an episodic memory, we become aware of what it was like to
experience a certain state of affairs; a state of affairs in the past. We can call this feature of epi-
sodic memory, a ‘feeling of past experience’.
A comparison with perception and introspection is also useful to highlight two of the char-
acteristic epistemic features of memory. One way in which episodic memory is epistemically
special is that it provides us with evidence, or grounds, for forming beliefs about the past, in an
analogous way to that in which perception provides us with evidence for forming beliefs about
the present. Episodic memories seem to put us in a position to form beliefs about the past that
91
Jordi Fernández
is, in a certain sense, immediate. It seems that, in virtue of having an episodic memory, we
have evidence for believing that some state of affairs happened in the past without the need to
perform any inference. Just like perception allows us to form perceptual beliefs by taking our
perceptual experiences at face value, it seems that episodic memory allows us to form beliefs
about the past by taking our episodic memories at face value. To that extent, it seems that epi-
sodic memory puts us in direct cognitive contact with the past. Let us call this epistemic feature
of episodic memory the ‘immediacy of the past’. A different way in which episodic memory is
epistemically special is that it provides us with evidence, or grounds, for forming beliefs about
our own past experience, in an analogous way to that in which introspection provides us with
evidence for forming beliefs about our present experience. Episodic memories seem to put
us in a position to form beliefs about our past experience non-inferentially. By simply taking
our episodic memories at face value, we can form beliefs about what our own experience was
like at certain moments in the past. Let us call this epistemic feature of episodic memory the
‘immediacy of past experience’.
The significance of an investigation of the intentional objects of episodic memory lies in its
potential to illuminate the phenomenology and epistemology of episodic memory. Suppose that
the intentional objects of our episodic memories involved the past as well as our own experi-
ence. Suppose, that is, that our episodic memories were, ultimately, about those two domains.
Then, it would make sense that our episodic memories enjoyed a feeling of pastness and a feel-
ing of past experience. The thought would be that the feeling of pastness and the feeling of past
experience is the way in which the intentional objects of episodic memory are presented to us,
i.e. the way in which we experience what our memories are about.6 Similarly, it would make
sense that our episodic memories enjoyed the features of immediacy of the past and immediacy
of past experience. The thought would be that, when we take our episodic memories at face
value, we are simply trusting what they convey to us. If it turned out that the intentional objects
of our episodic memories involved the past state of the world, and they also involved our own
past experience, then this would mean that our memories convey information about those two
domains. It would not be surprising, then, that we are in a position to form beliefs about those
two domains non-inferentially in virtue of having episodic memories.
The devil, however, is in the details. For different ways of specifying the precise involve-
ment of the past and that of our own experience in the intentional objects of our episodic
memories will illuminate the epistemic and phenomenal features of those memories with vary-
ing degrees of success. Let us turn, therefore, to some of the candidates for the intentional
objects of episodic memory.
4. Mind and world in episodic memory

Consider the following possible situation. I am looking at the door of my house, and my per-
ceptual experience presents my house keys to me as being in the door. Minutes later, I have an
episodic memory that originates in that perceptual experience; an episodic memory that I would
express by saying that I remember my house keys being in the door. Let us call this possible
situation ‘W0’, and let us call my past perceptual experience and my episodic memory in W0, ‘P’
and ‘M’ respectively. How should we conceive the intentional object of M? Let us assume, for
the sake of convenience, that propositions are identical with sets of possible worlds. Then, the
following two propositions spring to mind as candidates for being the intentional object of M:
OBJ: {W: In W, my house keys were in the door}

SUBJ: {W: In W, I had P}
92
If the intentional object of M is OBJ, then I remember something that is mind-independent;

something that was the case independently of my mental states at the time. Accordingly, let us
call the proposal according to which OBJ is the intentional object of M, the ‘objective proposal’.
By contrast, if the intentional object of M is SUBJ, then I remember something that is not
mind-independent, something that would not have been the case had some of my mental states
been different.7 Let us call, then, the proposal according to which SUBJ is the intentional object
of M, the ‘subjective proposal’. What are the virtues and shortcomings of these two proposals,
and what do those shortcomings teach us about the intentional objects of episodic memory?
The objective proposal performs quite well with regards to the feeling of pastness and the
immediacy of the past. If the objective proposal is correct, then M is about a past objective
state of affairs, namely, my house keys having been in the door. It is not surprising, then, that
I have a sense that, in the past, my house keys were in the door when I have M. After all, the
fact that my house keys were there in the past is simply what I am remembering if the proposal
is correct. For the same reason, it is also no wonder that M allows me to form the belief that a
certain state of affairs was the case in the past (namely, my house keys having been in the door)
without performing any inference. For if the objective proposal is correct, then what I am doing
by forming such a belief is simply accepting what I remember. That is, I am just assenting to the
information conveyed to me by my memory. The trouble for the objective proposal concerns
the feeling of past experience and the immediacy of past experience. The fact that, in virtue of
having M, I become aware of what it was like for me to perceive my house keys at some point
in the past seems odd if the objective proposal is correct. After all, what I remember, if the pro-
posal is correct, is my house keys, not my past perceptual experiences. For the same reason, it
seems odd that, in virtue of having M, I seem to be in a position to form a belief about what my
perceptual experiences were like when I looked at the door of my house without performing
any inference. If what I remember are my house keys, and not my past perceptual experiences,
then it is hard to see what the source of information that I am utilizing to form such a belief is.
Thus, the objective proposal seems to be capable of illuminating the feeling of pastness and the
immediacy of the past, but the feeling of past experience and the immediacy of past experience
seem to raise some difficulties for the proposal.
Perhaps unsurprisingly, the converse virtues and shortcomings apply to the subjective pro-
posal. If the subjective proposal is correct, then M is about a perceptual experience of mine. The
experience in question is P, a past perceptual experience wherein my house keys appeared to
me to be in the door. It is not surprising, then, that having M makes me aware of what it was
like to perceptually experience my house keys in the door. After all, if the subjective proposal
is correct, then having P is, strictly speaking, what I remember by having M. And how could
I episodically remember having P without remembering what it was like for me to have that
experience? For the same reason, it is also no wonder that M allows me to form a belief about
what my perceptual experience was like in the past without performing any inference. After
all, if what I remember is my past perceptual experience P, then one would expect me to be
in a position to form beliefs about that perceptual experience and, in particular, beliefs about
what it was like for me to have it. The trouble for the subjective proposal involves the feeling of
pastness and the immediacy of the past. The fact that, in virtue of having M, I become aware of
my house keys having been in the door seems odd if the subjective proposal is correct. Notice
that what I remember, if the proposal is correct, is my past perceptual experience of the keys
P; not the keys themselves. Why would my memory, then, convey the feeling that, in the past,
the state of the world involved my house keys being in the door? If the subjective proposal is
correct, then such a phenomenology goes beyond the experience of that which I remember.
For the same reason, it is hard to see how M could put me in a position to form the belief that
93
Jordi Fernández
the house keys were in the door without performing any inference. If what I remember, strictly
speaking, is my past perceptual experience of the keys P, then it seems that, in order to arrive
at the belief that the house keys were in the door, I do need to perform an inference; an infer-
ence from the belief that, by having P, I seemed to perceive the house keys in the door, and
the belief that my past perceptual experience P was trustworthy. It seems, therefore, that the
subjective proposal can shed some light on the feeling of past experience and the immediacy of
past experience, but it has trouble with the feeling of pastness and the immediacy of the past.
The outcome of our discussion of the subjective and objective proposals seems to be the
following. On the one hand, the past state of the world must have some kind of involvement in
the intentional objects of memory. Otherwise, it is hard to explain why our memories enjoy a
feeling of pastness and why they allow us to establish an immediate cognitive contact with the
past. On the other hand, our past experience must also be involved in the intentional objects of
memory in some way. Otherwise, it is hard to explain why our memories enjoy a feeling of past
experience and why they afford an immediate cognitive contact with our past experience. The
upshot, then, seems to be that somehow our episodic memories are simultaneously about two
domains: the past state of the world and our past experience. The question that naturally arises at
this point is how those two domains are involved in the intentional objects of memory precisely.
A straightforward way of construing the intentional object of an episodic memory as a
combination of the past state of the world and our past experience is by conjoining the two
elements. Consider again the case in which, in W0, I seem to perceive my house keys being in
the door by having P and, minutes later, I have episodic memory M, which I would express
by saying that I remember my house keys being in the door. The thought is that perhaps the
following candidate for the intentional object of M could overcome the challenges of OBJ as
well as those of SUBJ:
O&S: {W: In W, I correctly perceived that my house keys were in the door by
having P.}
Admittedly, if the intentional object of M is O&S, then it does make sense for M to enjoy
both a feeling of pastness and a feeling of past experience. For what I am remembering, by
having my memory, is both the world and my past experience of it. Similarly, one can see why
M would put me in a position to form, without the need to perform any inference, beliefs
about the past location of the keys and beliefs about my past perceptual experience of them
if the intentional object of M is O&S. After all, M is providing me with information about
both of those subject matters at the same time. Why should we not embrace, then, O&S as
the intentional object of M?
What this new proposal seems to miss is a further phenomenological feature of our episodic
memories. Part of the phenomenology of remembering episodically seems to be the feeling that
the mental state that we are undergoing originates in our own past experiences, as opposed to
testimony or reasoning.8 If we remember some state of affairs episodically, then the question
of whether we remember it because we have experienced that state of affairs in the past or we
remember it for some other reason is no longer open for us. In virtue of the fact that we are hav-
ing an episodic memory, it will seem to us that the relevant state of affairs has been experienced
by us in the past, and it will seem to us that this is the reason why we are now remembering it.
When I episodically remember my house keys in the door by having M, for example, it thereby
seems to me that the location of the keys has been experienced by me in the past, and that this is
why I am now remembering it. The trouble for the proposal according to which the intentional
object of M is O&S is that the proposal is silent on the causal origin of M. If the proposal is right,
94
then my memory M is about a correct episode of perception in the past: my past perception of
my house keys being in the door. But if this is what M is about, then M is neutral on its own
causal origin, which makes it difficult to explain why, in virtue of having M, it seems to me that
my memory comes from experiencing the house keys in the past. It seems, then, that what we
need as the intentional object of our episodic memories is not only a combination of the past
state of the world and our past experience, but a combination that links both elements to our
episodic memories in the right way.
5. Self-reference in episodic memory

In light of the fact that memories convey the feeling that we remember some state of affairs
because we have experienced that state of affairs in the past, it seems that the simplest way of
combining the past state of the world with our own past experience in the intentional objects
of our episodic memories is by identifying the intentional object of an episodic memory with its
own causal history. On this approach, memories are self-referential. They are about themselves.
In particular, they are about their own origin, which involves both a past perceptual experience
and an objective state of affairs.9 What does this mean for the case in which I have episodic
memory M minutes after seeming to perceive my house keys in the door by having P? The
suggestion is that the intentional object of M is the following proposition (where ‘SR’ stands
for ‘self-reference’):
SR: {W: In W, I have M because I correctly perceived that my house keys were in
the door by having P.}
If the intentional object of M is SR, then one can see why, in virtue of having M, I have the
feeling that I am remembering the keys in the door because I have perceived them to be there,
and not for some other reason. One would expect M to carry such a phenomenal feature, since
part of the information that it conveys to me when I have it is precisely where M is coming
from. Furthermore, it is also not surprising that M enjoys a feeling of pastness and affords an
immediate cognitive contact with the past. If a correct past perception of my house keys being
in the door is part of the intentional object of M, then it seems natural for M to produce a feel-
ing that, in the past, my house keys were in the door, since the past perception being correct
requires the keys to have been there. Likewise, it seems reasonable that I can form, on the basis
of M, the belief that my house keys were in the door without performing any inference if the
intentional object of M is SR. For the fact that my house keys were in the door is, once again,
part of what M is informing me of when it informs me that I correctly perceived the keys to be
there. Finally, if the intentional object of M is SR, then it makes sense that M elicits an aware-
ness of what it was like to have P in the past. After all, P is part of what M represents on the
self-referential approach. And, for the same reason, it is no wonder that M puts me in a position
to form, on its basis, beliefs about what it was like for me to have P in the past.
Perhaps a challenge for the self-referential view could arise from the fact that, according to
the view, memories are ‘token reflexive’, that is, they are a constitutive part of their own inten-
tional objects. The self-referential view predicts, in virtue of this aspect of it, that an episodic
memory cannot accurately represent a possible situation in which that very memory is not tak-
ing place. If part of what an episodic memory represents is that the memory itself has a certain
causal history, then the memory cannot accurately represent a possible situation in which the
memory never takes place. For, in that situation, the memory does not have any causal history
whatsoever. Consider, for example, a possible situation in which my house keys are in the door
95
Jordi Fernández
and, by having P (that is, the very same perceptual experience that I have in W0), I perceive
the keys to be there. However, in this possible situation, I quickly forget about my house keys.
Is my memory M, in W0, accurately representing this possible situation?10 If our pre-theoretic
intuitions are that M is accurately representing it, then this type of scenario presents a challenge
for the self-referential view, since the view commits us to the claim that M, in W0, is not cor-
rectly presenting such a possible situation.
A different challenge for the self-referential view might arise from the fact that, according
to the view, the causal link between past experience and memory is also a constitutive part of
the intentional object of a memory. The self-referential view predicts, in virtue of this aspect of
it, that an episodic memory cannot accurately represent a possible situation in which that very
memory takes place, and a correct past perception of the relevant objective state of affairs takes
place in the past, but the two are not causally linked. If part of what an episodic memory rep-
resents is that the memory itself has a certain causal history, then the memory cannot accurately
represent a possible situation in which the memory did not have that causal history.
Consider, for example, a possible situation in which my house keys are in the door and, by
having P (that is, the very same perceptual experience that I have in W0), I perceive the keys
to be there. Furthermore, in this possible situation, I have M (that is, the very same episodic
memory that I have in W0). However, I do not have M because I had P. Instead, in this pos-
sible situation, it turns out that, shortly after having P, I forgot about my seeing the keys in
the door, and M is now the product of my imagination. Is M, in W0, accurately representing
this possible situation?11 If our pre-theoretic intuitions are that M is accurately representing it,
then this type of scenario presents a challenge for the self-referential view too, since the view
commits us to the claim that M, in W0, is not correctly presenting such a situation. And yet, it
is hard to see how we could take care of the intuition that memories wear, so to speak, their
origin on their sleeves if we did not build those memories, with their causal histories, into their
own intentional objects.
An option might be to locate the reflexivity of episodic memories at the level of their phe-
nomenology, rather than at the level of their content. There are various ways of doing this.
According to some of these strategies, an episodic memory is reflexive in that a subject who has
it has the feeling that they themselves experienced whatever is being remembered. According
to other phenomenological strategies, an episodic memory is reflexive in that a subject who has
it has the feeling that the memory itself originates in some experience in the past. And, accord-
ing to yet another kind of phenomenological strategy, an episodic memory is reflexive in that
a subject who has it has the feeling that the memory itself originates in an experience that they
themselves had in the past.12 The three strategies differ, then, in the degree of reflexivity that
they build into the phenomenology of episodic memory. What is common to all three strate-
gies is the contention that the phenomenology of an episodic memory is not determined by,
and does not determine, its content. If a relation of dependence between the phenomenology
of a memory and its content did hold in either direction, then the reflexivity involved in the
phenomenology of that memory (whether this concerns a reference to the memory itself, a ref-
erence to its subject, or a reference to both) would be reflected in the content of that memory.
If the relation of dependence grounds phenomenology on content, then it would be reflected
because something in the content of the memory would need to be responsible for the relevant
self-reflexive aspect of the phenomenology of that memory. And if the relation of dependence
grounds content on phenomenology, then it would be reflected because the memory would
then represent either itself, or its subject, or perhaps both, in virtue of the relevant self-reflexive
aspect of its phenomenology. Proponents of the view that the self-reflexive character of epi-
sodic memory must be located at the level of phenomenology, but not at the level of content,
96
are therefore committed to a form of separatism (the view that the phenomenal and intentional
features of mental states are independent from each other). Whether this is too high a price to
pay for an account of the reflexivity of episodic memory or not, however, is an issue that would
take us far beyond the scope of this aspect to pursue.13
The conclusion to draw from our discussion in this chapter might be that our folk notion of
memory encompasses different ideas about what memories do, and those ideas motivate differ-
ent views about the intentional objects of episodic memory. On the one hand, we think that
memory is, in a certain sense, analogous to perception. Just like perception gives us knowledge
of the present, it seems that memory should give us knowledge of the past. And this intuitive
idea motivates the objective proposal about the intentional objects of memory. On the other
hand, we think that memory is, in a certain sense, analogous to introspection. Just like introspec-
tion gives us access to our present experiences, it seems that memory should give us access to our
past experiences. And this intuitive idea motivates the subjective proposal about the intentional
objects of memory. One is then naturally pushed towards some version of the self-referential
view to relieve the tension between these conceptions of the intentional objects of memory.
Adopting this approach seems reasonable, but the benefits of the self-referential view come at a
cost, a cost that should be disclosed. Admittedly, the more we build into the intentional objects
of episodic memory, the more phenomenal and epistemic features of episodic memories we
will be able to illuminate. But it also seems that the more we build into the intentional objects
of episodic memory, the more demanding it will be for a possible situation to qualify as being
accurately represented by our memories. As a result, the self-referential view raises the bar for
accuracy considerably high. Whether this aspect of the self-referential view is ultimately a prob-
lem will probably depend on our commitments elsewhere in the philosophy of memory.
Notes
1 The distinction is introduced in Endel Tulving (1972).The distinction has undergone a number of revi-
sions since then, but nothing in the discussion that follows hinges, as far as I can see, on the exact details
of how we distinguish episodic memory from semantic memory.
2 Here I am assuming that propositions are identical with states of affairs. In the fourth section, I will
also introduce the assumption that propositions are identical with sets of possible worlds. Despite the
fact that the two assumptions are convenient for expositional purposes, the present discussion of the
intentional objects of memory does not hang on the nature of propositions. If propositions are not
identical with states of affairs, then the question of whether the intentional objects of memory are
mind-independent or not can be read as the question of whether those propositions which constitute
the intentional objects of our memories concern, or are about, the objective world or not. Likewise, if
propositions are not sets of possible worlds and instead they are, let us say, ordered pairs of properties and
objects, then the question of whether the intentional objects of memory are mind-independent or not
can be read as the question of whether the constituents of those propositions which are the intentional
objects of our memories belong to the objective world or not.
3 Another way in which semantically remembered states of affairs may vary is temporally. Suppose that,
on Monday, you make an appointment to see your doctor; the appointment is for Thursday. Suppose
that your belief that you will see your doctor on Thursday is preserved until, let us say, Wednesday.
Then, on Wednesday, you are semantically remembering a future state of affairs (namely, that you will
see your doctor on Thursday). By contrast, in the Columbus case, the remembered state of affairs is
clearly in the past.
4 Sen Cheng and Markus Werning (2016) have proposed an account of the nature of episodic memory
which makes episodes their intentional objects. Episodes are, as Cheng and Werning construe them,
series of events in temporal succession and, for that reason, Chen and Werning’s view constitutes an
example of an events-based approach to episodic memory. In this particular version of the events-based
approach, however, the intentional objects of episodic memories are temporally extended. This intro-
duces the complication of determining how to individuate episodic memories. (Is the memory that
97
Jordi Fernández
the remembering subject is having when they remember the first event in the episode the same as the
memory that the subject is having when they remember the last event in the episode?) Fortunately, this
is not an issue that we will need to address here. Intentional objects of memory, for the purposes of this
discussion, will not be construed as temporally extended entities such as episodes.
5 It is sometimes argued that, since light takes some time to travel from the objects that we perceive to
our sensory organs, perception is always perception of something in the past. Nonetheless, it remains
true that perceived states of affairs appear to us, rightly or wrongly, to be in the present.
6 There is a strong representationalist view on consciousness according to which the phenomenal fea-
tures of our mental states are intentional features of those states. This is the view put forward, for
example, by Fred Dretske (1995). The point that the phenomenology of our episodic memories is the
way in which we experience their intentional objects is congenial to this view, but it does not entail it.
It is consistent with the possibility that the phenomenology of our episodic memories supervenes on
their intentionality, even though the phenomenal features of our episodic memories are distinct from
their intentional features.
7 Some expressions of this view can be found, for instance, in Alexius Meinong (1973: 256) and Wolfgang
Von Leyden (1961: 61).
8 This feature is highlighted, for example, in Aristotle (1972: 69), John Locke (1975: 83), and William
James (1890: 648–52). If we accept that mental states which originate in episodes of imagination,
testimony, or reasoning can qualify as memories, then we should grant that it is possible for this phe-
nomenology to be misleading. Nonetheless, it remains true that episodic memories appear to us, rightly
or wrongly, to originate in our own past experiences.
9 The earliest version of this view is, to my knowledge, that in John Searle (1983: 85). However, the
version of the view presented here differs from Searle’s in some details. For discussion of the relevant
differences, see Fernández (2006).
10 To be precise, the question is not whether M is true in that possible situation. (Since M never takes place
in that situation, the answer to that question is obviously in the negative.) The question is whether M is
true of that possible situation.
11 The question, once again, is not whether M is true when it takes place in this possible situation. (Since
M, in this possible situation, is the product of my imagination, the question to that answer is likely to be
in the negative.) The question is rather whether M, in W0, is true of that possible situation.
12 Examples of, at least, the first kind of strategy can be found in Klein and Nichols (2012) and Michaelian
(2016).
13 For a relevant discussion, see Horgan and Tienson (2002).
Related topics
• Phenomenology of memory
• Foundationalism
Further reading
Bernecker, S. (2010) Memory: A Philosophical Study. Oxford: Oxford University Press.
Lawlor, K. (2009) ‘Memory’, in B. McLaughlin, A. Beckermann and S. Walter (eds), The Oxford Handbook
of Philosophy of Mind. Oxford: Oxford University Press, 663–78.
References
Aristotle (1972) ‘De memoria’, in R. Sorabji (ed.), Aristotle on Memory. London: Duckworth.
Cheng, S. and Werning, M. (2016) ‘What is episodic memory if it is a natural kind?’, Synthese 193:
1345–85.
Dretske, F. (1995) Naturalizing the Mind. Cambridge, MA: MIT Press.
Fernández, J. (2006) ‘The intentionality of memory’, Australasian Journal of Philosophy 84: 39–57.
Horgan, T. and Tienson, J. (2002) ‘The intentionality of phenomenology and the phenomenology of
intentionality’, in D. J. Chalmers (ed.), Philosophy of Mind: Classical and Contemporary Readings. New
York: Oxford University Press, 520–33.
98
James, W. (1890) The Principles of Psychology. London: Macmillan.

Klein, S. and Nichols, S. (2012) ‘Memory and the sense of personal identity’, Mind 121: 677–702.
Locke, J. (1975) An Essay Concerning Human Understanding, P. Nidditch (ed.). Oxford: Oxford University
Press.
Meinong, A. (1973) ‘Toward an epistemological assessment of memory’, in R. Chisholm and R. Swartz
(eds), Empirical Knowledge: Readings from Contemporary Sources. Englewood Cliffs, NJ: Prentice Hall.
Michaelian, K. (2016) Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past.
Searle, J. (1983) Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press.
Tulving, E. (1972) ‘Episodic and semantic memory’, in W. Donaldson and E. Tulving (eds), Organization
of Memory. New York: Academic Press, 381–403.
Von Leyden, W. (1961) Remembering: A Philosophical Problem. London: Duckworth.
99
PART III
Memory, mind, and meaning

8
MEMORY AND
CONSCIOUSNESS
Paula Droege
1. Introduction
The connection between memory and consciousness is so deep that people often assume that
‘memory’ means conscious memory. This chapter will discuss the way consciousness figures
in relation to one form of memory, episodic memory. To begin, Section 2 proposes a general
definition of memory as the present use of past experience, where ‘experience’ does not imply
consciousness. Section 3 considers the role of consciousness in theories of episodic memory.
Originally defined by Endel Tulving (1972) as the capacity to remember a specific past event
or episode, this form of memory is usually identified with the conscious recollection of past
personal experience. The difficulty with this characterization is to account for the peculiar
‘feeling of pastness’ in episodic memory and how it relates to present experience. New research
on what is lost and what is preserved in the case of episodic amnesia suggests that conscious
re-experience is not necessary for the capacity to remember personal past events. Instead, I pro-
pose that the conscious feeling of pastness results from embedding the representation of a past
experience within a representation of the present. One benefit of this description of episodic
memory is the clarity it brings to the debate about whether animals are capable of episodic
memory. While animals demonstrate remarkable abilities to remember the past and plan for
the future, the role of consciousness in accomplishing these tasks remains unclear. Section 4
will consider why it is difficult to show how consciousness is involved in animal memory and
consider what tests might be persuasive. The lesson to be drawn from these considerations is
that representational function can account for the conscious aspect of episodic memory and
arbitrate between empirical paradigms investigating the role of consciousness in behavior.
2. Defining memory
John Locke’s famous and influential definition of memory firmly established the connection
between memory and consciousness. He described memory as a power “to revive Perceptions,
which it has once had, with this additional Perception annexed to them, that it has had them
before” (Locke 1689/1975: 150) [Book II, Ch. X, §2]. That this description implies conscious
recollection is made clear shortly after this passage, when he explains that to revive perceptions is
for the mind to “paint them anew on it self, though some with more, some with less difficulty;
103
Paula Droege
some more lively, and others more obscurely.” For Locke, memory is a matter of consciously
re-experiencing past ideas (see also Copenhaver, Chapter 38, this volume).
From a contemporary perspective, this definition is seriously limited. For one thing, the capac-
ity to ‘revive perceptions’ out of nothing remains a mystery. There was no role for the brain in
Locke’s dualism, nor any possibility of unconscious sensation or thought in his empiricism. The
statement that memory revives perceptions describes the situation without providing any hint
about what exactly is occurring or how to explain it. This point leads to another problem with
Locke’s definition: psychological research has amply demonstrated the existence of unconscious
forms of memory, such as habits, skills, conditioned learning, and priming. To limit memory to
conscious recall is to ignore a whole range of memory types.
In the twentieth century, Bertrand Russell offered a more inclusive definition of memory
in his description of mnemonic phenomena as a way to explain present behavior in terms of “past
occurrences in the history of the organism” (1921: 55). Russell himself followed tradition and
reserved the term ‘memory’ for conscious memory-images that are copies of past sensations
(108), so I will offer a definition that draws on Russell’s description of mnemonic phenomena
to account for both unconscious and conscious forms of memory.
Broadly speaking, memory is the present use of past experience. To use the more precise terms
of representational theory, when the mental representations that constitute past experience are
used in the present, they are memory representations. Anyone not fond of a representational theory
of mind can use the broader definition for now, so long as ‘experience’ includes both conscious
and unconscious mental states. In the following, I will offer reasons to think of memory and
consciousness in terms of teleofunctional representation, where a representation has the function
of varying in accord with an item, because that variation relation has contributed to survival
(Millikan 2004). On a teleofunctional theory, ‘memory’ is not a success word, because memory
representations may fail to fulfill their function. Memories can be false, as when I remember
locking the door but actually didn’t, or they can be accurate and still fail to be useful, as when I
remember putting the delicious leftovers in the refrigerator but someone else has eaten them in
the meantime. Memories succeed when present uses aid the organism in some way: when past
skill learning yields present performance, for example, or accurate recall of past events informs
decision making. Memory representations fail when a skill is badly executed or an event is con-
fabulated. Nonetheless, all past representations used in the present are memory representations.
3. The question of episodic memory

Episodic memory is the form of memory most closely connected to consciousness and is probably
the kind of memory Locke had in mind. The term ‘episodic memory’ was coined by psychologist
Endel Tulving under the constraints of behaviorism, when talk of the mind was methodologi-
cally problematic, and ‘consciousness’ was simply not a scientific term. Consequently, Tulving
(1972) proposed appropriately behavioristic criteria for episodic memory as the ability to store
and utilize temporal and spatial information about episodes and events. (We’ll come back to this
criterion in the next section when we consider whether animals have episodic memories.) Once
the grip of behaviorism had loosened, Tulving added two important conditions for the cur-
rent definition: (1) episodes are specific events in the subject’s personal past, and (2) the subject
re-experiences the past episode by means of autonoetic consciousness (Tulving 1985).
These additional criteria are meant to capture the fact that episodic memories have a specific
sort of phenomenology, in contrast to the way memories for facts (semantic memory) or abili-
ties (procedural memory) function without any particular conscious flavor. When I remember
that Berlin is the capital of Germany or remember how to ride a bike, those memories use past
104
Memory and consciousness
experience for successful present action, but those past learning experiences are not consciously
brought into the present. My consciousness simply produces the relevant information or capac-
ity. With episodic memory, images from the past event are consciously experienced in the
present as my past experience.
One reason the term ‘memory’ often refers just to episodic memory is the unique role that
this form of memory plays in our conscious lives.1 Re-experiencing past events can be a source
of knowledge—yes, I know she was at the party, because I remember she wore the red sweater.
Bringing the past into the present also establishes a connection between temporally distant
events in our lives. Locke’s famous ‘memory criterion’ for personal identity hinges on the capac-
ity to identify my past consciousness with my present consciousness to ground sameness of self
over time in sameness of consciousness (Locke 1689/1975: 335) [Book II, Ch. 27, §9] (see also
Nichols, Chapter 13, this volume). But the dual temporality of episodic memory is puzzling:
how could the past be in the present? The past is defined in contrast to the present, so the sug-
gestion that both times exist simultaneously seems contradictory. In his analysis of memory,
Bertrand Russell identified two necessary components for memory: (1) an image that copies past
experience, and (2) a belief that the image is past (1921: 125). A drawback of Russell’s account
is that he does not explain how the belief creates the phenomenological ‘feeling of pastness’ that
accompanies episodic memory, nor does he say what circumstances determine when the belief
is true (Droege 2013).
Theorists since Russell’s time continue to struggle with both the phenomenological and
epistemological features of episodic memory. Jordi Fernández (2007, 2014) argues that epi-
sodic memory must be true. The content of memory is past veridical perceptual experiences.
He claims that only this ‘factive view’ includes the causal relation to the past event as well as
our state of mind when we experienced that event. Because the perceptual experience of the
past event is presented as the cause of the memory, it necessarily occurred before the memory.
Consequently, the ‘feeling of pastness’ arises from the property of being a cause rather than the
property of being temporally past.
There are two problems with this theory. First is the problem of constructed episodic memo-
ries. Current research shows that memory systems reconstruct rather than replay experiences
of the past. Consequently, there is a great deal of room for error in details or sequencing even
while the overall narrative is correct (Michaelian 2016: Chapter 5; Addis et al. 2009, 2012). If
most memories are mistaken in some way or another, then either the theory fails to account for
most memory phenomena, or it must allow for partly true memories. In the first case, the theory
is simply inadequate; in the second, it requires clarification. How much of a memory must be
true for it to be a memory?
The second problem is the explanation of the ‘feeling of pastness’ in terms of the property
of being a cause. That the perceptual experience caused the memory is taken to be part of the
content of an episodic memory. But this description is both phenomenologically incorrect and
representationally inspecific. When I have an episodic memory, it presents itself to me as a
past experience, not as a perceptual experience that caused a memory experience. The memory
includes the representational content of the past, perhaps even the specific content of a particular
past time such as my fifth birthday. The property of being a cause fails to capture the aspect of
pastness in episodic memory.
Mohen Matthen (2010) solves the first problem of constructed memories by proposing that
experiences leave traces that are later reconstructed into images or beliefs. The reconstructive
process may be more or less accurate in forming memories about past experiences, yet they are
constructed from traces and so are all memories (see also Robins, Chapter 6, this volume). With
respect to the ‘feeling of pastness,’ Matthen argues against an explanation in terms of content.
105
Paula Droege
In his view, episodic memories are images, and images do not depict their causal or temporal origins.
Therefore, the ‘feeling of pastness’ must be an attitude or way of entertaining the image content:
The feeling of pastness is immediately felt; it is phenomenologically apparent . . . I am

suggesting that it is a way of entertaining a memorial image, a tense operator added
to an untensed image. The operator is subpersonally added: it is a signature of the
memory system.
(Matthen 2010: 11)
The idea that temporality is a kind of phenomenological relation to content of the past event is
plausible, but calling it a ‘tense operator’ is a description, not an explanation. As with memory
traces, Matthen suggests the mental work of temporality is done by the physical/causal system.
Of course, the work is done by the memory system, but the philosophical challenge is to explain
how a physical system constitutes mental phenomena such as episodic memory.
To explain the phenomenology of episodic memory requires, first, a theory of conscious-
ness. Without a clear idea about what makes any state conscious, proposals about what makes
episodic memory conscious will be fuzzy as well. I have argued that consciousness represents the
present moment; the contents of consciousness are the best approximation of what is happen-
ing now (Droege 2003, 2009). The indexical now is a function from context to content, where
context is determined by the current environment in relation to current goals. The content of
my consciousness at any given moment is the set of sensory and cognitive states that represents
how things are in the world relative to what I am doing. Since this is a teleofunctional theory, the
contents picked out as now are those that have contributed to survival. So, for example, the sound
of someone talking is synched to the sight of their lips moving, even though the physical signals
travel at different speeds and are processed by the brain at different rates. Our conscious experi-
ence represents these signals as simultaneous, because that temporal relation has been useful.
How might this theory deal with the conscious feeling of pastness, if conscious states repre-
sent the present? One possibility is to embed the past experience within the current conscious
state. When I have an episodic memory of my fifth birthday, the content of my consciousness
would include both current and past contents. As I work on this chapter, I can call up an episodic
memory with the following contents, where < > indicate my current conscious experience, and
[ ] indicate my fifth birthday experience: <9:47am 2016, laptop, desk [1968, green dress, cake,
‘Happy Birthday’ song]>.2 My conscious state represents the past experience as part of the con-
tent of the world as it is now. The higher-order representational structure resolves the apparent
contradiction of representing both present and past simultaneously. The representation of the
past as past is part of the content of the representation of the present. So its contents are no more
contradictory than the sentence ‘I am now representing that I went to the store yesterday.’
Add to this analysis the claim that representations are individuated in terms of function, and we
can see how episodic memories can misrepresent the events they are about. If my episodic mem-
ory system has the function of representing my own past experiences, then it might malfunction
completely and still produce representations as of the past. This is why the phenomenology of
memory alone is insufficient to ensure its accuracy. While episodic memory is basically reliable so
long as it is functioning properly, other forms of evidence are necessary for conclusive verification
of the accuracy of our memory constructions (Michaelian 2016; Droege 2015; Bernecker 2010).
Cases of amnesia offer further insight by showing what functions are and are not lost with a
failure of episodic memory. Severe damage to the hippocampus results in an inability to encode
and recall personal experiences (Dalla Barba and La Corte 2013). Episodic amnesics cannot
‘re-experience’ any events in their past nor imagine particular future events (Kwan et al. 2010).
106
Consequently, it is natural to assume that amnesic patients are ‘stuck in time’ (Roberts 2002),
unable to maintain a sense of themselves over the course of time.
This assumption has been undermined recently by studies done with K.C., who has a com-
plete deficit of episodic memory due to a motorcycle injury. As Carl Craver and colleagues
(2014) point out, preserved semantic memory for autobiographical details allows K.C. to order
past events on a personal timeline and to understand the difference between past and future.
K.C. comprehends regret, which relies on the possible future consequences of present action,
and is able to value greater future rewards over immediate benefits.
Research also shows that when episodic amnesics are cued to imagine a future event in
advance of a choice between a lesser immediate reward and a greater future reward, the ten-
dency to value the greater future reward is even stronger. This result is particularly striking given
that amnesics are impaired in the capacity for episodic projection; they can no more picture
future events than past events. Even to identify relevant future events for the study, amnesic
participants relied on calendars and other external memory devices (Kwan et al. 2015).
The preserved abilities of amnesics to make intertemporal decisions and generally to function
as a self within society demonstrates the adaptive nature of the mind in the face of disability.
There is also a cautionary tale to be told regarding functional theories of mental capacities such
as episodic memory. One such theory, proposed by Pascal Boyer (2008), takes the evolutionary
value of episodic memory to be a means of emotionally weighting the strength of the future
over the present. Boyer notes that involuntary episodic memories are generally biased toward
negative events, in contrast to the bias toward positive events in voluntary memory recall. Given
that most episodic memories are involuntary, the emotional weight of unpleasant events must
be useful. Boyer suggests that re-experiencing the negative emotional consequences of past
action helps a person avoid the temptation to repeat that action.
The value of avoiding bad consequences is obvious in the simple case of an associative connec-
tion such as between red berries and poison. According to Boyer, a more complicated episodic
memory is needed to navigate the nuances of negative social consequences. Social cooperation
demands an extra ‘oompf’ in order to get people to favor the unreliable future benefits of group
action over the immediate benefits of self-interested action. Unpleasant episodic memories are a
way to jigger the decision-making process in favor of future social good, especially in cooperative
action that offers few or no benefits at all to the actor. Consider two scenarios: I am tempted to
stay in bed, but instead I go to church; I am tempted to cheat on a test, but instead I accept a bad
grade. Social cohesion depends on people acting against their present interests in the interest of
structures that are in the interest of the group. The capacity to imagine future consequences—
guilt, penalty—is one means by which people are motivated to act toward social cooperation.
Boyer’s theory predicts that people with amnesia will be poor at valuing the future over the
present due to their inability to utilize the emotional power of episodic memory. He points to
evidence that amnesics fail to recognize that in the Iowa Gambling Task, one deck yields large
short-term gains and losses, but another has a better long-term yield (Gutbrod et al. 2006). Indeed,
K.C. also does badly on this task, yet shows no evidence of greater risk taking or impulsivity than
control subjects (Rosenbaum et al. 2016). As already noted, K.C. can prioritize the future over
the present, is calm and quiet, and is neither hedonistic nor fatalist (Craver et al. 2014).
So what should we conclude about the function of episodic memory given these inconsistent
results? Does K.C.’s ability to value the future over the present show that episodic memories do
not have this adaptive value? I think the lesson here is that function cannot be pulled apart from
representational content. On a teleofunctional theory, episodic memory represents past experi-
ence by embedding that past representational state within a representation of the present. There
may be many different advantages to representations of this kind: to bias present action toward the
107
Paula Droege
future, to help track the self through time, to provide material for imaginative planning, or even to
amuse or torture oneself with past pleasures and pains. Just as a mind might have many uses for a
representation of water, it might have many uses for a representation of a past experience of water.
In the fourth section, I will focus on the role of episodic memory in the development of a
temporally extended self. However, in the absence of episodic memory, other resources can fill
this role. If the mind cannot hold the events of life in view, calendars and journals and other
people can help maintain the structure of a self in time. From a clinical perspective, it is far more
important to focus on the capacities that are preserved and figure out how to supplement them
than to use the deficit profile to theorize about a necessary function for episodic memory.
Even so, these two goals need not be at odds. The research with K.C. is designed to take apart
the concept of ‘temporal consciousness’ (Dalla Barba 2002) in order to distinguish which capaci-
ties are lost and which are preserved in the case of episodic amnesia. This work shows that the
ability to consciously re-experience the past or imagine the future is indeed lost. However, the
ability to consciously think about the past and future remains. An adequate theory of conscious-
ness should be able to distinguish these very different roles for time within conscious experience.
Consistent with my previous arguments, I think that the best way to articulate the difference is in
terms of representation. In episodic memory, a past experience is represented as part of a conscious
representation of the present. Since the past experience is a representation of past events, on my
view, a memory is a conscious representation that contains as part of its content a representation of
the past. Semantic memory, in contrast, represents a past event without embedding the past experi-
ence of that event within the present conscious state. Someone with episodic amnesia represents
the past events that make up her life in the same way she represents other forms of information.
4. On animal memory
One advantage to thinking of episodic memory in terms of its representational function is the
clarity it can bring to empirical debates about the memory capacity of animals. Because animals
cannot report on the nature of their memories, we rely on behavioral evidence to demonstrate
what and how they remember. When it comes to the phenomenon of consciousness, however,
it seems that there could be no objective evidence sufficient to prove or disprove its occurrence
(Nagel 1974). Consequently, there seems no way to show whether animals re-experience the
past in the way necessary to episodic memory.
Out of respect for this epistemic barrier, animal researchers have focused on Tulving’s original
requirement that episodic memory is about a specific event in one’s personal past. The WWW
criterion is the capacity to remember What happened, Where it happened, and When it hap-
pened (Tulving 1972). Nicola Clayton and colleagues (Clayton and Dickinson 1998) dubbed this
WWW capacity episodic-like memory in their work with scrub jays, which cache a wide variety of
foods at varying time periods. These jays need to keep track of What food they hid, Where and
When they hid it in order to get to food before it degrades or, failing that, to avoid the waste of
energy by retrieving bad food. In one study, the jays cached both peanuts and their preferred food,
worms. After 4 hours, the birds went to the worm hiding place, but after 124 hours they went
directly to the peanut hiding place. They were able to remember that they hid the worms too
long ago to be worth retrieving. Since this pioneering study, tests on other animals using similar
paradigms have met with varying success. No other animal group exhibits the acuity of temporal
representation demonstrated by the scrub jay, suggesting that this talent was evolutionarily selected
due to the very specific niche these birds occupy (see Cheke and Clayton (2010) for a review).
In addition to the difficulty in applying the paradigm to other animal groups, the study was
criticized as failing to show the necessity for a representation of the past. William Roberts (2002)
108
argued that the jays could simply be representing the state of the food now as either degraded
or not. In other words, interval timing mechanisms operate like a red flag that signals the divi-
sion between good to eat and not good to eat. At any given point, the birds could utilize these
mechanisms to determine which food is good to eat now. There need be no capacity to remem-
ber a specific past episode such as ‘I cached the worms in the box on the left 124 hours ago’ in
order to know that they are degraded.
To try to address these problems, Thomas Zentall and colleagues (2008) proposed a way to
capture a different aspect of episodic memory: contextual information. Episodic memory dif-
fers from other ways of representing the past in that details of the episode are encoded, even
when those details aren’t relevant to the task. The green dress in my fifth birthday memory was
entirely unimportant to the event, yet that element and other contextual features—such as the
shape of the table (rectangular) and direction I was facing (my back to the window)—are part of
my memory of that event. In contrast, my semantic memory of the day that I was born entirely
lacks context; I know only the date and the place.
Researchers hypothesized that the ability to answer an unexpected question about a past
event shows that irrelevant contextual information had been encoded. Pigeons were trained to
indicate which side of a grid they had just pecked by pecking red for left and green for right.
Then they were trained to peck horizontal lines after a yellow cue and vertical lines after a blue
cue; correct answers were randomly presented either on the right or the left. In the test trial,
after the pigeons had pecked either the horizontal or the vertical lines, an unexpected red and
green stimulus appeared requiring the pigeon to remember which side of the grid it had just
pecked. The pigeons were right over 60 percent of the time, suggesting that they remembered
the task-irrelevant information about which side of the grid they had just pecked.
This paradigm is quite complicated and demonstrates how remarkably clever pigeons are.
It does not, however, demonstrate episodic memory. Since the response follows immediately
after the remembered event, it is likely supported by working memory processes which retain
information for a brief interval after an experience. A closer look at why contextual information
is a feature of episodic memory may help guide the development of better tests. I have suggested
that episodic memory represents past experience by embedding it within a present conscious
state. Contextual information is an aspect of episodic memory, on this account, because it is an
aspect of consciousness. When features from past experience are represented in current experi-
ence, the context is essential to recreating the past experience. Without contextual detail, the
past would be phenomenologically bare.
So, why have a phenomenologically rich memory? One reason is that the capacity to imagi-
natively project yourself into your personal past and future allows you to keep track of yourself
in time. While consciousness keeps track of the world in time, episodic memory keeps track of
the self in time. Going back to the pigeons, then, we can see that their attention to context is an
aspect of their consciousness. They collect information about how the world is now, including
task-irrelevant information, in order to be able to flexibly respond to new opportunities and
dangers in the world, such as the unexpected red and green stimulus. There is simply no need to
reproduce a past experience to accomplish this task. (Notice that there is a need to be conscious,
on my view, the demonstration of which is an accomplishment in itself.)
This idea that episodic memory allows a person to imaginatively project herself backward and
forward in time, called mental time travel, has inspired tests of future planning as a way to demonstrate
episodic memory capacity (see also Perrin and Michaelian, Chapter 18, this volume). Those same
amazing scrub jays demonstrate this sort of future planning when they cache specific sorts of food
in locations where they expect a deficiency in the future. This behavior does not necessarily show
episodic future planning, however, because it could rely on a simple caching heuristic like ‘this place
109
Paula Droege
gets low on food X, so hide more here.’ Another study showed that chimpanzees will choose a straw
over fruit in order to access the preferred reward of fruit juice at a later time in a different location.
Here again, it is possible that the chimpanzees learned to associate the straw with the juice, and that
association was sufficient to motivate the choice without prior episodic future planning (Suddendorf
and Corballis 2010). In a more recent study, apes were able to produce tools in advance of access
to an apparatus where they were needed for a reward. Significantly, the apes produced more tools
when there were more rewards, suggesting that they were able to modulate tool production toward
the specific goal (Bräuer and Call 2015). Even here, though, it is unclear that mental time travel is
required for this sort of planning, since the apparatus and tools are simultaneously available.
The issue is whether the animal has a conscious experience of the future situation, and it can-
not be resolved without a theory about the nature of that sort of experience. I have proposed
that episodic memory and imagination involve representing past and future experience as part
of the present moment. The experience of the past and future is critical to mental time travel; the
information about the events is not. That information could be stored and manipulated using
other forms of memory and thought. Consequently, paradigms that rely on demonstrating use
of information about the past and future will remain unconvincing unless they can capture the
role of consciousness. If the function of mental time travel is to keep track of ourselves in time,
the conscious experience of our personal past and projected future serves to create a temporally
extended self (Droege 2013).
It is not immediately obvious how to design an experimental paradigm that captures the
sense of a temporally extended self. In humans, this capacity arises with a suite of abilities for
navigating our social and cultural world: linguistic self-reference, an understanding of decep-
tion, an appreciation of the feelings of others. We develop an ability to track ourselves in time
in tandem with an ability to track other selves in time. Whatever social pressures demand this
particular distinction between self and other are the pressures that make episodic memory and
imagination more likely in other animals.
5. Conclusion
The phenomenology of episodic memory is its most important and elusive component. The
conscious ‘feeling of pastness’ in episodic memory generates a temporal puzzle of how an expe-
rience in the present can be about the past. Considerations above suggest that an adequate
theory of episodic memory must include at least three elements in order to resolve this puzzle.
First, memories are constructed and so may falsely represent past experience. A theory must be
able to say why inaccurate memories are still memories and what the conditions are for their
truth. Second, the ‘feeling of pastness’ is part of the conscious content of an episodic memory.
The theory needs to explain how this content is manifest and why a present experience of the
past is not a contradiction. Third, the theory should be able to explain what is lost and what
is preserved in cases of episodic amnesia. A person can lose the conscious ‘feeling of pastness’
without losing the capacity to think about past events in her life.
In animal research on memory, the conscious aspect of episodic memory and imagination
has stubbornly resisted empirical demonstration. Despite numerous approaches and very clever
animals, none of the experimental paradigms have shown that a conscious experience of the
past or future situation is necessary to complete the task. A theory of the role of consciousness
in mental time travel is needed to develop a convincing case that animals are capable of forming
a temporally extended self.
I have proposed a temporal representation theory of consciousness and episodic memory
in order to account for the role of consciousness in memory. A representation of the present
110
includes as part of its content a representation of a past experience. The fact that content
about the past is embedded in a representation of the present explains why episodic memory
is conscious, and why a present experience of the past is not a contradiction. Though episodic
memories have the function of representing past experience, they may fail to fulfill this function
and still be memories. Loss of this function entirely in episodic amnesia results in the loss of the
capacity to consciously represent past experiences. Yet amnesics can compensate for that loss
by representing their past (and future) with the help of words, photographs, and the support of
others. A better understanding of the function of the episodic memory system can illuminate
how its functions can be performed in other ways.
Likewise, an understanding of consciousness in the formation of a temporally extended self
can help researchers assess the capacity of animals for episodic memory and imagination. If expe-
riences of one’s personal past and future are important for distinguishing self from other over a
lifetime, then mental time travel will be more likely among animal groups with social conditions
that require this capacity.
In sum, episodic memory is a conscious representation of past experience. You re-experience
the past as part of your experience of the present. This sort of memory emerges in order to track
the self in time so as to navigate a complex social world. Episodic memory and imagination form
a personal past and future that grounds a temporally extended self. Consciousness unifies the
various experiences that you have had and will have by bringing them together in the present.
Notes
1 This point forms the basis of discussion in the Stanford Encyclopedia of Philosophy entry for ‘Memory’
(http://plato.stanford.edu/archives/spr2016/entries/memory/). Likewise, I will often use the unquali-
fied ‘memory’ to refer to episodic memory for the remainder of this section.
2 For a full exposition of the representational relations involved here, see Droege (2013).
Related topics
• The phenomenology of memory
• Memory traces
• Memory and personal identity
• Memory as mental time travel
• John Locke and Thomas Reid
Further reading
S. Bernecker, Memory: A Philosophical Study (Oxford: Oxford University Press, 2010) offers the best over-
all philosophical treatment of memory. K. Michaelian, Mental Time Travel: Episodic Memory and Our
Knowledge of the Personal Past (Cambridge, MA: MIT Press) provides a detailed discussion of the epis-
temological issues in memory and how they are related to conscious, metacognitive processes. The
Frontiers volume, The Long and Short of Mental Time Travel—Self-Projection over Time-Scales Large and
Small (2015) is available free online (https://doi.org/10.3389/fpsyg.2015.00668) and contains articles
from neuroscience, psychology, and philosophy covering a range of related topics.
References
Addis, D.R., Knapp, K., Roberts, R.P. and Schacter, D.L. (2012) “Routes to the past: Neural substrates of
direct and generative autobiographical memory retrieval,” NeuroImage, vol. 59, no. 3: 2908–22.
Addis, D.R., Pan, L., Vu, M.-A., Laiser, N. and Schacter, D.L. (2009) “Constructive episodic simulation
of the future and the past: Distinct subsystems of a core brain network mediate imagining and remem-
bering,” Neuropsychologia, vol. 47: 2222–38.
111
Paula Droege
Bernecker, S. (2010) Memory: A Philosophical Study, Oxford: Oxford University Press.

Boyer, P. (2008) “Evolutionary economics of mental time travel?” Trends in Cognitive Sciences, vol. 12,
no. 6: 219–24.
Bräuer, J. and Call, J. (2015) “Apes produce tools for future use,” American Journal of Primatology, vol. 77,
no. 3: 254–63.
Cheke, L.G. and Clayton, N.S. (2010) “Mental time travel in animals,” Wiley Interdisciplinary Reviews:
Cognitive Science, vol. 1, no. 6: 915–90.
Clayton, N.S. and Dickinson, A. (1998) “Episodic-like memory during cache recovery by scrub jays,”
Nature, vol. 395, no. 6699: 272–4.
Craver, C.F., Kwan, D., Steindam, C. and Rosenbaum, R.S. (2014) “Individuals with episodic amnesia
are not stuck in time,” Neuropsychologia, vol. 57: 191–5.
Dalla Barba, G. (2002) Memory, Consciousness and Temporality, Norwell, MA: Kluwer Academic Publishers.
Dalla Barba, G. and La Corte, V. (2013) “The hippocampus, a time machine that makes errors,” Trends in
Cognitive Sciences, vol. 17, no. 3: 102–4.
Droege, P. (2003) Caging the Beast: A Theory of Sensory Consciousness, Advances in Consciousness Research,
Amsterdam: John Benjamins Publishing.
Droege, P. (2009) “Now or never: How consciousness represents time,” Consciousness and Cognition, vol. 18,
no. 1: 78–90.
Droege, P. (2013) “Memory and consciousness,” Philosophia Scientiae, vol. 17, no. 2: 171–93.
Droege, P. (2015) “From Darwin to Freud: Confabulation as an adaptive response to dysfunctions of con-
sciousness,” in R. Gennaro (ed.) Disturbed Consciousness: New Essays on Psychopathology and Theories of
Consciousness, Philosophical psychopathology, Cambridge, MA: MIT Press, pp. 141–165.
Fernández, J. (2007) “Memory and time,” Philosophical Studies, vol. 141, no. 3: 333–56.
Fernández, J. (2014) “Memory and immunity to error through misidentification,” Review of Philosophy and
Psychology, vol. 5, no. 3: 373–90.
Gutbrod, K., Kroužel, C., Hofer, H., Müri, R., Perrig, W. and Ptak, R. (2006) “Decision-making in
amnesia: Do advantageous decisions require conscious knowledge of previous behavioural choices?”
Neuropsychologia, vol. 44, no. 8: 1315–24.
Kwan, D., Carson, N., Addis, D.R. and Rosenbaum, R.S. (2010) “Deficits in past remembering extend to
future imagining in a case of developmental amnesia,” Neuropsychologia, vol. 48: 3179–86.
Kwan, D., Craver, C.F., Green, L., Myerson, J., Gao, F., Black, S.E. and Rosenbaum, R.S. (2015)
“Cueing the personal future to reduce discounting in intertemporal choice: Is episodic prospection
necessary?” Hippocampus, vol. 25, no. 4: 432–43.
Locke, J. (1689/1975) An Essay Concerning Human Understanding, Oxford: Clarendon Press.
Matthen, M. (2010) “Is memory preservation?” Philosophical Studies, vol. 148: 3–14.
Michaelian, K. (2016) Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past,
Millikan, R.G. (2004) Varieties of Meaning, Cambridge, MA: MIT Press.
Nagel, T. (1974) “What is it like to be a bat?” Philosophical Review, vol. 83, no. 4: 435–50.
Roberts, W.A. (2002) “Are animals stuck in time?” Psychological Bulletin, vol. 128, no. 3: 473–89.
Rosenbaum, R.S., Kwan, D., Floden, D., Levine, B., Stuss, D.T. and Craver, C.F. (2016) “No evidence
of risk-taking or impulsive behaviour in a person with episodic amnesia: Implications for the role of
the hippocampus in future-regarding decision-making,” The Quarterly Journal of Experimental Psychology,
vol. 69, no. 8: 1606–18.
Russell, B. (1921) The Analysis of Mind, Rockville, MD: Arc Manor.
Suddendorf, T. and Corballis, M.C. (2010) “Behavioural evidence for mental time travel in nonhuman
animals,” Behavioural Brain Research, Special issue on episodic memory, vol. 215, no. 2: 292–8.
Tulving, E. (1972) “Episodic and semantic memory,” in E. Tulving and W. Donaldson (eds.) Organization
of Memory, New York: Plenum, pp. 381–403.
Tulving, E. (1985) “Memory and consciousness,” Canadian Psychology, vol. 26: 1–12.
Zentall, T.R., Singer, R.A. and Stagner, J.P. (2008) “Episodic-like memory: Pigeons can report location
pecked when unexpectedly asked.” Behavioural Processes, vol. 79, no. 2: 93–8.
112
9
MEMORY AND PERSPECTIVE
1. Introduction
In an essay in the London Review of Books, the late Jenny Diski describes a remembered scene
from her childhood. Aged 6 or so, she is seated on her father’s knee. Her father, she tells us,
looks just like he does in the pictures she has of him: ‘silvery hair, moustache, brown suede
lace-ups’. Diski doesn’t have many pictures of her childhood self but she’s pretty sure her
remembered image of herself at that age is accurate. The layout of the room is also correct:
‘Door in the right place; chair I’m sure accurate, a burgundy moquette; patterned carpet; win-
dows looking out onto the brick wall of the offices opposite’. Indeed, Diski had even gone back
to the block of flats and ‘sat in the living-room of the flat next door’ just to verify the layout
and confirm its accuracy. Nonetheless, for Diski, there is still something rather ‘odd’ about this
particular memory. She writes:
Here’s the thing, though: I can see the entire picture. I can . . . see myself. My obser-
vation point is from the top of the wall opposite where we are sitting, just below the
ceiling, looking down across the room towards me and my father in the chair. I can
see me clearly, but what I can’t do is position myself on my father’s knee and become
a part of the picture, even though I am in it. I can’t in other words look out at the
room from my place on the chair. How can that be a memory? And if it isn’t, what is
it? When I think about my childhood, that is invariably one of the first ‘memories’ to
spring up, ready and waiting: an untraumatic, slightly-moving picture. It never crossed
my mind to notice the anomalous point of view until I was middle-aged. Before then
it went without saying that it was a ‘real’ memory. Afterwards, it became an indicator
of how false recollection can be.
(Diski 2012: 12)
The scene Diski describes is recalled from what’s known as an ‘observer perspective’. In this
external visual perspective, Diski views herself in the remembered scene. For Diski, the falsity of
this memory stems solely from this ‘anomalous point of view’. In all other respects, the memory
is accurate: her father’s appearance; the layout of the flat – a fact which she even verified in
adulthood; and the phenomenology of the mental image is such that it presents as a memory,
113
not merely imagination. The only reason Diski doubts this memory is that she is recalling the
episode from an external visual perspective. As we will see, the allegedly ‘anomalous point of
view’ of observer perspectives is often taken to show that such memories simply cannot be
genuine.
In this chapter, we discuss the phenomena of perspectival memory. While surveying the
field, we suggest that visual perspective alone is not a guide to the truth or falsity of memory,
and that genuine memories can be recalled from an observer perspective. Such memories can
satisfy conditions placed on genuine memory. Observer perspectives can satisfy factivity con-
straints, and can stand in appropriate causal connections to the past. In the next section, we
identify the phenomena and provide an overview of some of the empirical evidence related
to point of view in personal memory. We articulate some doubts about remembering from an
observer perspective, before responding to these worries. We suggest that observer perspectives
may retain other forms of internal imagery: there is no neat division between internal and exter-
nal perspectives. We suggest that external perspectives may help in understanding the past, and
we question the primacy of egocentricity.
2. Field and observer perspectives

The imagery involved in remembering past episodes in one’s life often involves visual points of
view. When we recall a past event, we usually adopt the same perspective that we had at the
time of the original experience. We see the scene as we originally saw it from a first-person or
‘field’ perspective. Sometimes, however, we recall the past event from an external visual per-
spective, from a position we didn’t occupy at the time of the original episode. In such cases,
we view ourselves in the remembered scene, as from a third-person ‘observer’ perspective.
Nigro and Neisser conducted the first systematic experimental studies on visual perspective
in memory, and their terms ‘field’ and ‘observer’ memories became part of the vocabulary of
memory studies.
Since Nigro and Neisser’s paper, empirical research has produced a number of consistent
findings concerning these differing points of view. The field perspective is more common.
Observer perspectives are more common, though, in certain circumstances. One robust empiri-
cal result is that observer perspectives are more common for older memories, such as memories
of childhood (Nigro and Neisser 1983). Observer perspectives are also more common for events
that involve a high degree of emotional self-awareness (Nigro and Neisser 1983; Robinson and
Swanson 1993). Field perspectives seem to be related to remembering the emotional details,
feelings, or psychological states associated with an event; in contrast, observer perspectives tend
to include less sensory and affective detail but more information related to concrete, objective
details (Nigro and Neisser 1983; McIsaac and Eich 2002; Rice 2010).
Another study looking at emotion and visual perspective in memory found that although
there was no difference in reports of emotional intensity between field and observer perspec-
tives, when subjects switched from a field to an observer perspective there was a resulting
decrease in reported emotional intensity. There was no corresponding change in emotional
intensity, however, when switching from an observer to a field perspective (Robinson and
Swanson 1993; see also Rice 2010: 233–4).
This last point also suggests that these visual perspectives in memory are not fixed. In
Robinson and Swanson’s (1993) study, participants recalled an event from a particular perspec-
tive (e.g. field), and sometime later recalled the same event from the alternate perspective (e.g.
observer). However, evidence indicates that one can often switch between perspectives within
a single episode of remembering a past event. That is, remembering a past event may involve
114
Memory and perspective
adopting not just a field or an observer perspective, but may involve adopting both perspectives
in the same retrieval attempt (Rice and Rubin 2009).
Even if the term ‘perspective’ bears a visual bias, it refers more generally to the range of
imagery or ‘standpoints’ in distinct modalities that informs one of one’s body, the world, or
even other perspectives (Behnke 2003: 52). There are many different kinds, domains, and
modes of ‘perspective’. Perspectives can be cognitive, embodied, emotional, or evaluative in
nature; they occur in many domains, including imagination, perception, and memory, and they
can be first-, second-, or third-personal. These distinct perspectives and forms of perspective
stand in many different relations to each other. By initially insisting on such distinctions between
different kinds, domains, and modes of perspective, we can then investigate their coexistence,
fusion, integration, and coordination.
The distinction between field and observer perspectives in episodic memory is paralleled
in other cognitive domains: in imagination (e.g. Vendler 1979; Walton 1990; Williams 1973;
Wollheim 1984); in dreaming (e.g. Cicogna and Bosinelli 2001; Rosen and Sutton 2013; Windt
2015). Even in the domain of spatial cognition, one can adopt points of view that are internal or
external to the subject. Spatial information can be processed and communicated from egocentric
(route or embedded) points of view and allocentric (extrinsic or survey) perspectives. In fact,
just as an episodic memory may involve both field and observer perspectives, spatial informa-
tion is often interpreted and conveyed by integrating and blending these distinct points of view
(Tversky 2011).
3. Remembering from an observer perspective:

truth and authenticity
In most studies on visual perspective in memory, the observer perspective is simply taken as one
particular instance of remembering a past event. These studies do not normally question the
authenticity of such memories in which one sees oneself from an external perspective. Some
psychologists studying the phenomena of point of view in memory are interested in the ques-
tion of whether, as a matter of fact, more observer perspectives than field perspectives tend to
be false. This is quite distinct from the question of whether there can in principle be genuine or
veridical memories in which one adopts an observer perspective.
We saw that Diski casts doubt on her childhood observer perspective memory solely because
of its anomalous point of view. But if one takes oneself to be remembering, and one is accurately
representing some past event in all aspects other than occupying the original point of view, what
motivates the claim that such representations are not ‘real’ memories?1 Why would an external
perspective entail a false memory? Diski seems to assume an idea which is also apparent in some
philosophical work, the idea that memory should preserve the content of perception. In percep-
tion, one sees an event unfold from a particular point of view. And because memory preserves
the content of perception, the remembered event should be recalled from the same point of
view one had at the time of the original experience.
The idea that memory exactly reproduces a past experience seems to put pressure on the
status of observer perspectives as genuine memories. Since ‘remember’ in relevant senses is typi-
cally used as a success word, memory implies truth: this is a factivity constraint on remembering.
On the view we are here describing, truth in memory is taken to require duplication of the past.
But observer perspectives are not duplicates of the past event, because they present the event
from an ‘anomalous’ point of view. Thus, on this line of thought, observer perspectives cannot
appear in genuine remembering. On such a preservationist view, genuine memories should be
recalled from a field perspective.
115
Yet this cannot be the whole picture. It is possible to accept the factivity constraint on
memory, yet deny that memory involves strict preservation; a degree of change may still be
compatible with truth. This is a point accepted, for example, by Sven Bernecker, a moderate
preservationist: ‘Memory implies truth, but it does not imply that the memory content is an
exact duplicate of the past thought content. Sometimes memory allows for moderate transfor-
mations of the informational content’ (2008: 155).
Further, the preservationist account of memory is itself called into question by reconstruc-
tive models of memory, which emphasise the flexible and dynamic nature of remembering
(e.g. Schacter and Addis 2007). For Frederic Bartlett, who conducted pioneering work on
reconstruction in memory, ‘Remembering is not the re-excitation of innumerable fixed, life-
less and fragmentary traces. It is an imaginative reconstruction, or construction, built out of the
relation of our attitude towards a whole active mass of organised past reactions or experience’
(1932: 213). But construction in memory should not be equated with error or invention: mal-
leability is not in itself unreliability (Barnier et al. 2008). Memories can be influenced, ‘worked
over’, constructed, compiled, and still be functional, faithful, accurate, true.
A broadly preservationist line of thought lies behind Richard Wollheim’s rejection of the
possibility that I might see myself in a (genuinely) remembered scene. Wollheim claims that
‘would require that I be represented as from the outside, but the fact that it is an event memory
forbids this, for this isn’t how I experienced myself in the course of the event’ (1984: 103).2
Zeno Vendler articulates a similar worry: ‘one cannot remember seeing oneself from a different
perspective simply because it is impossible to have seen oneself from an outside perspective’ (1979:
169; original emphasis). And, Vendler explains, this conclusion simply follows from the truism
that ‘one cannot remember doing something that one has not done’ (1979: 170).
On such views, because one did not (indeed cannot) see oneself from an external perspective
at the time of the original experience, one cannot have a memory in which one sees oneself
from an external perspective: one cannot recall from an observer perspective. But perhaps this is
to set an unrealistic standard for what a genuine observer perspective in memory would have to
be – a requirement of having visually perceived oneself during the original event. We suggest,
in contrast, that in order to ‘see’ oneself in memory from an observer perspective, one does not
need to have visually perceived oneself from an external perspective at the time of the original
event. Even if we grant that one cannot see oneself from an external perspective, one can still
have a memory in which one ‘sees’ oneself from an external perspective.
This point is nicely made by Dominic Gregory:
my own observer memories do not involve its seeming to me that things once looked
to me the ways that the visual mental images show things as looking; I do not seem to
be recalling episodes in which I somehow saw myself. Rather, they involve its seem-
ing to me that there were once past scenes in which I played a certain part and which
looked – ‘from somewhere’ rather than ‘to someone’ – the ways that the visual mental
images show things as looking.
(Gregory 2011: 2; see also Gregory 2013: 203–4)
Discussing the impact of present context on the content of memory, Peter Goldie argues that
what one now knows, thinks, or feels may infuse the memory of a past event. For Goldie, the
content of memory can be influenced at the point of retrieval by present knowledge and emo-
tion. He tells us that ‘in effect, I remember it as I now feel it’ (2012: 52). According to Goldie,
observer perspectives are more likely to occur when there is an epistemic, emotional, or evalu-
ative gap – what Goldie terms a triply ironic gap – between past and present. In other words:
116
what one now knows, thinks, and feels, is different to what one then knew, thought, and felt. It
is the (ironic) gap that opened between the past and the present that affords the possibility of a
memory from an observer perspective.
Goldie provides the example of remembering drunkenly singing at the office party: feeling
at that time a ‘heady delight’ but now shamefully realising that his colleagues were laughing at
him and not with him. For Goldie, such a memory will typically be recalled from an observer
perspective: ‘I can see myself now, shamefully making a ridiculous fool of myself in front of
all those people, getting up on the table and gleefully singing some stupid song’ (2012: 52).
Importantly though, ‘field episodic memories – memories of what happened “from the inside” –
can also be infected with irony, with what one now knows, and how one feels about what one
now knows’ (Goldie 2012: 52).
That both field and observer perspectives memories involve constructive elements is a point
acknowledged by Dorothea Debus (2007). Debus also argues that observer perspective mem-
ories are consistent with a causal theory of memory: observer perspectives can maintain an
appropriate causal connection to the past. Debus argues that, despite the external visual perspec-
tive of observer memories, the information involved in such imagery has its source in the original
experience. For Debus, the shift in point of view between the original perceptual experience
and the subsequent observer memory results from a systematic modification of the spatial infor-
mation available at the time of encoding. Spatial information available at the time of the original
experience – and hence appropriately causally connected to the past – is systematically manipu-
lated into an observer perspective image. This seems to be the case for Jenny Diski’s memory, in
which spatial relations between the elements of the remembered scene appear to be maintained
through the shift in visuospatial perspective.
Nonetheless, a further related preservationist argument may be levelled against observer per-
spective memories. Even if it is accepted that perfect preservation is unrealistic, it could be
claimed that memory should still broadly preserve the content of a past perceptual experience.
Aspects of the original perceptual content may be lost from the memory – memory degrades
with time and forgetting is natural – but nothing should be added to the content of a genuine
memory. In just such a moderate departure from strict preservationism, Bernecker argues that
‘In the process of remembering, the informational content stored in traces may stay the same or
decrease (to a certain degree); but it may not increase’ (2008: 164).3
An argument against observer perspective memory can then be formulated thus: genu-
ine memory involves only content that was available at the time of the original experience.
Observer perspectives seem to involve a representation of the self that was not available at the
time of perception. Therefore observer perspectives involve additional content and cannot be
genuine memories.
One response to this argument is to urge that genuine memory can be generative. Kourken
Michaelian argues that on a (re)constructive model of memory new content can be generated:
The generation of new content occurs when memory produces content in addi-
tion to that which it took as input; this can occur either before retrieval, by means
of transformation of content received from other sources, or at retrieval, by means of
transformation of content stored by memory.
(Michaelian 2011: 324)
Memory processes allow that new content can be added to genuine memory. Therefore, even
if observer perspectives have an additional representation of the self, they can still count as
genuine memories. In a recent paper, Bernecker (2015) addresses visual memory and the extent
117
to which its content can differ from the content of a previous perception. Bernecker discusses
the possibility that observer perspectives may be counted as genuine cases of inferential memory.
Inferential memory is ‘remembering with admixture of inferential reasoning involving back-
ground knowledge or fresh evidence’ (Bernecker 2010: 77). For example, one may see a par-
ticularly beautiful bird in the park without knowing what type of species it is. Being something
of an amateur ornithologist, one then consults one’s book on Australian birds and finds out that
the bird was a Kookaburra. In saying that one remembers seeing a Kookaburra, one is inferen-
tially remembering, because one did not know it was a Kookaburra at the time of the original
experience (adapted from Malcolm 1963: 223; Bernecker 2010: 25). Non-inferential memory
does not involve such inferential reasoning.4 Importantly, and in line with Michaelian’s proposal
for generative memory, Bernecker holds that ‘While non-inferential memory allows only for
the decrease of information, inferential memory also allows for the increase or enrichment of
information’ (2015: 457).
Bernecker writes:
Should observer memories count as genuine memories? The main reason to answer
in the negative is that observer memories contain information that wasn’t available to
the subject at the time of the original representation. But then all inferential memo-
ries are admixed with inferential reasoning involving background knowledge or fresh
information. What, if anything, distinguishes observer memories from other inferential
memories? To not count observer memories as inferential memories it would have
to be shown that the fresh information contained in observer memories is false or
unreliable. However, there is no evidence to suggest that memories from the observer-
perspective are any less reliable than memories from the field-perspective.
Bernecker suggests that the main difference between field and observer perspectives lies in their
emotional content and concludes that ‘given that memories from the observer-perspective are
not less reliable than memories from the field-perspective I see no reason to not count them as
instances of inferential memory’ (2015: 461–2). Observer perspectives are therefore permissible
as inferential memories because such memory allows for the generation of content.
But even if observer perspectives can be classed as genuine (inferential) memories, and
hence not outright false memories, they are sometimes still taken to be examples of ‘distorted
memories’ (e.g. De Brigard 2014; Fernández 2015).5 Again, the thought is that because the
event remembered did actually happen the memory is not false, and the factivity condition
is satisfied; but because that event is remembered from an observer perspective, and so the
content of the memory is different from that of perception, the memory is distorted. This
understanding reflects a distinction Bernecker appeals to between truth and authenticity: ‘a
memory state must accord not only with objective reality but also with one’s initial perception
of reality’ (Bernecker 2010: 214). For Bernecker, genuine (non-inferential) memory must sat-
isfy both conditions – truth and authenticity (2010: 39). We suggest that observer perspectives
need not be considered distorted memories. Observer perspectives can satisfy both truth and
authenticity conditions.
How can observer perspectives accord with one’s initial perception of reality? We suggest that
observer perspectives may be constructed in part from external perspectival information available
during perception. Emotions, thoughts, and images which are experienced during the original
episode may be used in the construction of observer perspective memories of the past event.
Even though these experiences are internal, they can involve adopting an external perspective
118
on oneself. Recall that observer perspectives are more common for events that involve a high
degree of self-awareness. We suggest that during such emotionally charged events, one’s literal
(visual) perspective is internal, but one may adopt an external thoughtful or emotional perspec-
tive on oneself. And it is from this ‘external’ perspectival information that observer perspectives
can be constructed.
During perceptual experience an agent may make use of both egocentric and allocentric spa-
tial information. Observer perspective memories may be constructed from this non-egocentric
information available at the time of encoding. Mohan Matthen tells us that:
Field-perspective memory presents scenes in egocentric terms – how they look through
the eyes of the observer. Observer-perspective memory is in allocentric terms: it is an
expression of observer independent spatial relations in the remembered scene . . . Now,
we know that visual perception incorporates both forms simultaneously . . . In view
of this, many cognitive scientists hypothesize that visual content contains allocentric
information as well – perhaps we have a map or model stored away in visual memory.
The point to take from easy switching between field and observer perspectives is that
in episodic memory, the egocentric and allocentric forms are somehow separated out
and expressed in two different, alternating perspectives.
(Matthen 2010: 13)
In most cases, one attends to egocentric visual information available during a perceptual experi-
ence. But non-egocentric perspectives are available during perceptual experience too, and some-
times one’s attention is focused on this non-egocentric information.6 These different perspectives
provide different information on the same scene: they provide different ways of thinking about
the same episode. Distinguishing episodic from semantic memory, Mark Rowlands remarks
that ‘What is distinctive of episodic memory is the way in which facts are presented: they are
presented by way of experiences. And these experiences, in turn, are presented as ones that the
subject had at the time of the episode’ (2009: 337). We suggest that field and observer perspec-
tives involve different ways of thinking about the same past event, presenting the same event in
different ways. They involve different forms of information that are both available at the time of
encoding (see also McCarroll and Sutton 2016).
Rowlands suggests that in observer perspectives ‘you may well be accurately remembering
the episode itself . . . However, you do not accurately remember the experiences that presented
the episode in its occurrence. You seem to be remembering visual experiences that you could
not have had’ (2009: 340–41). In contrast, we suggest that observer perspectives can be both
true and authentic: they can both represent an event that occurred and the experiences occur-
ring at the time of the event. Further, not only can observer perspectives make use of allocentric
information that was available at the time of the original experience: such memories can also
maintain perspectival information in distinct modalities. In this way, observer perspectives may
accurately represent the experiences (kinaesthetic, emotional, even imaginative) that one had
during the original event, and that are recalled in genuine episodic memory.
4. The plurality of perspectives

In many cases of memory imagery, one may adopt an external visual perspective and yet main-
tain an internal perspective in other embodied, emotional, or cognitive modalities. Yet many
theorists retain, if only implicitly, an exclusive association between kinaesthetic, embodied, or
emotional imagery and an internal visual perspective. This is coupled with the parallel position
119
that an external visual perspective is (necessarily) isolated from such forms of embodied imagery
(Vendler 1979; Williams 1973). In this section, we discuss the complex relations between inter-
nal and external perspectives in distinct modalities.
Perspectival imagery need not be consistently either internal or external across all modali-
ties. An external visuospatial perspective on a past experience is compatible with an internal
embodied (kinaesthetic or emotional) perspective. We can underline the way different per-
spectival modalities can thus come apart in memory by considering the parallel case of film.
In film, point of view (POV) shots represent the visual perspective of a character involved in
the action; even though they are removed from the domain of memory, they may be roughly
analogous to a field perspective. It has been argued that such POV shots invite the viewer
to take up the position of a character in the narrative the film portrays, perhaps through
imagining from-the-inside, or empathising with the character (Messaris 1994: 33). The vis-
ual perspective invoked in POV shots may sometimes thus invite the spectator to adopt, or
empathise with, the character’s perspective in other respects or modalities too: but this is not
necessarily so. As Murray Smith notes:
POV may be particularly effective in rendering how a character sees, and so enabling
our imagining from the inside how the character sees, but it is not particularly use-
ful in evoking, say, a character’s joy or humiliation or anxiety. Emotional simulation
certainly does not need a POV shot in order to be prompted.
(Smith 1997: 418)
Consider POV shots that represent an evil or monstrous character in the film, stalking or lying in
wait for another character. In such cases, even though one shares or adopts the visual perspective
of the monster, say, one’s emotional and kinaesthetic perspectives may be far from in harmony
with that creature. One may feel the emotions of the individual the beast is watching; one may
feel the fear or terror of the victim rather than the excitement or bloodlust of the fiend. Here,
different perspectives can come apart: one shares the (internal) visual perspective with one char-
acter, while at the same time not in any way sharing that character’s affective perspective, and
perhaps even adopting the external emotional perspective on that character which is held by the
victim. The point of view may be visually internal, as if one were seeing the action through the
eyes of one of the characters, but in other modalities – such as emotional or kinaesthetic – one’s
perspectives need not neatly align with those of that character. This example shows us how a
visual field perspective can be coupled with an ‘external’ emotional perspective.
One powerful and disturbing example of the divergence of perspectives within cinematic
point of view is found in Jonathan Demme’s The Silence of the Lambs.7 In the concluding
sequence, Detective Clarice Starling (Jodie Foster) confronts the notorious serial killer ‘Buffalo
Bill’. We see Clarice, groping around in the pitch dark, wielding her gun but unable to see,
visibly shaking and terrified for her life; but we see her from the terrible ‘night goggles’ perspec-
tive of Buffalo Bill. We are visually aligned with the killer, we see Clarice from his point of
view through the eerie green of the lens he is wearing; yet our emotional and even kinaesthetic
perspectives are more aligned with Clarice: we feel her terror, her helplessness.8 One’s visual
perspective on a remembered, imagined, or filmed scene can diverge from or align with per-
spectives in other modalities: there is no simple internal/external dichotomy.
Sports psychology offers further indications of the multimodality of perspectives. There is
evidence that embodied imagery is not exclusively tied to an internal visual perspective. Morris,
Spittle and Watt tell us that:
120
Researchers have found that participants are able to form kinesthetic images equally
well with either [visual] imagery perspective . . . and more recent research even sug-
gests that for some tasks, kinesthetic imagery may have a stronger association with
external [visual] imagery than with internal imagery.
(Morris et al. 2005: 129–31)
This illustrates that an observer perspective in visuospatial imagery can be coupled with internal
kinaesthetic imagery. The tasks which are purported to have a stronger association with exter-
nal visual imagery are open rather than closed skills: football rather than darts, for example. In
open skills, the external environment (the position of other players, say) may have an impact on
successfully performing the action, and bodily form in movement may be important. In these
open skills, egocentric and allocentric information are integrated.
Consider the following example in which the professional footballer Wayne Rooney discusses
his use of imagery in preparation for matches:
Part of my preparation is I go and ask the kit man what colour we’re wearing – if it’s
red top, white shorts, white socks or black socks. Then I lie in bed the night before
the game and visualize myself scoring goals or doing well. You’re trying to put yourself
in that moment and trying to prepare yourself, to have a ‘memory’ before the game. I
don’t know if you’d call it visualizing or dreaming, but I’ve always done it, my whole
life . . . when you get older and you’re playing professionally, you realize it’s impor-
tant for your preparation – and you need to visualize realistic things that are going to
happen in a game.
(Winner 2012; see also Sutton 2012)
This ‘memory before the game’ appears to involve Rooney visualising himself performing from
an observer perspective: note his attention to external details such as the colour of the kit.
Rooney is using external visual imagery to prepare for professional football matches. As part of
his preparation, Rooney cultivates internal kinaesthetic imagery which coheres with his external
visual imagery, such that internal and external perspectives fuse.
To return to autobiographical memory, in their 1897 survey of earliest recollections Victor
and Catherine Henri note that observer perspective memories are common in memories
from childhood. But they suggest that while such memories present a visual representation
of oneself as a child, they are in a sense distanced from any internal feeling accompanying the
memory:
A large number of responses contain the same affirmation about the way that remem-
berers see themselves in memory: they see themselves as children, they do not feel
themselves children, they have a representation in which a child appears, and they
know that they are that child: “I see myself in sickness like someone who is outside
of me.” “I’m at the seashore and my mother is holding me upon her arms; this scene
appears to me as though I were far away from it.” Such are the observations that are found
in many of the responses.9
(Nicolas et al. 2013: 370; original emphasis)
Observer perspectives may tend to involve less emotion or embodied imagery: but this is not
necessarily so, as the following response from the Henris’s survey demonstrates:
121
I had the croup when I was 12 months old, and they had to burn all the lumps in my
throat. I have a very clear visual image of the scene; I distinctly see four people holding
me down by force, laid out on one side; what I see most of all is the scorching brazier
where two red irons are heating until they are red-white; right now, I still seem to feel
that burning iron approaching my lips.
(2013: 370; original emphasis)
On one reading of this passage, it is a memory recalled from an observer perspective: the
respondent sees him or herself in the scene, being pinned down to receive the gruesome treat-
ment. But the memory is also infused with kinaesthetic and, perhaps, emotional elements. The
memory articulates the external visual perspective as well as the simultaneous emotional and
embodied perspectives: it evokes the fear of the searing heat as well the sense of danger looming
towards the subject, invading personal space.10
So not all the features or qualities that are experienced from-the-inside are lost in the observer
perspective. Visual, emotional, kinaesthetic, and other embodied perspectives may come apart:
there is a plurality of perspectives (Sutton 2010). If we consider that ‘neither affect or kinaes-
thesis need be determined by visual perspective, or even inevitably follow it, we make room for
a range of relations between these distinct modalities to operate in different contexts’ (Sutton
2014: 143).
The plurality of perspectives is not only restricted to embodied, experiential, or emotional
imagery. As Goldie (2012) argues, it may also involve cognitive or evaluative perspectives
on the past which may be either internal (reflecting considerations at that time) or external
(bearing knowledge that was not available in the past). These cognitive or evaluative per-
spectives may or may not align with visual perspective in personal memory: there is no neat
internal/external divide.
The view we discussed above – that genuine autobiographical memories can only involve
field perspectives – reflects the thought that egocentric perspectives are natural and primary. One
example of this tendency to favour egocentricity can be seen in a study on the use of drawings as
a means of lie-detection. Aldert Vrij and colleagues argue that people who draw a remembered
scene truthfully are likely to sketch it from an internal perspective, as if from a shoulder-mounted
camera, while ‘liars’ will draw it as from an overhead or external vantage-point. Truth-tellers use
more direct phrases, phrases such as ‘I saw’ that imply direct perceptual experience, while ‘liars
are more likely to convey indirect, hypothetical knowledge (e.g. “I would see . . .”)’ (Vrij et al.
2010: 588). The authors hypothesise that this distinction will hold for scenes that participants
draw, either:
From a ‘shoulder camera’ (observer) position, where someone sketches what she/he
could actually see, or from an ‘overhead’ (actor) position, where someone sketches the
location as it could be seen from the air. The former is more direct and likely to be the
result of actual first-hand experience than the latter, which ‘removes’ the participant
from the scene. We thus predicted that more truth tellers than liars would sketch the
drawing from a shoulder camera position.
(Vrij et al. 2010: 588)
Yet this way of thinking arguably misses the ordinary mingling of route (internal, field) and
survey (external, observer, overhead) perspectives in spatial cognition.11
We argued above that we think about, process, and communicate spatial information from
both egocentric and allocentric perspectives. Indeed, work on spatial cognition calls into question
122
any notion of egocentric primacy: ‘The primacy of egocentric perspective has been challenged
by research showing that rats, monkeys, and people on first encountering an environment
immediately form multiple representations of space, in particular, allocentric representations’
(Tversky and Hard 2009: 124). In studies demonstrating how we often naturally adopt another’s
spatial perspective as a means to improve action understanding, Tversky and Hard conclude that
‘the deep meaning of embodied cognition is that it enables disembodied thought’ (2009: 129).
The mind is not always bound by limitations of the physical world.
We suggest therefore that external perspectives offer another way to interpret the world. The
intermingling of the multiple internal and external perspectives that one can adopt when remem-
bering provides a way of understanding the past that goes beyond a purely egocentric point of view.
This intermingling of perspectives is seen in the anthropologist Bradd Shore’s (2008) research
on memory work at long-running annual religious camp meetings at Salem. At these camps,
older adults spend time watching the younger campers engage in a range of activities: bible
readings, sports, and arts and crafts:
Over time at camp meeting, people come to watch their kids doing exactly what they
did. This effects an alternation between field and observer memories and a kind of
blurring that allows campers to ‘participate’ in the lives of their offspring at the same
moment as they gain reflexive distance . . . In its subtle orchestration of memories of
doing and of watching over time, Salem provides perfect conditions for the fusion of
observer and field memory; conditions that ultimately inform narrative expression and
create a powerful sense of identification in ‘family’ over the generations.
(Shore 2008: 114)
This blurring of perspectives is a fusion that affords a greater degree of understanding of the
past. It may take a mix of internal and external perspectives to fully understand and appreciate
a past event.
5. Conclusion
The imagery of personal memory involves a plurality of perspectives. In remembering the past,
we can adopt a range of viewpoints, internal and external, visual and non-visual, which can
fuse or integrate in various ways. Even in the present moment, we have ways of getting outside
ourselves. Remembering from an observer perspective, from an external visual point of view,
is but one way we have of thinking about and understanding our past. Sometimes adopting an
external point of view can help put the past in perspective.
Notes
1 Context and pragmatic considerations change the goals and functions of remembering. Whether one is
testifying in a court of law or reminiscing round the dinner table affects how we think about the truth
of a memory. See Sutton (2003), Campbell (2014), Harris et al. (2014).
2 On Wollheim’s conception, event-memory as a species of memory for events that one experienced
is closely related to episodic memory. Wollheim stresses that it is an exaggeration to say that in event-
memory one must remember the event exactly as one experienced it. Rather, in a genuine memory
one should broadly remember the event as one experienced it (Wollheim 1984: 103–4). But these
broad limits to the mnemonic content must not include ‘gross deviations’, which include ‘structural
deviations, or deviations in identity’ (1984: 103). The external visuospatial representation of the self
in observer perspectives would, on his view, amount to such a deviation. In response, we suggest that
observer perspectives need not involve structural deviations or deviations in identity.
123
3 This idea reflects a distinction in psychology between errors of omission and errors of commission.
When memory fails, it can do so by way of either errors of omission – typically errors of forgetting
or memory failures; or errors of commission – when details are remembered that were not part of the
original event. Errors of commission are often called false memories, in which one ‘falsely remembers
details, words, or events that weren’t actually experienced’ (Intraub and Dickinson 2008: 1007).
4 Non-inferential and inferential memory are sometimes referred to as ‘pure’ and ‘impure’ memory
respectively. Bernecker finds these labels unfortunate because they imply that inferential memory is
somehow inferior even though it is a pervasive form of memory (2010: 25). Nonetheless, Bernecker’s
analysis of memory (2010) concentrates predominantly on non-inferential memory.
5 De Brigard says that distorted memories ‘present the remembered content in a somewhat distorted way,
that is, as a distortion of the content encoded during the original experience’ (2014: 160). Fernández
distinguishes two types of distorted memories corresponding to ‘storage’ (preservationist) and ‘narrative’
(reconstructivist) conceptions of memory.The former is important in this context: ‘On the storage con-
ception of memory, a subject’s faculty of memory has produced a distorted memory when the content
of that memory does not match the content of the subject’s past experience on which the memory
originates’ (Fernández 2015: 539).
6 It is beyond the scope of this chapter to go into detail on how non-egocentric information may be
constructed into an observer perspective. For a more comprehensive exposition, including a discussion
of the cross-modal translation of non-visual to visual information, see McCarroll (2015).
7 Thanks to Robert Sinnerbrink for this example.
8 The scene can be viewed here: www.youtube.com/watch?v=OQZYz7qR0Fo.
9 The Henris’s 1897 paper in French was partly translated in the American journal Popular Science Monthly
in 1898. Our quotations are from the complete translation by Nicolas et al. (2013: 370).
10 Arguably, the ambiguity between a field and an observer perspective inherent in this description points
to the fact that both types of memories can be emotional.
11 While Vrij et al. did find that ‘significantly more truth tellers (53 per cent) than liars (19 per cent) sketched
the drawing from a shoulder camera position’ (2010: 592), nonetheless almost half of the truth tellers drew
the scene from the ‘anomalous point of view’ said to characterise liars. Indeed, we doubt that an ‘own-eyes’
point of view is intrinsically tied to reality. In a study on point of view in spontaneous waking thought,
observer perspective thoughts were more likely to be memory reports, whereas field perspective reports
included more fantasies such as seeing ‘a slice of ham hovering in space’ (Foulkes 1994: 682).
Related topics
• Phenomenology of remembering
• Memory and truth
• Intentional objects of memory
• Memory and imagination
• Memory images
• Memory and emotion
• Memory and narrativity
• Frederic Bartlett
Further reading
Nigro and Neisser (1983) provide the first empirical investigation into the phenomena of visual perspective
in memory. Nigro and Neisser’s main hypothesis is that observer perspectives are the products of recon-
struction in memory, and are a sign of the malleability of memory. The idea that observer perspectives are
the product of mnemonic change can be traced to Freud’s essay Screen Memories (1899/2001). Nonetheless,
Nigro and Neisser also point to the possibility of observer perspective experiences. Rice (2010) provides
an excellent overview of some of the key empirical work on visual perspective in memory.
For the view that observer perspectives are false or distorted memories, see Vendler (1979) and
Fernández (2015). For views defending observer perspectives as genuine instances of memory, see Sutton
(2010); Debus (2007) shows that observer perspectives can be appropriately causally connected to the past,
124
and Bernecker (2015) suggests that observer perspectives can count as genuine inferential memories. For a
longer defence of observer perspective memories, see McCarroll (forthcoming).
The notion of perspective in memory touches on a range of other interesting issues such as self-repre-
sentation in imagery and how perspectival information from distinct domains is integrated in imagery. As
such, there are interesting connections between the issues raised by perspective in memory and work in a
diverse range of disciplines: areas such as sports psychology (Morris and Spittle 2012), philosophy of film
(Smith 1997), imagination (Wollheim 1984), dreaming (Rosen and Sutton 2013), philosophy of language
(Recanati 2007), personal identity (Goldie 2012), and moral psychology (Mackenzie 2007).
References
Barnier, A. J., Sutton, J., Harris, C. B. and Wilson, R. A. (2008). A conceptual and empirical framework
for the social distribution of cognition: The case of memory. Cognitive Systems Research, 9(1): 33–51.
Bartlett, F. C. (1932). Remembering: A Study in Experimental and Social Psychology. Cambridge: Cambridge
University Press.
Behnke, E. A. (2003). Contact improvisation and the lived world. Studia Phaenomenologica, 3(Special):
39–61.
Bernecker, S. (2008). The Metaphysics of Memory. Dordrecht: Springer.
Bernecker, S. (2015). Visual memory and the bounds of authenticity. In D. Moyal-Sharrock, V. Munz and
A. Coliva (eds), Mind, Language, and Action: Proceedings of the 36th International Wittgenstein Symposium
(pp. 445–63). Berlin: de Gruyter.
Campbell, S. (2014). Our Faithfulness to the Past: The Ethics and Politics of Memory. Oxford: Oxford University
Press.
Cicogna, P. C. and Bosinelli, M. (2001). Consciousness during dreams. Consciousness and Cognition, 10(1):
26–41.
De Brigard, F. (2014). Is memory for remembering? Recollection as a form of episodic hypothetical thinking.
Synthese, 191(2): 155–85.
Debus, D. (2007). Perspectives on the past: A study of the spatial perspectival characteristics of recollective
memories. Mind and Language, 22(2): 173–206.
Diski, J. (2012). The me who knew it. London Review of Books, 34(3): 12–13. www.lrb.co.uk/v34/n03/
jenny-diski/the-me-who-knew-it.
Fernández, J. (2015). What are the benefits of memory distortion? Consciousness and Cognition, 33: 536–47.
Foulkes, D. (1994). Point of view in spontaneous waking thought. Perceptual and Motor Skills, 78: 681–2.
Freud, S. (1899/2001). Screen memories. In J. Strachey (ed.), The Standard Edition of the Complete Works of
Sigmund Freud (Vol. 3, pp. 303–22). London: Vintage.
Goldie, P. (2012). The Mess Inside: Narrative, Emotion, and the Mind. Oxford: Oxford University Press.
Gregory, D. (2011). Sensory memories and recollective images. Paper presented at conference ‘Perceptual
Memory and Perceptual Imagination’, September 2011, University of Glasgow. https://docs.google.
com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnxkb21pbmljZ3JlZ29yeXBoa
Wxvc29waHl8Z3g6NjQ0NTczZWRkZjJlMjZhOA.
Gregory, D. (2013). Showing, Sensing, and Seeming: Distinctively Sensory Representations and Their Contents.
Oxford: Oxford University Press.
Harris, C. B., Rasmussen, A. S. and Berntsen, D. (2014). The functions of autobiographical memory: An
integrative approach. Memory, 22: 559–81.
Intraub, H. and Dickinson, C. (2008). False Memory 1/20th of a second later: What the early onset of
boundary extension reveals about perception. Psychological Science, 19(10): 1007–14.
McCarroll, C. J. (2015). Point of view in personal memory: A philosophical investigation. PhD Thesis.
Sydney: Macquarie University.
McCarroll, C. J. (forthcoming). Remembering from the Outside: Personal Memory and the Perspectival Mind.
New York: Oxford University Press.
McCarroll, C. J. and Sutton, J. (2016). Multiperspectival imagery: Sartre and cognitive theory on point of
view in remembering and imagining. In J. Reynolds and R. Sebold (eds), Phenomenology and Science:
Confrontations and Convergences (pp. 181–204). London: Palgrave.
McIsaac, H. K. and Eich, E. (2002). Vantage point in episodic memory. Psychonomic Bulletin and Review,
9(1): 146–50.
125
Mackenzie, C. (2007). Imagination, identity, and self-transformation. In K. Atkins and C. Mackenzie (eds),
Practical Identity and Narrative Agency (pp. 121–45). London: Routledge.
Malcolm, N. (1963). Knowledge and Certainty. Ithaca, NY: Cornell University Press.
Matthen, M. (2010). Is memory preservation? Philosophical Studies, 148(1): 3–14.
Messaris, P. (1994). Visual Literacy: Image, Mind, and Reality. Boulder, CO: Westview Press.
Michaelian, K. (2011). Generative memory. Philosophical Psychology, 24(3): 323–42.
Morris, T. and Spittle, M. (2012). A default hypothesis of the development of internal and external imagery
perspectives. Journal of Mental Imagery, 36(1 and 2): 1–30.
Morris, T., Spittle, M. and Watt, A. P. (2005). Imagery in Sport. Champaign, IL: Human Kinetics.
Nicolas, S., Gounden, Y. and Piolino, P. (2013). Victor and Catherine Henri on earliest recollections.
L’année psychologique/Topics in Cognitive Psychology, 113(3): 349–74.
Nigro, G. and Neisser, U. (1983). Point of view in personal memories. Cognitive Psychology, 15(4): 467–82.
Recanati, F. (2007). Perspectival Thought. A Plea for (Moderate) Relativism. Oxford: Oxford University Press.
Rice, H. J. (2010). Seeing where we’re at: A review of visual perspective and memory retrieval. In
J. H. Mace (ed.), The Act of Remembering: Toward an Understanding of How We Recall the Past
(pp. 228–58). Chichester: Wiley-Blackwell.
Rice, H. J. and Rubin, D. C. (2009). I can see it both ways: First- and third-person visual perspectives at
retrieval. Consciousness and Cognition, 18(4): 877–90.
Robinson, J. A. and Swanson, K. L. (1993). Field and observer modes of remembering. Memory, 1(3): 169–84.
Rosen, M. and Sutton, J. (2013). Self-representation and perspectives in dreams. Philosophy Compass, 8(11):
1041–53.
Rowlands, M. (2009). Memory. In J. Symons and P. Calvo (eds), The Routledge Companion to Philosophy of
Psychology (pp. 336–46). London: Routledge.
Schacter, D. L. and Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering
the past and imagining the future. Philosophical Transactions of the Royal Society B: Biological Sciences, 362:
773–86.
Shore, B. (2008). Spiritual work, memory work: Revival and recollection at Salem camp meeting. Ethos,
36(1): 98–119.
Smith, M. (1997). Imagining from the inside. In R. Allen and M. Smith (eds), Film Theory and Philosophy
Sutton, J. (2003). Truth in memory: The humanities and the cognitive sciences. In I. McCalman and
A. McGrath (eds), Proof and Truth: The Humanist as Expert (pp. 145–63). Canberra: Australian Academy
of the Humanities.
Sutton, J. (2010). Observer perspective and acentred memory: Some puzzles about point of view in personal
memory. Philosophical Studies, 148(1): 27–37.
Sutton, J. (2012). Memory before the game: Switching perspectives in imagining and remembering sport
and movement. Journal of Mental Imagery, 36(1–2): 85–95.
Sutton, J. (2014). Memory perspectives (editorial). Memory Studies, 7(2): 141–5.
Tversky, B. (2011). Visualizing thought. Topics in Cognitive Science, 3(3): 499–535.
Tversky, B. and Hard, B. M. (2009). Embodied and disembodied cognition: Spatial perspective-taking.
Cognition, 110(1): 124–9.
Vendler, Z. (1979). Vicarious experience. Revue de Métaphysique et de Morale, 84(2): 161–73.
Vrij, A., Leal, S., Mann, S., Warmelink, L., Granhag, P. A. and Fisher, R. P. (2010). Drawings as an innovative
and successful lie detection tool. Applied Cognitive Psychology, 24: 587–94.
Walton, K. L. (1990). Mimesis as Make-Believe. On the Foundations of the Representational Arts. Cambridge,
MA: Harvard University Press.
Williams, B. (1973). Imagination and the self. In Problems of the Self (pp. 26–45). Cambridge: Cambridge
University Press.
Windt, J. M. (2015). Dreaming: A Conceptual Framework for Philosophy of Mind and Empirical Research.
Winner, D. (2012). Beautiful game, beautiful mind. ESPN The Magazine. www.espnfc.com.au/england/
story/1071240/beautiful-game-beautiful-mind.
Wollheim, R. (1984). The Thread of Life. New Haven, CT: Yale University Press.
126
10
MEMORY AND IMAGINATION
Felipe De Brigard
In the history of philosophy, the relationship between memory and imagination has been a
matter of debate. This chapter critically surveys the controversy. Section 1 explores some rea-
sons that have led philosophers to assume that memory and imagination are distinct. Section 2
offers a historical overview of the main views concerning the distinction between memory and
imagination. I suggest that much of the philosophical discussion surrounding the nature of this
distinction obfuscates at least three different senses in which memory and imagination could
differ. One sense concerns the difference between mental events that should be considered
memories versus those that should be considered imaginations. A second sense concerns the
nature of the relationship between the mental faculties or systems of memory and imagination.
Finally, a third sense concerns the phenomenology of remembering versus that of imagining.
Section 3 concludes with a brief overview of some important behavioral and neuroscientific
results relevant to this discussion.
1. Should memory and imagination be distinct?

For years, the received view in the philosophy of memory has been that memory and imagination
are distinct. Although most of the discussion has focused on the precise nature of the distinc-
tion, it is worth examining the reasons that have driven philosophers to assume that memory
and imagination must differ. Some philosophers have supported this assumption on conceptual
grounds. Aristotle, for instance, does so by deploying a content-based principle (De Brigard 2014a),
according to which two or more cognitive faculties are different from one another if the inten-
tional contents they purportedly operate upon are distinct. Employing this principle, Aristotle
distinguishes memory from expectation and perception in de Memoria. Briefly, he argues that
memory ought to be different from both expectation and perception because, unlike expecta-
tion, which is about things that haven’t happened yet, and unlike perception, which is about
things that are happening now, memory deals with things that already happened (Sorabji 1972:
449b10–25). Aristotle also points out that, unlike memories, imaginations are typically about
things that did not happen or are false (de Anima 428a12–15). Thus, the fact that memories are
about things that happened and are true, whereas imaginations are about things that did not hap-
pen and are false, leads Aristotle to use the content-based principle to claim that remembering
must be different from imagining.
127
Felipe De Brigard
Unfortunately, the employment of such a content-based principle to buttress the distinction

between memory and imagination is controversial. Aristotle wants us to accept that remember-
ing and expecting, for instance, are distinct because the content of mental states that are about
things that happened in the past are relevantly distinct from the contents of mental states that are
about things that may happen in the future. But what reason do we have to accept the second
clause? Moreover, even if we grant that, somehow, such contents are relevantly distinct, we still
need a reason to believe that a difference in mental content maps onto a distinction in psycho-
logical processing. What could prevent us from making an analogue case for mental contents
about things that are smaller than 6 feet versus things that are bigger than 6 feet? Why should
time be a better psychological wedge than size? Without a principled way to determine, first,
when a certain kind of mental content is relevantly distinct from another and, second, when
such a distinction in content maps onto a distinction in cognitive processing, Aristotle’s strategy
to individuate cognitive faculties risks delivering arbitrary cognitive taxonomies.
A second reason to be suspicious of content-based conceptual strategies to distinguish mem-
ory from imagination comes from the fact that confining the contents of memories solely to
things that happened in the past may leave out clear instances of remembering that are not about
past things. As Munsat (1966) pointed out, we often use the locution “remember” to refer to
events that have not happened yet. Suppose that you leave your office with a friend, and she
invites you over for a drink. “Thank you,” you answer, “I’d like to very much. Oh, wait a
minute, I just remembered that I have to be home because Jerry is coming for dinner.” In what
sense, Munsat wonders:
is this ‘remembering’ of the past? As far as I can see, the only thing ‘past-ish’ about
my suddenly remembering that I have to do something, or be somewhere, is that we
always say “I just remembered.” But what was supposed to be in the past was what I
remember, not the remembering.
(Munsat 1966: 5)
Trying to find something about the moment in which you arranged the dinner with Jerry won’t
do, as not only the content of the memory “I remember inviting Jerry over for dinner” is dif-
ferent from the content of “I remember I have to be home in a few minutes,” but also is not
even necessary for you to remember anything about the moment in which you invited Jerry, or
about the intention of having him over, to suddenly remember that you have to go home. True,
all these could be causal factors leading to your sudden realization that you cannot accept the
invitation, but they needn’t feature as the contents of your memory. Still, to suddenly remember
seems like a genuine case of remembering. Thus, confining all memorial contents to be about
past events would leave out clear instances of remembering.
A second strategy to defend the claim that memory and imagination must differ involves
linguistic reasons. “Remember,” it is said, is a factive verb, meaning that when it is used to
express a relation between a subject, S, and a proposition, p, the corresponding propositional
attitude report “S remembers p” can only be true if p is true. Thus, if one utters “John remem-
bers that the car is out of gas,” the utterance can only be true if, in fact, the car is out of gas.
“Imagine,” on the other hand, is not factive. One can say, truthfully, “John imagines that the
car is out of gas,” even if the car is not out of gas. Arguably, this grammatical difference gives us
reason to believe that memory and imagination are distinct.
However, this factivity constraint can be objected. The first objection pertains linguistic
chauvinism. Let’s grant that the English verb “remember” expresses a relation between a subject
and a proposition referred to by a predicative that-clause. Why should that matter? Currently,
128
Memory and imagination
there are around 6,500 languages spoken in the world (Evans and Levinson 2009). It would be
a miracle if all languages would have a lexicalized verb not only sharing the same semantic field
as the English verb “to remember,” but also taking as complement a sentential clause. Indeed,
although less than 10 percent of today’s languages have been decently documented, we already
find counterexamples. Dalabon, a Gunwinyguan language of Arnhem Land (Australia), simply
has no lexical verb for remembering. To talk about memories, speakers of Dalabon employ dif-
ferent tenses and aspectual transformations on words that are also used to express other mental
processes, such as “realize,” “attend,” and “decide” (Evans 2007). Moreover, some languages
don’t even make distinctions between verbs and direct complements, like Straits Salish, an
endangered language in the American Pacific Northwest, which only contains one major class
of lexical item functioning as predicate (Jelinek 1995). In neither of these languages do their
approximate cognates of “remembers” look relational—or at least not in the same way in which
“remembers” looks relational in English. Arguing in favor of a psychological or metaphysi-
cal difference between memory and imagination on the contingent fact that we speak English
seems, if not unwarranted, at least chauvinistic.
A defender of the linguistic strategy may contend that the evidence comes not from surface
but from deep grammar. Thus, even if a language lacks a lexicalized verb for “remembering,” it
may be possible that remembering-like constructions can conform, at the deep grammar level,
to the canonical form of propositional attitude reports. This strategy is problematic too, for even
at the deep grammar level, the distinction between complement, relative, and adverbial clauses
isn’t always clear-cut. For instance, sometimes determining whether a subordinate clause pattern
conforms to one or another structure may be, more or less, a matter of taste. “I remember when
I used to play” seems to take as a complement an adverbial clause, but for certain purposes it
could be taken as a relative clause with an elided head noun, e.g., “I remember [the days] when
I used to play.” Forcing all construction patterns to look like nominal phrases taking as comple-
ments sentential clauses of the form “S remembers that p” may look like an attempt to make the
data fit the theory rather than the other way around.
Linguistic variability aside, there are other reasons to be suspicious of the factivity constraint.
Although virtually every endorser of the factivity constraint assumes it as obviously true from
the way in which competent speakers allegedly use the word “remembering” (Audi 1998;
Malcolm 1963; Shoemaker 1972), a handful of additional reasons have been offered in support.
One line of argument is to claim that the conjunction of a memory claim with the negation
of its embedded clause is contradictory. According to this line of argument, if someone utters
(a) “I remember I was drinking tequila but I wasn’t drinking tequila,” then she would be con-
tradicting herself. But, as Hazlett (2010) has recently argued, this isn’t really a contradiction. It
is only incoherent in the same way in which Moore’s famous “It is raining outside but I don’t
believe it” is incoherent. For it would be wrong to think that no competent user of the verb
“to remember” can rationally hold (a) true. At most, it is only pragmatically incoherent, and the
incoherence appears solely when we keep fixed the conditions under which the claims before
and after the “but” are evaluated. However, evaluative conditions between claims can shift.
Suppose a tequila snob utters (a) so that the first claim is evaluated as a description of the event,
whereas the second claim is evaluated as a value judgment of what constitutes tequila. His utter-
ance of (a) wouldn’t be contradictory then.
Another argument for the factivity constraint consists in applying Vendler’s criterion to the
verb “to remember.” According to Vendler (1972), one could distinguish between factive and
non-factive verbs in that the former, but not the latter, can be transformed into wh-clauses.
Thus, “to remember” is factive because “Jose remembers that the car was scratched” can take
the form of “Jose remembers where the car was scratched,” “Jose remembers when the car was
129
Felipe De Brigard
scratched,” etc. But what is really the scope of this argument? First, albeit grammatically correct,
it seems to make little psychological sense. It may be that Jose does not really remember where
the scratch was or when it happened. Thus, some cases in which the wh-clauses produced by
Vendlerizing remembering statements may involve information about the remembered event
that need not be part of the expressed intentional content. Should we say that these aren’t
genuine cases of remembering? If we say that they are, then the Vendler criterion isn’t really
that useful. At most, it indicates something interesting about the grammar of the English verb
“to remember” that has no bearing on the truth of its contents. But if we say that they are not
genuine cases of remembering, precisely because there is some information “missing,” then we
are unduly constraining our cases of genuine remembering to cases in which a large amount
of information about a particular event needs to be brought to mind. Many of our memories
wouldn’t then count as genuine memories.
Second, this strategy is hard to apply to memory statements about atemporal events or events
that haven’t occurred. Consider:
(1) I remember that I am going to see you next week.

(2) I remember that the number of planets is 8.
In these cases, the following wh-clauses derived from (1) and (2) sound odd:
(1′) *I remember when I will be seeing you next week.

(1′′) *I remember what I will be seeing you next week.
(2′) *I remember what the number of planets is 8.
(2′′) *I remember when the number of planets is 8.
A possibility is to try to translate these statements into remembering-statements about events

that happened in the past. But, as discussed above, this strategy won’t do. When I remember
that I will be seeing you next week, I needn’t bring to mind anything regarding the way in
which I encoded the information about our future meeting. Likewise, we often remember facts
without recalling when or how we learned them. To say that I remember that the number of
planets is 8 does not imply that I also remember having learned it. I may have forgotten learning
about it while still remembering the fact.
Finally, perhaps the most damaging argument against the factivity constraint is the simple fact
that competent speakers just don’t abide by it when they use the word “remembering” (Hazlett
2010). Talk of false memories, for instance, is nowadays ubiquitous. Most people feel comfort-
able using the word “remembering” when referring to things that didn’t happen, or things that
didn’t happen exactly as remembered. Are people just misapplying the word “remembering”?
Traditionally, this has been the philosophers’ reply, as they dismiss the concern about false
memories by distinguishing ostensive from veridical remembering (Shoemaker 1972). Ostensive
remembering is only “seeming to remember,” whereas veridical remembering is, well, true rem-
embering, i.e., recollecting what was the case. The thought is that when we use the word
“remembering” when referring to events that did not happen, we are speaking loosely, for we
should have used instead the locution “seeming to remember.” But this strategy isn’t warranted.
Notice that “seeming” has, at least, an epistemic and a phenomenological sense (Schwitzgebel
2008). In the epistemic sense, we use “it appears/seems” to indicate hesitation or uncertainty.
But false and distorted memories just don’t come to us in a way that makes us hesitant about
them being memories. Conversely, in the phenomenological sense, we use “it appears/seems”
to indicate the way in which a particular mental content presents itself to consciousness, as when
130
we say, looking at the Müller-Lyer illusion, that one line appears longer than the other. In this
case, we express no hesitation but a mere phenomenological report. The problem is that, when
it comes to the phenomenological sense of “seeming to remember,” distorted and veridical
memories are often indistinguishable. The distinction between seeming to remember and actu-
ally remembering only makes sense from the point of view of epistemology, but this is because
the philosopher has already confined her notion of remembering to veridical memories—
a decision that isn’t grounded in the way competent speakers use the word “remembering.”
To summarize: the claim that memory and imagination must be distinct is often assumed
a priori. The few arguments given in support of this assumption take the distinction to be
grounded either on conceptual or linguistic facts. I offered reasons to be skeptical of both
strategies. Despite it being part of the dogma in philosophy for centuries, the best reasons phi-
losophers have mustered in support of the assumption that memory and imagination must be
distinct are, in fact, controversial.
2. On the distinction between memory

and imagination
Notwithstanding the controversy, it is still legitimate to ask what, if any, is the difference
between memory and imagination. Here, again, philosophers disagree. Unfortunately, the
disagreement is obfuscated by the fact that traditional attempts to clarify the distinction often
confuse at least three non-mutually exclusive senses in which one can ask how memory and
imagination differ. First, we may wonder how a particular memory is different from a particular
imagination. Attempts to distinguish memories from imaginations in terms of mental contents,
mental representations, or even the causal relationship to their intentional objects, can be seen
as speaking to this first way of understanding the question. Second, we may also wonder how
memory, qua cognitive faculty (or system), is different from imagination. Traditional views also
vary here, with some suggesting that memory and imagination are entirely different faculties,
others that they are the same, and some that it may be best to think of them as having common
and distinct processes. Importantly, the answer to this second question is fairly independent of
the answer to the first question, as one can hold, for instance, that memories and imaginations
are different mental events even though they are processed by the same faculty, by faculties that
share common processes, or by entirely independent systems.
Finally, we may also wonder how memory, qua psychological experience, differs from imagi-
nation. Again, the answer to the question about the phenomenological differences between
remembering and imagining is, to some extent, independent of the answers to the first two
questions. One could, for instance, suggest that memories are experienced with more vivac-
ity than imaginations, and argue that such phenomenological difference is due to a difference
in the corresponding mental representations, or to the way in which such representations are
processed. As mentioned, traditional philosophical views on the distinction between memory
and imagination are often unclear as to what sort of difference they are speaking to. In what
follows, I review some prominent philosophical views on the difference between memory and
imagination, while trying to elucidate the precise sense(s) of the question each view addresses.
Philosophical discussions about the nature of memory and imagination can arguably be
traced back to Plato (MacKisack et al. 2016; Chapter 30, this volume), not so much because he
offered a psychological theory of either of them, but rather because he helped to introduce the
terms of the debate. Specifically, Plato’s metaphorical allusions to both memory and imagina-
tion consistently employed representation-like terms—such as “impressions” (Theaetetus: 191d),
“copies” (Republic: 509e; Sophist: 266b), and “images” (Philebus: 39b)—also used to refer to
131
Felipe De Brigard
non-mental representations (e.g., paintings, portraits). While Aristotle inherits Plato’s

terminology—particularly the locutions phantasma and eikon—he didn’t treat them on par.
Instead, he argued for a distinction between memories and imaginations qua mental particulars.
After suggesting that memory must be different from perception and expectation (see Section 1
of this chapter), Aristotle wonders (de Memoria: 450a25) how something that is not present—the
remembered—could be recalled by something that is present—the remembering. Prima facie, his
answer resembles Plato’s: when we remember, we perceive an image of the past. However, this
solution raises two problems. First, if remembering is to perceive something present, viz. the cur-
rent image, then we must say that what we remember is present. By Aristotle’s own assumption,
this cannot be: memory is about the past, not the present. But if what we remember is the past,
and remembering is a kind of perception, how can we perceive something that is not present?
This looks like an aporia. The second problem can be illustrated with a thought experiment.
Suppose that Edgar witnesses a certain object, say, Mount Fuji, and later on recalls this moun-
tain by perceiving a mental image resembling it. If remembering consists in perceiving a mental
image resembling a past object, then this would be a case of remembering. But now suppose that
Ava, who has never seen Mount Fuji, is asked to imagine a snowy mountain. With magnificent
detail, Ava manages to conjure up a very elaborate mental image that happens to be identical to
Edgar’s. As such, both Edgar and Ava’s perceived mental images resemble the same object to an
equal degree. Yet, intuitively, only Edgar’s constitutes a memory; Ava’s is an imagination. But if
all there is to remembering is to bring to mind and presently perceive a mental image that resem-
bles a past event, then we would have to say that both are memories. So, what gives?
Aristotle’s solution is to differentiate two kinds of mental images: phantasma, which is used as
a generic term for mental image (de Anima: 631a16), and eikon, which is a mental image similar
to and causally derived from the object it represents (de Memoria: 450a27–b11). Remembering,
then, consists in bringing back to mind an eikon. If the currently perceived image merely resem-
bles but it is not causally derived from a past object, then it is not a memory: it is merely an
imagination (451a2; a8–12). This causal criterion constitutes Aristotle’s strategy to distinguish
memories, qua mental particulars, from imaginations. Critically, the causal criterion is external,
in that the individual need not remember that the currently perceived mental image was actu-
ally caused by a past event, as it is the case with some of the cases introduced by Martin and
Deutscher (1966; Chapter 5, this volume). Edgar could have forgotten seeing Mount Fuji and,
later on, being asked to imagine a snowy mountain; if the mental image Edgar conjures up both
resembles and is causally derived from the previously witnessed mountain, then it constitutes an
act of remembering even if he thinks he’s just imagining.
Aristotle also employs the aforementioned content-based principle to distinguish the faculties
of memory and imagination. For him, memories represent both past objects and time lapsed (de
Anima: 450b27), so he reasons that memory must be related to the part of the mind with which
we perceive time, viz. common sense. This is because Aristotle takes time-perception to be
essentially change-perception, which is possible only if one can contrast two or more images of
that which changes. Since such contrasts can be multimodal, they occur in the common sense.
Now, the contrasted images—phantasmata—are the objects upon which phantasy, or imagination,
operates. Therefore, memory is not only part of but also depends on imagination.
Years later, Aquinas developed a more nuanced model of the relationship between memory
and imagination (MacKisack et al. 2016; Chapter 37, this volume). First, Aquinas uses “phan-
tasia,” not as a synonym for “imagination,” but as a general term for the mental contents that
result from common sense funneling information from the “outer senses.” He then takes imagi-
nation to be one of three faculties of the “inner senses,” all of which process phantasmata as
132
their contents. Specifically, Aquinas considers imagination (vis imaginativa) as the “storehouse”
(thesaurus) of sensory forms. Memory (vis memorativa), the second inner-sense faculty, links the
information from imagination to a particular individual object or event at a particular time; that
is, it stores the spatiotemporal context of the phantasmata kept in imagination. Finally, cogni-
tion (vis cogitativa), abstracts away from the pure sensory and spatiotemporal information of the
mental image in order to apprehend “primary substances,” akin to “categories,” or “concepts”
(Lisska 2016). The relationship between memory and imagination is, therefore, hierarchical:
memory depends on imagination—which also has a retentive function—but goes beyond it by
its capacity to store the spatiotemporal information associated with the acquisition of a particular
mental image. Additionally, imagination has a creative or “compositive” ability, which allows it
to piece together images that were not jointly perceived. Memory may remember these mon-
grel phantasmata, but they were imagination’s work.
Aquinas’ account contrasts with the neo-Platonic view of Augustine (Chapter 35, this volume).
For the latter takes memory, not imagination, to be fundamental. Specifically, while he accepts
that memory stores images from the senses (Confessions X: 8), he believes that memory also
contains non-imaginistic information, such as facts (X: 9), feelings, bodily reactions (X: 14), and
abstract cogitations (X: 12–13). Unlike Aristotle and Aquinas, Augustine does not think that
memory and imagination process the same contents, i.e., mental images. Instead, he endows
imagination with the lowlier task of either filling-in or putting together memories in novel ways
(Breyfogle 1996). Thus, imagination can play positive roles in our mental life, such as using past
memories to generate predictions and expectations (XI), but also negative, as it can also be the
source of error and confusion (I: 16; De Vera Religione: 64). We see, then, a contrast between
an Aristotelian and a neo-Platonic view of the distinction between memory and imagination.
For the former, memory (qua faculty) depends on imagination, and memories (qua mental
particulars) are distinguished from imaginations, not by not being mental images, but by being
processed by distinct faculties. For the latter, imagination (qua faculty) depends on memory, as
it cannot operate unless there are already mental images stored, and it also differs from it in that
the set of all memories (qua mental particulars) is larger than the set of images, restricting thus
its processing domain.
Descartes is harder to fit within this dialectic, partly because he is more interested in squar-
ing the distinction between memory and imagination with biological and phenomenological
observations than with abstract principles. For instance, Descartes suggests that the brain activity
responsible for perceptions is similar to that underlying both memory and imagination (Clarke
2003). However, in the Passions, he states that there is a difference between perceptions that
are caused by real events, such as sensed and remembered ones, and those that are triggered by
merely imaginary ones: “The impressions that come into the brain through the nerves are usually
more lively and more distinct than those stimulated in the brain by the spirits” (xi: 348). Thus,
unlike Aristotle’s causal criterion, Descartes lively criterion is internalist: it pertains to the experi-
ence of imagining and remembering, rather than to the causal origin of the experienced mental
content. True, Descartes—as many of his successors—thought that increased vivacity correlated
with actually experienced events, but he still placed emphasis on the phenomenology of the
experience as a criterion for distinguishing particular memories from imaginations. Although
less developed, a second internalist criterion is also evident in Descartes’ work: coherence. At
the end of his Meditations, he briefly suggests that experiences of real events cohere better with
our beliefs than experiences of merely imagined events (vii: 89–90). Thus, in Descartes, we find
a view in which memory and imagination share a common functional substrate, and whose sole
differences are to be found internally, in the way those contents are experienced.
133
Felipe De Brigard
At around the same time, Hobbes was arguing for a stronger claim: “That Imagination and
Memory are but one thing” (Hobbes 1651: 2). Hobbes’s view of cognitive contents is inspired
by the notions of motion and inertia. Roughly, his idea is that every object is in constant motion
unless it is hindered by something else, such as our senses. When our eyes, say, contemplate an
object, its motion is hindered and as an effect, it leaves an image in our senses. As long as the
object is perceived, its effect endures, and the image is clear. But if the item is removed from our
presence, or we shut our eyes, its effect begins to decay, and thus the image becomes more and
more obscure. Imaginations and memories are, therefore, faded mental images, whose perceived
obscurity is due to their objects no longer exerting pressure upon our senses. This view implies,
not only that memory and imagination are identical qua faculties, but also qua particular mental
contents: both are faded images of previously perceived objects. How can Hobbes reconcile this
view with the obvious differences in our experiences of remembering and imagining? Here,
I suggest, Hobbes employs another internalist criterion, based on the purpose to which these
faculties are put to use. Remembering, he suggests, is bringing to mind “simple imaginations”:
images as they were perceived, unchanged, by the senses. Imagining, on the other hand, is
bringing to mind “compound imaginations”: images composed of simple imaginations pieced
together—e.g., the compound image of a centaur, which combines the simple images (memories)
of a horse and a man.
Since then, we see modern empiricist and rationalist philosophers offering internalist criteria
for distinguishing memories from imaginations. Spinoza, for instance, considers both memories
and imagination to be ideas associated with previous modifications of the body (Ethics, XVIII).
However, he suggests that memories, unlike imaginations, “are accompanied with the thought
to determine the duration of the sensation” (TIE: 83). Memories, accordingly, are connected to
the sensation by the accompanying temporal thought. We mistake imaginations for memories
when the unamended intellect erroneously assents to the unconnected sensation as if it was con-
nected. Surprisingly, a similar internalist criterion is offered by his contemporary John Locke.
In an oft-quoted passage of the second edition of his Essay, Locke suggests that memory is the
mind’s power “to revive perceptions which it has once had, with this additional perception
annexed to them, that it has had them before” (X: 2).
Distinguishing memories from imaginations by suggesting that the former, as opposed to
the latter, are accompanied by an added mental content at retrieval, is a popular strategy (De
Brigard 2014b) even to this day (e.g., Fernández 2015). I’ve already mentioned how Spinoza
and Locke thought of this added state as a kind of affective state of duration or “pastness.”
Others—e.g., Leibniz (1714) and Kant (1787)—thought of the accompanying mental state as a
second-order representation or apperception the content of which is that the currently entertained
image is the same as one entertained before. Finally, some—e.g., James (1890) and Russell
(1921)—take this additional content to be a belief. James, for instance, suggests that memories,
unlike imaginations, are retrieved “with the additional consciousness that we have thought or
experienced [them] before” (1890: 648). Similarly, Russell claimed that while memory-images
and imagination-images have identical intrinsic qualities, only the former
are accompanied by a feeling of belief which may be expressed in the words ‘this
happened’. The mere occurrence of images, without this feeling of belief, constitutes
imagination; it is the element of belief that is the distinctive thing in memory.
(Russell 1921: 14)
However, no account of the difference between imagination and memory is as influential

as Hume’s two internalist criteria (Chapter 39, this volume). The first criterion is vivacity:
134
“The ideas of the memory are much more lively and strong than those of the imagination” (I: I, 3).
The second criterion appeals to structural preservation between the original impression and
the retrieved one: “Imagination is not restrain’d to the same order and form with the original
impressions; while the memory is in a manner ty’d down in that respect, without any power of
variation” (ibid.). Hume, however, was aware of the obvious difficulties of the second criterion:
one simply cannot bring back to mind the original impression to compare it with the retrieved
one (a point Reid (1785) emphasizes). Thus, he placed more importance to vivacity as an inter-
nalist criterion for distinguishing memories from imagination (I: III, 5).
Nevertheless, Hume’s vivacity criterion has been widely criticized (MacNabb 1966; Passmore
1968). Influenced by Wittgenstein’s anti-imaginistic stance on psychology (e.g., 1963, I: 905),
Ryle famously argued that Hume’s use of the term “vivacity” was mistaken in one of two ways.
On one hand, if “vivacity” means “lively,” then it makes sense to say that one doll looks more
“lively” than another one, in the sense of seeming to be more life-like. But this notion, claims
Ryle, makes no sense when it comes to impressions. On the other hand, if “vivacity” means
“intense,” then Hume is wrongly assuming that what can be said of sensations can also be said of
mental images, for while a sensation can be said to be stronger or weaker than another, the same
is not the case for images: “While I fancy I am hearing a very loud noise, I am not really hear-
ing either a loud or a faint noise; I am not having a mild auditory sensation” (Ryle 1949: 270).
Shortly after, R.F. Holland—echoing a concern previously expressed by others (Kemp-Smith
1941; Russell 1921)—states what for many is the Achilles heel of Hume’s vivacity criterion:
“The suggestion that the haziest of recollections must be somehow clearer and more vivid than
the most powerful products of a lively imagination seems implausible, if not senseless” (Holland
1954). Holland articulates two versions of this objection. The first version pertains individual
memories and imaginations. According to this version, Hume’s vivacity criterion is to be under-
stood as stating that all memories are more vivid than imaginations. That is, for every particular
memory, m, and every particular imagination, i, m is more vivid than i. However, it is a fact
that, sometimes, a particular i is more vivid than a particular m. Therefore, it is not the case that
all m are more vivid than all i.
Unfortunately, Hume had already anticipated this objection, when he claimed that
frequently a memory may become weak and feeble, and that
by losing its force and vivacity, may degenerate to such a degree, as to be taken for an
idea of the imagination; so on the other hand an idea of the imagination may acquire
such a force and vivacity, as to pass for an idea of the memory, and counterfeit its
effects on the belief and judgment.
(I: III, 5)
Still, Holland offers a second version of the argument that, allegedly, may prove fatal. This ver-
sion takes Hume’s vivacity criterion as suggesting that, on average, memories are more vivid
than imagination. However, Holland suggest that, in the case of memories and imaginations,
the notion of “average” makes no sense, as there is no measure according to which the respec-
tive ranges of vivacity can be compared, as opposed to, say, things made of lead, which are on
average heavier than things made of cork—even if some things made of cork are heavier than
some things made of lead—because weight is a common measure that allows us to compare
them. Vivacity does not seem to be a good criterion to distinguish memories from imaginations.
Urmson (1967) suggested an ingenious strategy to safeguard Hume’s vivacity criterion.
According to Urmson, there is an ambiguity in the way “imagining” and “remembering” are
used in this discussion. On one hand, imagining and remembering can differ in terms of their
135
Felipe De Brigard
success criteria. One may successfully imagine something when one attempts to invent some-
thing freely and does it. Conversely, one may successfully remember when one attempts to
bring to mind a past experience and does it. In this sense, Urmson suggests, Hume was right in
taking the internal character of the mental activity as sufficient criteria to determine whether
one is imagining or remembering. However, if “imagining” and “remembering” are under-
stood in terms of accuracy, then internal criteria won’t work, as we can’t simply tell, from the
subjective experience alone, whether what we take to be a memory is or is not an accurate
representation of what happened, just as we can’t tell whether or not what we are imagining
does, in fact, correspond to an actual past event. E.J. Furlong (1970), alas, is not convinced, for
he believes that Urmson is making Hume draw a distinction he did not have in mind: that of
attempting to imagine versus attempting to remember. Furlong correctly points out that such
was not Hume’s intention and, therefore, that Urmson’s isn’t an adequate defense of Hume’s
internal criterion of vivacity.
This historical review is both opinionated and selective. Still, it shows how philosophers
bestride three ways to ask about the nature of the distinction between memory and imagination,
viz. to mean either (1) mental particulars, (2) cognitive faculties, or (3) psychological experiences.
Regarding (1), philosophers’ views fall into two groups: externalists, for whom the difference
between memories and imaginations depends on something extrinsic to their contents—
e.g., an appropriate causal connection to their objects (Aristotle)—and internalists, for whom the
difference is to be found in the content, either because of an intrinsic difference (e.g., Aquinas’
memories are spatiotemporal) or because, at retrieval, otherwise indistinguishable contents are
recovered in conjunction with additional mental states, be them affective (e.g., Spinoza, Locke,
maybe Hume), cognitive (e.g., James, Russell), or apperceptive/meta-cognitive (e.g., Leibniz,
Kant). As for (2), we see philosophers endorsing either a hierarchical relationship between them
(Aristotle), a complete dissociation (Augustine, Aquinas), a full-fledged identification (Hobbes),
or an intermediate stance in which imagination and memory have distinct and common pro-
cesses (Hume). Finally, regarding (3), philosophers have used criteria such as liveliness, coherence
(Descartes), clarity (Hobbes), vivacity (Hume) and familiarity (Russell, James), to characterize the
differences in the phenomenology of remembering and imagining.
3. Can science distinguish memory and imagination?

Recently, findings from research in the cognitive psychology and neuroscience of memory
and imagination have been brought to bear on the philosophical discussion about the nature
of their distinction. One relevant line of research pertains work conducted by psychologist
Marcia Johnson and collaborators. In the 1970s, there were a couple of important findings
pertaining people’s ability to remember information that they either came up with or merely
perceived. The first finding is reported in an influential paper by Slamecka and Graf (1978),
where they show a memory advantage for self-generated versus other-generated information.
Importantly, they also report that this manipulation makes people really good at determining
the source (i.e., self or other) of the remembered information. The second set of results came
from the work of Larry Jacoby (1978), who showed that self-generated solutions to a prob-
lem are better retained than solutions that are simply repeated. Repetition, however, tends to
increase perceptual fluency which in turn tends to increase false alarms to repeated relative to
non-repeated information at retrieval (i.e., the fluency effect; Jacoby and Dallas 1981). These,
and similar observations, lead Johnson to propose the source monitoring framework (SMF:
Johnson et al. 1993; previously “reality monitoring framework”: Johnson and Raye 1981).
The SMF starts with the assumption that the phenomenological features of every encoded
136
experience are multidimensional: some features are sensory, some are related to the special
arrangement of experienced items, some are emotional or affective, and some pertain to the
cognitive operations we are engaged in while experiencing the event. At retrieval, we rely on
the relative saliency of these features to bias our judgment as to whether the recovered mental
content is from an event that was previously perceived or previously imagined. Specifically, the
sensory and spatial features of the retrieved contents of previously perceived events tend to be
experienced more clearly and vividly than the phenomenological characteristics associated with
the cognitive operations we were engaged in during encoding, which are more salient when
remembering previously imagined events. Accordingly, discriminating previously experienced
from imagined events is a meta-cognitive decision-making process that depends upon the way
mental contents are experienced at retrieval. Moreover, the SMF can also help to explain many
well-documented memory distortion effects, where disruptions of source monitoring processes
during encoding reliably increase false recognition (for a review, see Johnson 1997).
In addition to making the distinction between memories and imaginations qua mental par-
ticulars a matter of meta-cognitive operations on the phenomenological characteristics of the
retrieved contents, the SMF also suggests that memory and imagination, qua cognitive systems,
share common mechanisms. Further support for this claim comes from a second line of research.
For most of the twentieth century, exploring imagination deficits in individuals with amnesia
was rare (but see Talland 1965). Perhaps motivated by the observation that individuals with
medial temporal lobe (MTL) damage allegedly had only declarative memory impairments, the
standard model took memory to be a relatively independent system (Squire 1982). But this view
started to change, partly with Tulving’s observation that amnesic patient K.C. had difficulty
imagining possible personal future events, which prompted Tulving to suggest that remember-
ing one’s personal past and imagining one’s possible personal future may be two processes of a
single system for mental time travel (Tulving 1985). This hypothesis began to receive stronger
support in the early 2000s, as a number of neuropsychological (Klein et al. 2002), developmental
(Atance and O’Neill 2001), behavioral (D’Argembeau and van der Linden 2004), and neuro-
imaging (Okuda et al. 2003) studies further suggested that episodic memory and future thinking
share a common mechanism. Since, the number of studies corroborating these observations
have risen steeply (Schacter et al. 2012; De Brigard and Gessell 2016), strengthening the view
that remembering the past and imagining the future engages a common core brain network
(Schacter et al. 2007; Spreng et al. 2009; Spreng and Grady 2010).
Some philosophers, who tend to advocate constructivist theories of remembering (Michaelian
2011; De Brigard 2011, 2012), have taken note of these recent findings and have suggested—a
bit in the spirit of Hobbes and Hume—that just as remembering consists in piecing together
fragments of previous experiences into mental simulations of past events, imagining possible
future events also involves piecing together bits of previous experiences. As such, these con-
structivist theories of memory align with the view that, understood as particular mental contents,
memories—or, at least, episodic memories—and imaginations—or, at least, mental simulations
about possible future events—may not be intrinsically distinct, and that the difference in the
phenomenology of remembering the past and imagining possible hypothetical scenarios could
be accounted for by the meta-cognitive resources afforded by the SMF (Michaelian 2016). In
addition, the fact that episodic memory and future thinking require the same reconstructive
mechanisms, not only helps to account for the overlap in brain structures involved in remem-
bering the past and imagining the future (Addis et al. 2007), but also buttresses the view that
memory and imagination, qua cognitive faculties, may not be independent after all. Michaelian
(2016), for instance, defends the view that remembering is a form of imagining, and that episodic
memory is a process of a larger system that supports mental time travel (Chapter 18, this volume).
137
Felipe De Brigard
By contrast, De Brigard (2014a), while agreeing that remembering should be seen as a process of
a larger system for imaginative simulations, holds that whether the simulations are about the past
or the future may not be as essential as the fact that they constitute hypothetical scenarios whose
mental simulation unfolds over time (De Brigard and Gessell 2016).
Taken together, the scientific evidence seems to overwhelmingly support the view that
memory and imagination are profoundly intertwined. Understanding the precise ways in which
they interact, the conditions under which they come apart, and the different constraints they
impose upon one another, constitute exciting open questions for future research in the philosophy
and the science of memory and imagination.
Acknowledgment
Thanks to Bryce Gessell, Greg Stewart and Kirk Michaelian for their comments on a previous draft.
Further reading
Bernecker, S. (2009). Memory: A philosophical study. Oxford: Oxford University Press.
Markman, K.D., Klein, W.M.P. and Suhr, J.A. (2008). Handbook of imagination and mental simulation. New
York: Psychology Press.
Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge
MA: MIT Press.
Michaelian, K., Klein, S.B. and Szpunar, K.K. (2016). Seeing the future: Theoretical perspectives on future-
oriented mental time travel. New York: Oxford University Press.
Warnock, M. (1989). Memory. London: Faber & Faber.
References
Addis, D.R., Wong, A.T. and Schacter, D.L. (2007). Remembering the past and imagining the future:
Common and distinct neural substrates during event construction and elaboration. Neuropsychologia.
45: 1363–77.
Atance, C.M. and O’Neill, D.K. (2001). Episodic future thinking. Trends in Cognitive Sciences. 5(12):
533–9.
Audi, R. (1998). Epistemology. London: Routledge.
Augustine (2008). The works of Saint Augustine: A translation for the 21st century (Vol. 25). Hyde Park, NY:
New City Press.
Breyfogle, T. (1996). Memory and Imagination in Augustine’s Confessions. In: Breyfogle, T. (ed.), Literary
imagination, ancient and modern: Essays in honor of David Grene. Chicago, IL: Chicago University Press.
Clarke, D. (2003). Descartes’ theory of mind. Oxford: Oxford University Press.
D’Argembeau, A. and van der Linden, M. (2004). Phenomenal characteristics associated with projecting
oneself back into the past and forward into the future: Influence of valence and temporal distance.
Consciousness & Cognition. 13: 844–58.
De Brigard, F. (2011). Reconstructing memory. PhD Dissertation. University of North Carolina, Chapel
Hill.
De Brigard, F. (2012). Predictive memory and the surprising gap. Commentary on Andy Clark’s “Whatever
Next? Predictive Brains, Situated Agents and the Future of Cognitive Science.” Frontiers in Psychology.
3: 420.
De Brigard, F. (2014a). Is memory for remembering? Recollection as a form episodic hypothetical thinking.
Synthese. 191(2): 155–85.
De Brigard, F. (2014b). The nature of memory traces. Philosophy Compass. 9(6): 402–14.
De Brigard, F. and Gessell, B.S. (2016). Time is not of the essence: Understanding the neural correlates of
mental time travel. In: Klein, S.B., Michaelian, K. and Szpunar, K.K. (eds.), Seeing the future: Theoretical
perspectives on future-oriented mental time travel. New York: Oxford University Press.
138
Descartes, R. (1991). The philosophical writings of Descartes, Vol. III: Correspondence, J. Cottingham,
R. Stoothoff, D. Murdoch and A. Kenny (trans.). Cambridge: Cambridge University Press.
Evans, N. (2007). Standing up your mind: Remembering in Dalabon. In: Amberber, M. (ed.), The language
of memory in a cross-linguistic perspective. Amsterdam: John Benjamins.
Evans, N. and Levinson, S.C. (2009). The myth of language universal: Language diversity and its importance
for cognitive science. Behavioral and Brain Sciences. 32(5): 429–48.
Fernández, J. (2015). Epistemic generation in memory. Philosophy and Phenomenological Research. 92(2):
620–44.
Furlong, E.J. (1970). Mr. Urmson on memory and imagination. Mind. 79(313): 137–8.
Hazlett, A. (2010). The myth of factive verbs. Philosophy and Phenomenological Research. 80(3): 497–522.
Hobbes, T. (1651/1994). Leviathan, E. Curley (ed.), with selected variants from the Latin edition of 1668.
Indianapolis, IN: Hackett.
Holland, R.F. (1954). The empiricist theory of memory. Mind. 63(252): 464–86.
Hume, D. (1978). A treatise of human nature. Oxford: Oxford University Press.
Jacoby, L.L. (1978). On interpreting the effects of repetition: Solving a problem versus remembering a
solution. Journal of Verbal Learning and Verbal Behavior. 17(6): 649–67.
Jacoby, L.L. and Dallas, M. (1981). On the relationship between autobiographical memory and perceptual
learning. Journal of Experimental Psychology: General. 110(3): 306.
James, W. (1890). The principles of psychology. New York: Henry Holt & Co.
Jelinek, E. (1995). Quantification in Straits Salish. In: Bach, E., Jelinek, E., Kratzer, A. and Partee, B. (eds.),
Quantification in natural languages. Norwell, MA: Kluwer.
Johnson, M.K. (1997). Source monitoring and memory distortion. Philosophical Transactions of the Royal
Society of London B. 352: 1733–45.
Johnson, M.K., Hashtroudi, S. and Lindsay, D.S. (1993). Source monitoring. Psychological Bulletin.
114(1): 3.
Johnson, M.K. and Raye, C.L. (1981). Reality monitoring. Psychological Review. 88(1): 67.
Kant, I. (1787/1997). Critique of pure reason. P. Guyer and A. Wood (trans. and ed.). Cambridge: Cambridge
University Press.
Kemp Smith, N. (1941). The philosophy of David Hume: A critical study of its origins and central doctrines.
London: Palgrave Macmillan.
Klein, S.B., Loftus, J. and Kihlstrom, J.F. (2002). Memory and temporal experience: The effects of episodic
memory loss on an amnesic patient’s ability to remember the past and imagine the future. Social Cognition.
20: 353–79.
Leibniz, G.W. (1714/1989). Philosophical writings. Indianapolis, IN: Hackett.
Lisska, A.J. (2016). Aquinas’s theory of perception: An analytic reconstruction. Oxford: Oxford University Press.
Locke, J. (1979). An essay concerning human understanding. New York: Oxford University Press.
MacKisack, M., Aldworth, S., Macpherson, F., Onians, J., Winlove, C. and Zeman, A. (2016). On picturing
a candle: The prehistory of imagery science. Frontiers in Psychology. 7: 1181–96.
MacNabb, D.G.C. (1966). David Hume: His theory of knowledge and morality. Oxford: Blackwell.
Malcolm, N. (1963). Knowledge and certainty. Ithaca, NY: Cornell University Press.
Martin, C.B. and Deutscher, M. (1966). Remembering. Philosophical Review. 75: 161–96.
Michaelian. K. (2011). Generative memory. Philosophical Psychology. 24(3): 323–42.
MA: MIT Press.
Munsat, S. (1966). The concept of memory. New York: Random House.
Okuda, J., Fujii, T., Ohtake, H., Tsukiura, T., Tanji, K., Suzuki, K., et al. (2003). Thinking of the future
and the past: The roles of the frontal pole and the medial temporal lobes. Neuroimage. 19: 1369–80.
Passmore, J. (1968). Hume’s intentions. London: Gerald Duckworth & Co Ltd.
Plato (1997). Complete works. Indianapolis, IN: Hackett.
Reid, T. (1785/1849). Essays on the intellectual powers of man. Edinburgh: McLachlan, Stewart, & Co.
Russell, B. (1921). The analysis of mind. London: George Allen and Unwin.
Ryle, G. (1949). The concept of mind. Oxford: Oxford University Press.
Schacter, D.L., Addis, D.R. and Buckner, R.L. (2007). The prospective brain: Remembering the past to
imagine the future. Nature Reviews Neuroscience. 8: 657–61.
Schacter, D.L., Addis, D.R., Hassabis, D., Martin, V.C., Spreng, R.N. and Szpunar, K.K. (2012). The
future of memory: Remembering, imagining, and the brain. Neuron. 76(4): 677–94.
Schwitzgebel, E. (2008). The unreliability of naive introspection. Philosophical Review. 117(2): 245–73.
139
Felipe De Brigard
Shoemaker, S. (1972). Memory. In: P. Edwards (ed.), Encyclopedia of philosophy. New York: Macmillan,
vol. V, 265–74.
Slamecka, N.J. and Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experi-
mental Psychology: Human Learning and Memory. 4(6): 592.
Sorabji, R. (1972). Aristotle on memory. London: Duckworth.
Spinoza, B. (1985). The collected works of Spinoza. Vol. I. E.M. Curley (ed.). Princeton, NJ: Princeton
University Press.
Spreng, R.N. and Grady, C.L. (2010). Patterns of brain activity supporting autobiographical memory,
prospection, and theory of mind, and their relationship to the default mode network. Journal of Cognitive
Neuroscience. 22(6): 1112–23.
Spreng, R.N., Mar, R.A. and Kim, A.S. (2009). The common neural basis of autobiographical memory,
prospection, navigation, theory of mind, and the default mode: A quantitative meta-analysis. Journal of
Cognitive Neuroscience. 21(3): 489–510.
Squire, L.R. (1982). The neuropsychology of human memory. Annual Review of Neuroscience. 5(1): 241–73.
Talland, G.A. (1965). Deranged memory. New York: Academic Press.
Tulving, E. (1985). Memory and consciousness. Canadian Psychology. 26(1): 1–12.
Urmson, J.O. (1967). Memory and imagination. Mind. 76(301): 83–91.
Vendler, Z. (1972). Res cogitans: An essay in rational psychology. Ithaca, NY: Cornell University Press.
Wittgenstein, L. (1953). Philosophical investigations. New York: Blackwell.
140
11
MEMORY IMAGES
Elizabeth Irvine
1. Introduction
The aim of this chapter is to assess arguments, based on research into visual short-term memory,
that the contents of consciousness are more rich and detailed than we can evidence in reports.
Several authors claim that the rich and possibly non-conceptual contents of iconic memory
contribute to the contents of consciousness, but much of this is not processed into working
memory so is soon ‘forgotten’ (Block 2011, 2007; Dretske 2007, 2006, 2004; Fodor 2008; Tye
2009, 2006). This short-lived, non-attended and unreported conscious content is argued to exist
on a ‘phenomenal’ (Block) or ‘object’ (Dretske) level of awareness. Content that is attended and
reported on is cognitively accessed (Block), or experienced on a ‘fact’ level (Dretske) as well.
This reflects Block’s well-known distinction between phenomenal and access consciousness.
Phenomenal consciousness refers to content that is experienced, and access consciousness refers
to whatever content is made available to consumer systems, including those systems that gener-
ate (phenomenological) reports. Block uses evidence about short-term visual memory to lend
support to the idea that phenomenal consciousness can exist without access: we can experience
things we cannot act or report on. This particular claim is situated within a biological account
of consciousness, where converging lines of empirical evidence are used to argue that some
contents are present in consciousness, even if the subjects do not (perhaps cannot) report on that
content specifically (Block 2011, 2007; Lamme 2006).
This claim provides a challenge to currently popular Global Workspace theories of con-
sciousness (Dehaene et al. 2011), which suggest that conscious content is limited to the contents
of working memory. The basic idea behind these accounts is that the contents of conscious-
ness are fairly limited at any one time, but change as shifts of attention populate working
memory with new content. This position in turn often stems from a mix of functionalist and
operationalist perspectives on consciousness (e.g. Cohen and Dennett 2011). If the contents of
consciousness must be accessible to the subject (and so to experimenters via report) to serve any
function, then if some content appears to be currently unaccessed or unreportable, then there is
no reason to think of it is as being part of the contents of consciousness. This is a straightforward
denial of the idea that phenomenal consciousness can exist without accessibility. So, since the
contents of working memory are those contents that are currently accessed and reportable, this
141
Elizabeth Irvine
means that the contents of consciousness are tied to the contents of working memory (non-
overflow accounts).
To understand this debate, this chapter focuses on evaluating the two types of short-term
visual memory usually invoked to support claims about rich/overflowing phenomenal content.
The first is iconic memory, which lasts less than a second, stores reasonably large amounts of
visual detail, and components of which support the existence of visual after-images (the ‘memory
images’ of the title). The second is termed ‘fragile visual short-term memory’, and is somewhere
between iconic memory and working memory in both duration and capacity. The contents of
these types of memory are richer than that of working memory, and are argued to form part of
the contents of phenomenal consciousness, so providing short-lived but rich visual experiences.
Several broad themes emerge from discussions on how to interpret this experimental work.
One recurring issue is how to conceive of the relationship between information processing, the
contents of memory, subjective reports, and phenomenology. If there is a ‘mis-match’ between
the format of processed or stored information, and what seems to be evidenced in phenom-
enology or in subjective reports, then it is possible to challenge simple claims that this content
is present in consciousness. Rich/overflow and sparse/non-overflow accounts treat these mis-
match relationships very differently, which in turn impacts on their plausibility.
In addition, later sections raise methodological and conceptual questions about how to iden-
tify the capacity and contents of working memory. This potentially challenges any strong claims
(overflow and non-overflow) about the relationship between working memory and conscious-
ness. This is because if we are fundamentally unsure about what is in working memory, and what
format it takes (and how attention comes into it), then claims about the relationships between
conscious content, working memory and attention become difficult to assess, both philosophically
and empirically.
Arguments that iconic memory contributes content to phenomenal consciousness are
reviewed first, followed by two alternative explanations of the phenomenon, and a discussion
of the ‘mis-match’ problem. This is followed by a review of newer claims about rich conscious
content based on fragile visual short-term memory, and potential consequences for both sides of
the debate, given developing concerns about the nature of working memory. The conclusion
sums up. A brief note: in these debates the devil is in the experimental detail. The following
sections are therefore quite technical in places, but less empirically inclined readers can skim or
skip as required.
2. Iconic memory and overflowing phenomenology

The usual form of short-term visual memory that is used to support claims about the rich/
overflowing content of visual experience is ‘iconic memory’, illustrated via Sperling’s (1960)
experimental paradigm (see Figure 11.1).
The paradigm is fairly straightforward: subjects are shown a display of letters (usually in a 3 by
4 grid) for a short time (15–500 ms), followed by a variable delay. In the ‘full report’ condition,
subjects are asked to report as many letters from the display as they can. They can typically cor-
rectly report 4.5 out of 12 letters, which is roughly what you’d expect based on the capacity of
working memory. In the ‘partial report’ condition, a cue is shown to subjects just after the delay,
either in the form of a visual arrow, or a high/medium/low-pitched audio tone, meant to orient
attention to a particular row in the grid. In this condition, subjects report as many letters as they
can from the cued row, and on average report 3.03 letters from a 4-letter row.
The important feature of the partial report condition is that although subjects can only report
3 letters at a time, they are able to report this many letters from any of the rows that might be
142
Memory images
Figure 11.1 Sperling’s ‘partial report’ paradigm.
cued. In order to support this level of performance, information about 9.01 letters from the
12-letter display must therefore be available to subjects, in some form, at the time of the cue,
most of which is later ‘forgotten’.
Sperling’s paradigm was originally seen as establishing the existence of a non-conceptual and
high-capacity ‘iconic’ visual memory store. This store appears to be short-lived and is likely
pre-categorical, as only some of its content can be processed and reported at any one time.
Importantly for the current discussion, subjects also report seeing ‘all the letters’ of the display,
suggesting that the contents of this kind of iconic memory are present in consciousness.
A range of philosophers have picked up on this language of images, ‘icons’, and pre-
categorical information, and used it to support views both about the richness of visual
phenomenology and the existence of non-conceptual content. For example, Dretske notes
that: ‘subjects extract [letter identity] information from what they describe as a conscious
but rapidly fading image (“icon”) that persists for a short time after removal of the stimulus’
(2006: 175). Block says something similar in discussing the partial report condition: ‘[subjects]
could also report 3–4 items from any row that was cued after stimulus offset, suggesting that
subjects did have a persisting image of almost all the letters’ (2011: 567) (see also Block 2007;
Fodor 2008: 189–90; Tye 2006: 511–13).
And indeed, taking subjects’ reports at face value, it does appear as though these are reasonable
conclusions to draw. However, looking in more detail at contemporary work on iconic memory,
and the temporal nature of consciousness, highlights alternative explanations of the Sperling para-
digm, and so provides challenges to the idea of rich/overflowing visual phenomenology.
3. Iconic memory and the Sperling paradigm: where’s the ‘icon’?

The standard philosophical presentation of iconic memory is as a persisting visual image, from
which letter identities can be read off when a cue is shown. In contrast, the contemporary
scientific understanding of iconic memory suggests that it in fact refers to a range of different
short-term memory stores. These support different visual and memory phenomena, but none of
them provide what supporters of the rich/overflow view need.
One important distinction within iconic memory is between visible persistence and
informational persistence. Visible persistence essentially refers to the phenomenon of visual after-
images; when presented with high-contrast stimuli (e.g. bright areas on a dark background),
stimulus-related activity in early visual areas can continue for up to 100 ms after the stimulus
has disappeared (Brockmole and Wang 2003; Di Lollo 1980). However, what explains per-
formance in the Sperling paradigm is not visible persistence, which is usually well over by the
143
Elizabeth Irvine
time the cues are shown to subjects. Instead, performance in the Sperling paradigm is explained
via the phenomenon of informational persistence, which is supported by a short-term store of
visual information. Informational persistence splits into a visible analogue representation, which
preserves shape and location information for 150–300 ms after stimulus offset, and a non-visual
post-categorical store which preserves abstract information such as letter identity for 500 ms
after stimulus offset (see e.g. Di Lollo 1980; Irwin and Yeomans 1986).
Given the different rates at which these different information stores decay, varying the tem-
poral aspects of the Sperling paradigm can lead to some interesting results. For example, using
a 300–500 ms delay between the letter display and the cue leads to subjects making ‘location
errors’. They still have access to letter identities (from the post-categorical store), but not to
shape and location (from the analogue representation), so can correctly identify some of the
letters in the display, but cannot allocate them to the correct row.
This then generates a series of problems for standard interpretations of the Sperling paradigm.
First, according to the contemporary understanding of visual memory, what enables subjects to
identify letters is not a persisting visual ‘icon’ or ‘image’. The classic Sperling paradigm does not
assess visible persistence, which is related to persisting visual images, only informational persis-
tence, which is not. As Luck and Hollingworth state: ‘the partial-report technique [= when a
row is cued] does not measure directly the visible aspect of visual sensory memory, but rather
that information persists after stimulus onset’ (2008: 16).
However, authors using the Sperling paradigm to argue for rich/overflowing conscious
content tend to conflate visible and informational persistence. For example, in the quotation
from Block above (2011), he notes that: ‘[subjects] could also report 3–4 items from any row
that was cued after stimulus offset, suggesting that subjects did have a persisting image of almost
all the letters’ (2011: 567; emphasis added) (see also Block 2007: 488–90, 494, 532; Tye 2006:
511–13, for similar claims). But the Sperling paradigm shows no such thing; visible persis-
tence, which generates after-images, decays well before cues are showed to subjects, and stored
(unconscious) information is all that is needed to drive subjects’ performance in the task.
Second, it is also difficult to provide positive arguments for the claim that whatever persists
informationally also persists phenomenologically. One issue is the existence of different memory
stores, and how their contents are linked to the contents of visual consciousness. It seems to be
assumed that the link is a transparent one, such that the contents of iconic memory are straightfor-
wardly present in consciousness. However, as noted above, the different stores in informational
persistence decay (and are populated) at different rates. Yet this is apparently not noticed by sub-
jects, who only report general claims of having seen the array, or all the letters. They do not, for
example, report phenomenological changes in where, or whether, letters are spatially located, as
might be expected when letter identities are still available but spatial information is not.
If this is the case, then even if the contents of iconic memory are present in visual phenom-
enology, they are certainly not present in a straightforward way. No such account of how this
works has been offered, but one is necessary to support the claim that the contents of sensory
memory are the contents of experience. Two alternative explanations are given below for what
subjects experience in the Sperling paradigm which are more consistent with the contemporary
science of iconic memory.
4. Alternative explanations: generic phenomenology

Commentaries on Block (2007) abound with suggestions that subjects do not perceive all
the letters in the display in detail, but that they have an experience of something like generic
phenomenology, or an experience of ‘letter-iness’ without an experience of specific letter shapes
144
Memory images
or identities (see also Kouider et al. 2010 on ‘partial awareness’). These suggestions are certainly
plausible on psychological grounds as scene gist is routinely processed quickly and efficiently, and
if subjects are expecting to be presented with an array of letters, then this will be made even easier.
Indeed, the generation of reports consistent with visual richness from gist processing is a
reliable phenomenon, so reliable that it can be utilized to test other features of perceptual pro-
cessing. For example, Castelhano and Henderson (2008) tested how colour contributes to gist
processing using a contextual bias paradigm. Here, subjects were shown a photograph of a scene
for 20–250 ms followed by a 50 ms mask, and then asked whether a target object was present in
the photograph. For a city scene, a target object could be a fire hydrant (consistent with scene)
or a tea set (inconsistent with scene). Subjects often responded that consistent items were present
and that inconsistent items were not.
However, the twist in the paradigm is that none of the target objects were actually present
in the scenes, something which subjects failed to notice. In this paradigm then, subjects gener-
ate reports consistent with having rich/overflowing experiences of specific items even in the
absence of item-specific processing (since here the items simply were not present). This puts
pressure on accounts that take subjects’ reports of having ‘seen all the letters’ at face value (see
Irvine 2011 for a more in-depth discussion of this material).
5. Alternative explanations: postdictive experiences

Phillips (2011a) argues for a slightly different account of the relationships between what sub-
jects report and what they experience. This makes use of a distinction between two alternative
ways that (visual) experience could unfold over time. An ‘Orwellian’ possibility is that a (visual)
experience of an event matches what happens in the event, when it actually happens. However,
this experience might be quickly forgotten and ‘replaced’ with a memory of having seen some-
thing slightly different. Alternatively, a ‘Stalinesque’ possibility is that a (visual) experience of
an event lags behind the event itself, where, as above, you may end up seeing the event in a
slightly different way (see Dennett 1993 for distinction). One reason for seeing the event slightly
differently is if information around the event is integrated over time; in particular, this could
potentially lead to the subject experiencing a (visual) illusion.
The standard interpretation of the Sperling paradigm is an Orwellian one: subjects experi-
ence the contents of the display when it is present, but details of the display are rapidly forgotten,
and subjects can only report the products of later processing (the three-letter identities that make
it into working memory).
Phillips’s argument is that there is also a plausible Stalinesque or ‘postdictive’ interpretation of
the paradigm. Here, subjects have no experience of specific letters of the display before the cue,
and only have letter-specific experiences of a few specific letters after the cue. Phillips argues
for this using examples of other postdictive phenomena where what subjects experience and
report about a stimulus depends on what one shows after it. Importantly, there are examples of
postdictive effects that can occur over the kind of time-spans found in the Sperling paradigm
(e.g. several hundred ms), and that can occur cross-modally (so can account for both visual and
audio cues used in the paradigm).
One of these examples is the sound-induced visual bounce (Sekuler et al. 1997). Here, sub-
jects are presented with a video of two balls moving towards each other across a screen, which,
when they touch, can be seen as bouncing off each other, or streaming past/through each other.
If a sound is played around the point where they touch, subjects are more likely to report seeing
bouncing rather than streaming. The authors report that sounds played 150 ms after the point of
coincidence make perception of bouncing more likely, and Choi and Scholl (2006) make this
145
Elizabeth Irvine
up to 200 ms after coincidence. Watanabe and Shimojo (2001) also found that when a sound is
played at the point of coincidence, the likelihood of seeing a bounce is affected by other sounds
played up to 500 ms either side of it.
So, here is a clear example of a cross-modal postdictive phenomenon with parameters similar
to those found in the Sperling paradigm. Subjects’ reports about the stimuli (two moving balls)
are affected by a subsequent stimulus (sound) that can occur up to 500 ms later. Phillips (2011b)
also outlines an account of attention in which attending to a location amplifies whatever infor-
mation has already gained from that location. This makes it possible for subjects to attend to
information that is no longer physically present, or present in consciousness.
These two ideas can be used to suggest an alternative Stalinesque interpretation of the Sperling
paradigm: subjects only experience specific letters in the cued rows after the cue is shown, where
cues direct attention to amplify pre-conscious representations of the letters. Importantly, this is also
consistent with the existence of generic phenomenology before the cue, thus enabling Phillips to
‘explain away’ subjects’ reports of visual richness in a similar way to the account offered earlier.
The temporal sequence of experiences is then as follows: when the display is present, subjects
have an experience of generic phenomenology (general letter-iness in a grid formation), which
is added to several hundred milliseconds later by an experience of the specific letters in the cued
row that they are asked to report on. Reports of seeing ‘all the letters’ stem from a combina-
tion of generic phenomenology and the fact that subjects can identify most of the letters in the
cued row. Instead of a persisting experience of specific letters enabling performance at the task,
short-term memory (informational persistence) stores letter identities that can be accessed with
an attentional cue. Having provided an alternative account of the Sperling paradigm, Phillips
concludes that the onus is on proponents of the Orwellian/rich/overflow account to argue why
their account is superior (something he thinks it difficult to do).
6. The mis-match problem

One problem with the sparse/non-overflow accounts above is how the contents of various bits
of processing come together in phenomenology in a way that preserves how things ‘seem’ to us.
In particular, it looks like proponents of the sparse/non-overflowing view need to give some
account of what generic phenomenology is like, how it differs from ‘specific’ phenomenology,
and how the two might contribute to conscious experience. To illustrate this problem, Block
writes that before and after the cue in the Sperling paradigm, which is supposed to mark a
shift between pre-cue generic phenomenology and post-cue specific phenomenology: ‘subjects
report no such phenomenological shift . . . The vast literature on this topic . . . contains no
mention of such a thing as far as I know. I myself can testify that even looking for such a shift,
one does not experience it’ (Block 2007: 532).
In response, it is essential first to reiterate that, according to the discussion above, there
is no persisting visual image of the display that lasts until the cue (visible persistence is very
short-lived), so there is no continuing visual image in which a shift from generic to specific
phenomenology could be marked. It could be that generic phenomenology is experienced first
(when the display is first presented), and specific phenomenology of individual letters, recon-
structed from memory, only comes much later (with a gap in the middle). Yet even if there is
not a sudden shift in the phenomenology in ongoing experience, one might think that the dif-
ferent types of phenomenology would ‘look’ different and so be noticed anyway.
Here, though, one can point out that all parties face general problems in linking up informa-
tion processing, memory content, and phenomenology. Proponents of the rich/overflowing
view, as above, need to give an account of why subjects do not report phenomenological changes
146
Memory images
as the contents of iconic memory change and decay over time. Proponents of the sparse/non-
overflowing view of conscious content need to explain why subjects do not report on differences
between generic and specific phenomenology. For both parties then, there is no straightforward
link between how things appear and the contents of reports, memories, and perceptual process-
ing. In this case, neither side is at a clear advantage from the point of view of accounting for
visual phenomenology.
However, this ‘mis-match’ problem may be based on a ‘movie-screen’ model of con-
sciousness, which assumes that visual phenomenology is, more or less, like a movie screen:
where spatial or temporal gaps are noticeable, low-grade images are noticeably different to
high-grade ones (e.g. sharp vs. fuzzy), and where parts of images have to be located and col-
oured and determined all the time. If one assumes a movie-screen model of consciousness,
then the lack of a noticeable jump between generic and specific phenomenology is a problem
for the sparse/non-overflow view, because on a movie screen, it would be very obvious.
However, if one rejects the movie-screen model, then the mis-match problem can be
explained away. And indeed, sparse/non-overflow accounts are amenable to ‘illusions’ of rich-
ness via generic phenomenology, mis-matches between reports and information processing, and
the mind having clever ways of dealing with gappy, changing, indefinite content as and when
needed. That is, there need be no noticeable shift between generic and specific phenomenol-
ogy, nor between rapidly changing content, because perceptual experience is just not of the
format where this could be noticed. We can be mistaken about how much access we have to
the world at any point in time via perceptual experience.
Although the idea of illusions in consciousness are often deemed problematic, illusions or
mis-matches are arguably just what you end up with if you reject a movie-screen model of con-
sciousness and take note of empirical and theoretical work on perception (see early discussions of
these issues in Noë 2002). As such, the sparse/overflow accounts described above, that reject the
movie-screen model and are based on well-known perceptual phenomena, seem well placed to
account for the complex relations between the contents of iconic memory and phenomenology.
One might wonder however what a rich/overflow account would look like without the
movie screen assumption. Yet dropping this assumption removes much of the motivation
behind the view. The primary reason for thinking that the contents of iconic memory are in
the contents of phenomenal consciousness is that subjects’ reports suggest as much; they report
seeing ‘all the letters’. Yet if visual experience is not as of a movie screen, and mis-matches
between reports and information processing or memory are possible, then there is no strong
reason to continue to take reports at face value, given conflicting empirical evidence. Instead,
we allow other routes to generating these kind of reports (e.g. based on generic phenomenology
and expectations), and so end up at the sparse/non-overflow view.
While more work needs to be done on exactly how these mis-matches work, explanations of
the Sperling paradigm featuring generic phenomenology may be on the winning side. However,
more recently, another kind of short-term memory store has been proposed to contribute rich
phenomenal content to consciousness, that cannot be explained away using the strategies here.
This leads into deeper questions still about the nature of working memory and its relationship
to phenomenal content, discussed below.
7. Fragile VSTM
More recently, variations of change detection paradigms have been used to suggest that there
is another type of memory that potentially provides a richer, and longer lasting, set of conscious
content that overflows working memory (Block 2011; also see Lamme 2006). Fragile visual
147
Elizabeth Irvine
short-term memory (fragile VSTM) is argued to sit between iconic memory and working mem-
ory in terms of capacity and duration (e.g. Sligte et al. 2010, 2008; Vandenbroucke et al. 2011).
Because of its longer duration, postdictive explanations can’t be applied (they only work for
short time spans), and while appeals to generic phenomenology can still be used, some mecha-
nism other than informational persistence needs to be used to explain how subjects are able to
identify apparently large numbers of letters long after they have disappeared from view.
First, it is important to understand the source of the evidence in favour of the existence of
fragile VSTM. The paradigm used to identify fragile VSTM uses three conditions as follows
(see also Figure 11.2):
• Iconic-cue condition: A ‘memory array’ of items is displayed for around 250 ms, and sub-
jects are asked to remember as many of them as possible. A cue to a location of a particular
item in the array is displayed to subjects within the temporal window used in the Sperling
paradigm. After this, a probe or test display of similar objects to the memory array is shown,
and subjects have to state whether the cued item has changed. As the name suggests, this
condition tests the capacity of iconic memory, which is fairly high.
• Retro-cue condition: As above, a ‘memory array’ of items is displayed for around 250 ms,
and subjects are asked to remember as many of them as possible. This time a cue is shown
long after iconic memory has decayed (1,000 ms). After this, a test display is shown of similar
objects to the memory array, and subjects have to state whether the cued item has changed.
• Post-cue condition: As above, a ‘memory array’ of items is displayed for around 250 ms,
and subjects are asked to remember as many of them as possible. After a delay interval of up
to 1,000 ms, a test array is displayed, and then disappears, and then a cue to a particular item
location is displayed. Subjects have to state whether the cued item has changed. The post-
cue condition is supposed to measure the capacity of working memory, as the presentation
of the test array before the cue provides a lot of visual interference, so should overwrite any
information that is not already in working memory.
Memory array, Memory array, Memory array,

250 ms 250 ms 250 ms
10 ms delay 1000 ms delay 900 ms delay
Cue, 490 ms Cue, 500 ms Test array,

joined by cue
10 ms later,
Delay, 1500 ms Delay, 500 ms until response
Test array until Test array until

response response
Iconic-cue condition Retro-cue condition Post-cue condition

(short delay (long delay (test array
between memory between memory shown before cue)
array and cue) array and cue)
Figure 11.2 The basic differences between iconic-cue, retro-cue, and post-cue conditions. Adapted
from Sligte et al. (2008: Figure 1).
148
Memory images
The interesting condition here is the retro-cue condition. The memory capacity measured in this
condition varies widely over the specifics of the task (e.g. up to 10 low-resolution simple items,
down to 3 high resolution complex items, see Sligte et al. 2008, 2010), but the capacity appears to
be robustly above that of working memory (measured using the post-cue condition) and below
that of iconic memory (measured using the iconic-cue condition). This evidence is then used as
evidence of a third type of visual memory: fragile visual short-term memory, or fragile VSTM
(see Sligte et al. 2008 for defence of this). It is argued that subjects in these studies use their visual
experiences, supported by fragile VSTM, to perform the change detection and identification tasks.
Again then, this can be used to provide support for rich/overflowing accounts of the contents of
consciousness; subjects seem to have seen more items than they can report at any one time.
Although the strategies used above against rich/overflowing content cannot be straight-
forwardly applied to untangle fragile VSTM, there are other ways to challenge the idea that
fragile VSTM contributes rich content to visual experience. These focus on how the capac-
ity of fragile VSTM is measured, and if it actually forms a separate memory store to working
memory after all. If it does not, then fragile VSTM cannot be used as part of a rich/overflow
account of conscious content. This raises further important questions about the nature of
working memory, and of its links with consciousness.
8. Challenges to fragile VSTM

The first challenge is that change detection (used by Sligte et al. 2008) is a lot easier than iden-
tification (used in the Sperling paradigm). Subjects in the change detection paradigm just have
to process enough information to tell if an item has changed (e.g. changed orientation, changed
colour), which is a much less stringent way of assessing memory capacity than requiring subjects
to freely identify letters. Phillips (2011a: 405–6) raises this challenge and suggests that given that
change detection has such low informational demands, it might even be the case that it can be
performed successfully using unconscious priming mechanisms. Similarly, it could be the case
that subjects use feelings of familiarity generated from unconscious processing to make a judg-
ment about whether the display has changed, rather than comparing specific visual experiences.
Sligte and colleagues (2010) pick up this criticism and test change detection and an identi-
fication task with complex items (harder), instead of a change detection task only on oriented
rectangles (easier), to measure the capacity of fragile VSTM. However, presumably partly to
preserve the structure of the experimental paradigm, the identification task is still only a forced
decision task among 4 alternatives (so still far easier than in the Sperling paradigm, in which
there are 26 alternatives). Yet even in these conditions measured capacity drops sharply: capacity
of fragile VSTM measured according to change detection is 4.6 items, and capacity measured
according to the identification task is 3.3 items. Importantly, these capacity measures are at the
high end, but within, the bounds of working memory capacity.
Further questions can be raised about measures of the confines of fragile VSTM based on
the type of stimuli used. Earlier experiments from Landman and colleagues (2003) using the
change detection paradigm were problematic, as subjects could have been ‘chunking’ oriented
rectangles into larger spatial groups. In this case, capacity measures would have massively over-
estimated the capacity of fragile VSTM: instead of storing 32 individual rectangles, the array
could have been stored as 4 larger chunks. Sligte and colleagues (2008) controlled for this by
rotating all the non-test items in the array by 90 degrees in the test array (intended to rule out
chunking strategies), and in Experiment 3, by using 4 orientations rather than only 2. Under
these conditions, again capacity measures dropped significantly to 5.5 items, which is at the
high-end of working memory capacity.
149
Elizabeth Irvine
Another feature of these experiments investigated by Matsukura and Hollingworth (2011)

is the amount of practise that subjects need to get up to the capacity measures found in the
Sligte studies. In the initial study (Sligte et al. 2008) subjects had 3 hours of practise, and could
also select to re-do trials in the experimental phase. Matsukura and Hollingworth found that
without a significant amount of practise, their subjects had much lower capacity measures, and
even with 80 minutes of practise they still had lower capacity measures than those found by
Sligte and colleagues. Practise and training also remain features of the later studies, with Sligte
and colleagues (2010) giving less training time, but requiring a certain performance level before
subjects engaged in the main task.
The worry here is that practise need not necessarily increase memory capacity, but can make
coding, storage, and retrieval more efficient, making it look like capacity has increased. In this
case, subjects may be working at the high end of working memory capacity, rather than using a
separate, higher-capacity memory store.
The challenges above are aimed mainly at the capacity measures of fragile VSTM, to show
that they are actually within standard capacity measures for visual working memory, though
often at the higher end. Yet there is still the fact that the capacity measures for the retro-cue
conditions (supposed to measure fragile VSTM) are usually double those for the post-cue condi-
tions (supposed to measure working memory). So, even if the capacity measures are not actually
that high, there still seems to be evidence in favour of a separate higher-capacity memory store.
However, there are reasons to question this. Matsukura and Hollingworth (2011) also tested
whether there is a capacity difference between a partial report condition (the cueing condition in
the Sperling paradigm) and something similar to a full report condition (condition with no cues
in the Sperling paradigm). In the partial report condition, a cue is shown after the main display
(as is standard), but in the full report condition here, all the items in the display are cued, effec-
tively giving a neutral cue, but also providing visual interference. Under these conditions, one
would expect much higher-capacity measures for the partial report condition, which assesses frag-
ile VSTM, than the full report condition, which due to the visual interference provided should
only assess the contents of working memory. The authors found a small difference between the
two conditions, but the capacity limits of fragile VSTM were still within the bounds of working
memory capacity (the higher capacity measure for fragile VSTM = 4.7 items).
Somewhat turning the tables in these debates, Matsukura and Hollingworth suggest that
the differences in capacity measures can be explained as attentional effects within working
memory itself: attention can prevent decay or interference to selected items in working memory
(Makovski et al. 2008). This goes against standard conceptions of attention and working mem-
ory; that attention directs content into working memory, which provides (potentially only some
of) the contents of consciousness. Here though, attention also seems to modulate the content of
working memory, and so the contents of consciousness, making them more or less ‘available’
to other consumer systems.
Importantly, this idea that attention can work even within the contents of working memory
presents a strong challenge to the way that fragile VSTM is identified. Makovski (2012) and
Rerko and colleagues (2014) show that retro-cues still benefit performance at change-detection
tasks, even if the cue comes after visual or cognitive inference, or a switch of attention, both of
which should wipe out fragile VSTM. Note that the post-cue condition is used to test work-
ing memory capacity precisely because the test array provides visual inference that wipes out
anything else. So here, even in conditions that are only supposed to test the capacity of working
memory, it is still beneficial to give subjects an attentional cue before the test array is shown.
If these retro-cue benefits occur for content that is squarely in working memory, then higher-
capacity measures gained during retro-cue conditions vs. post-cue conditions (or vs. no cue or
150
Memory images
neutral cue condition) cannot be used as a criterion to separate two kinds of memory; they simply
illustrate the effects of attention on the contents of working memory.
Here then, not only is the capacity of fragile VSTM not particularly high, but it is not a
separate memory store from working memory. What is termed ‘fragile VSTM’ is the product
of efficient encoding, storage, and retrieval within working memory, with storage and retrieval
made yet more targeted by using an attentional cue. There is therefore no need to posit fragile
VSTM as an additional memory store that ‘overflows’ the contents of working memory. If frag-
ile VSTM simply doesn’t exist, then its contents obviously cannot provide support in favour of
the rich/overflowing view of conscious content.
9. What’s in working memory?

More pressing for discussions about the relationship between short-term memory and the con-
tents of visual phenomenology, is that there now appear to be serious questions about how to
assess the capacity and contents of working memory. In addition to the experimental work
above on showing the effects of attention within working memory, it is fairly well known that
measured working memory capacity is sensitive to the complexity of the items it stores (Alvarez
and Cavanagh 2004), and to the type of testing used (Makovski et al. 2010). While it is still
controversial whether working memory has a limited numbers of slots or is a more divisible
resource (Bays and Husain 2008 and challenges), there seem to be several sources of variation
in capacity.
This is problematic for rich/overflow views, as stating that conscious content overflows
working memory relies on there being a fairly foolproof way of assessing what the contents of
working memory are. Yet if the capacity of working memory can be manipulated by atten-
tion, by practise, by the task and the visual stimuli used, then it is potentially quite difficult to
identify what would count as convincing empirical evidence of content that overflows working
memory.
It is also problematic for sparse/non-overflow views, but in a slightly different way. If work-
ing memory is defined functionally (roughly, as that which makes content available to other
consumer systems, including allowing subjects to act and report on content), then whether or
not we have a stable way of measuring its content is perhaps not pressing. The claim is just
that the contents of working memory are the contents of consciousness, whatever the contents
of working memory end up being. Yet the very same problems that affect the measurement of
working memory also affect how to define its function. If the functional definition of working
memory is abstract enough that these problems of measurement can be ignored, then it may
become an ambiguous and/or empty definition, and so leave theories of consciousness that
involve working memory also problematically ambiguous.
10. Conclusion
This chapter has analysed arguments that short-term visual memory provides rich conscious
content that overflows the contents of working memory. These arguments were based on
Sperling’s partial report paradigm (iconic memory) and Sligte’s change blindness studies (fragile
VSTM) which appear to show that more content is stored than can be reported at any one time,
and that this content is present in consciousness. It was suggested above that there are alterna-
tive explanations for subjects’ behaviours and reports of richness in the Sperling paradigm that
fit with what is known about the structure of iconic memory and perception more generally.
In addition, it has been argued that fragile VSTM likely does not exist as a separate store to
151
Elizabeth Irvine
working memory. Both iconic memory and fragile VSTM are therefore poor candidates for
supporting the idea that the contents of experience overflow the contents of working memory.
These debates have highlighted deeper issues about how to relate the contents of information
processing and memory to the contents of consciousness, and how to identify the contents of
working memory in the first place; questions to which both sides of the debate need to have
an answer.
And one final point that is often skated over: even if we were to accept that the contents
of iconic memory and fragile VSTM are present phenomenally, it arguably would not provide
much support for a truly ‘rich’ view of experience. If the contents of iconic memory and fragile
VSTM overflow working memory, they do not do so by very much. In the Sperling paradigm,
subjects can store information about 9 (out of 12) letters in iconic memory, and in the more
conservative tests of fragile VSTM, capacity is around 4–5 items (less for complex objects).
While having a detailed visual experience of 9 letters is more impressive than experiencing
only 3, it does not support a picture of particularly rich visual detail. In this case, even if the
arguments above fail to convince, the experimental work still presents a rather sparse picture of
the contents of consciousness.
Related topics
• Taxonomy and unity of memory
• Memory and consciousness
References
Alvarez, G. A., Cavanagh, P. (2004). ‘The capacity of visual short-term memory is set both by visual
information load and by number of objects’. Psychol. Sci. 15: 106–11.
Bays, P. M., Husain, M. (2008). ‘Dynamic shifts of limited working memory resources in human vision’.
Science 321: 851–4.
Block, N. (2007). ‘Consciousness, accessibility, and the mesh between psychology and neuroscience’.
Behav. Brain Sci. 30: 481–99.
Block, N. (2011). ‘Perceptual consciousness overflows cognitive access’. Trends Cogn. Sci. 15: 567–75.
Brockmole, J., Wang, R. F. (2003). ‘Integrating visual images and visual percepts across time and space’.
Vis. Cogn. 10: 853–73.
Castelhano, M. S., Henderson, J. M. (2008). ‘The influence of color on the perception of scene gist’.
J. Exp. Psychol. Hum. Percept. Perform. 34: 660.
Choi, H., Scholl, B. J. (2006). ‘Perceiving causality after the fact: Postdiction in the temporal dynamics of
causal perception’. Percept.-Lond. 35: 385.
Cohen, M. A., Dennett, D. C. (2011). ‘Consciousness cannot be separated from function’. Trends Cogn.
Sci. 15: 358–64.
Dehaene, S., Changeux, J.-P., Naccache, L. (2011). ‘The global neuronal workspace model of conscious
access: From neuronal architectures to clinical applications’, in S. Dehaene and Y. Christen (eds.),
Characterizing Consciousness: From Cognition to the Clinic? Berlin: Springer, pp. 55–84.
Dennett, D. C. (1993). Consciousness Explained. Harmondsworth: Penguin UK.
Di Lollo, V. (1980). ‘Temporal integration in visual memory’. J. Exp. Psychol. Gen. 109: 75.
Dretske, F. (2004). ‘Change blindness’. Philos. Stud. 120: 1–18.
Dretske, F. (2006). ‘Perception without awareness’, in T. S. Gendler and J. Hawthorne (eds.), Perceptual
Experience. New York: Oxford University Press, pp. 147–80.
Dretske, F. (2007). ‘What change blindness teaches about consciousness’. Philos. Perspect. 21: 215–20.
Fodor, J. (2008). LOT 2: The Language of Thought Revisited: The Language of Thought Revisited. Oxford:
Irvine, E. (2011). ‘Rich experience and sensory memory’. Philos. Psychol. 24: 159–76.
152
Memory images
Irwin, D. E., Yeomans, J. M. (1986). ‘Sensory registration and informational persistence’. J. Exp. Psychol.
Hum. Percept. Perform. 12: 343.
Kouider, S., De Gardelle, V., Sackur, J., Dupoux, E. (2010). ‘How rich is consciousness? The partial
awareness hypothesis’. Trends Cogn. Sci. 14: 301–7.
Lamme, V. A. (2006). ‘Towards a true neural stance on consciousness’. Trends Cogn. Sci. 10: 494–501.
Landman, R., Spekreijse, H., Lamme, V. A. (2003). ‘Large capacity storage of integrated objects before
change blindness’. Vision Res. 43: 149–64.
Luck, S. J., Hollingworth, A. (2008). Visual Memory. Oxford: Oxford University Press.
Makovski, T. (2012). ‘Are multiple visual short-term memory storages necessary to explain the retro-cue
effect?’ Psychon. Bull. Rev. 19: 470–76.
Makovski, T., Sussman, R., Jiang, Y. V. (2008). ‘Orienting attention in visual working memory reduces
interference from memory probes’. J. Exp. Psychol. Learn. Mem. Cogn. 34: 369.
Makovski, T., Watson, L. M., Koutstaal, W., Jiang, Y. V. (2010). ‘Method matters: Systematic effects of
testing procedure on visual working memory sensitivity’. J. Exp. Psychol. Learn. Mem. Cogn. 36: 1466.
Matsukura, M., Hollingworth, A. (2011). ‘Does visual short-term memory have a high-capacity stage?’
Psychon. Bull. Rev. 18: 1098–104.
Noë, A. (ed.) (2002). ‘Is the visual world a grand illusion?’ Journal of Consciousness Studies: Controversies in
the Science & the Humanities. Imprint Academic.
Phillips, I. B. (2011a). ‘Perception and iconic memory: What Sperling doesn’t show’. Mind Lang. 26:
381–411.
Phillips, I. B. (2011b). ‘Attention and iconic memory’. In Mole, C., Smithies, D., Wu, W. (eds.), Attention:
Philosophical and Psychological. Essays. Oxford: Oxford University Press, pp. 204–27.
Rerko, L., Souza, A. S., Oberauer, K. (2014). ‘Retro-cue benefits in working memory without sustained
focal attention’. Mem. Cognit. 42: 712–28.
Sekuler, R., Sekuler, A. B., Lau, R. (1997). ‘Sound alters visual motion perception’. Nature 385: 308.
Sligte, I. G., Scholte, H. S., Lamme, V. A. (2008). ‘Are there multiple visual short-term memory stores?’
PLOS One 3: e1699.
Sligte, I. G., Vandenbroucke, A. R., Scholte, H. S., Lamme, V. (2010). ‘Detailed sensory memory, sloppy
working memory’. Front. Psychol. 1: 175.
Sperling, G. (1960). ‘The information available in brief visual presentations’. Psychol. Monogr. Gen. Appl. 74: 1.
Tye, M. (2006). ‘Nonconceptual content, richness, and fineness of grain’, in Gendler, T. S., Hawthorne,
J. (eds.), Perceptual Experience. New York: Oxford University Press.
Tye, M. (2009). Consciousness Revisited: Materialism without Phenomenal Concepts. Cambridge, MA: MIT
Press.
Vandenbroucke, A. R., Sligte, I. G., Lamme, V. A. (2011). ‘Manipulations of attention dissociate fragile
visual short-term memory from visual working memory’. Neuropsychologia 49: 1559–68.
Watanabe, K., Shimojo, S. (2001). ‘When sound affects vision: Effects of auditory grouping on visual
motion perception’. Psychol. Sci. 12: 109–16.
153
12
MEMORY AND EMOTION
Ronald de Sousa
Truly, though our element is time,

We are not suited to the long perspectives
Open at each instant of our lives.
They link us to our losses: worse,
They show us what we have as it once was,
Blindingly undiminished, just as though
By acting differently, we could have kept it so.
‘Reference Back’, Philip Larkin
At the highest level of biological generality, we can discern a rough functional analogy between
the most primitive tropisms and the most sophisticated human emotions. Both serve to detect and
respond differentially to situations that affect the well-being of an organism. Similarly, elemen-
tary changes are perpetuated through time in the most primitive organism as well as in explicit
individual memory. Thus, in the words of the psychologist Nico Frijda, emotions ‘can . . . be
traced back in evolution to elementary principles of being alive’, notably autonomy, arising from
autopoiesis (Varela et al. 1974), a capacity for detection or perception of relevant cues, guided by
a set of concerns (Frijda 2016: 610). While the ‘concerns’ of bacteria might seem too remote to
mention in the same breath as human emotions, the term draws attention to the parallel between
sophisticated emotions and wants on the one side, and the needs served by tropisms towards nutri-
ents or away from threats, on the other. Similarly, a common function is served by memory and
by the mere preservation of an acquired change in the simplest organism. Memory and emotion
together embody fundamental enabling processes of natural selection: differential success, among
a variety of possible processes or behaviours, and the preservation of successful forms or strategies.
Such a broad biological perspective obviously leaves out all interesting peculiarities and dif-
ferences. By being too inclusive about what falls under the label we are likely to ‘reduce the
term memory to a meaningless designator’ (Klein 2015: 2). And the same goes for emotions.
Tropisms are not emotions, any more than the effects of phylogenetic antecedents are literally
memory. Frijda conceded that ‘emotions . . . do not . . . reflect a unitary function’ (2016: 618).
Neither does memory. We need to look more closely at how emotions and memory intertwine
as their complexity culminates in the ways they function in human life.
154
Memory and emotion
1. The personal and the subpersonal

Following Endel Tulving (1972, 1985), psychologists distinguish between procedural, semantic, and
episodic memory. The first, characterized as ‘anoetic’ because it does not require consciousness,
consists in an acquired ability to exercise a skill. That includes the ability to say something, so
procedural memory is presupposed by semantic memory, which is the retention of knowledge.
Semantic memory is ‘noetic’, i.e. it can normally be brought to consciousness, but the circum-
stances of its acquisition are lost. Episodic memory is ‘autonoetic’: it takes itself to be knowledge
of the past. As such, it could be viewed a species of semantic memory of which the differentia is
that its content can be located from a given perspective within a temporal frame.
There are obvious and intimate links between memory and emotion. In Proust’s famous
madeleine experience (Proust 1966: 48), a present experience seems to bring back a past time.
It brings a deep and initially inexplicable feeling of joy that is not, at first, attributed to specific
past events. Marcel’s search of the origin of his present feelings uncovers not only episodic
memories but also a feeling of their pastness in a first-person perspective. They are memories
par excellence, and some have suggested that only full-fledged episodic memories should count as
true memories (Klein 2015); others, however, have urged that semantic and episodic memory
form a single natural kind from which only procedural memory should be excluded (Michaelian
2015). Both about memory and about emotion, we can ask what exactly each is most helpfully
regarded as comprising.
Disputes as to what should count as a natural kind are common on the somewhat unsettled
border between philosophy and science. There are clear cases where what appear to be the same
kind are really two. Nephrite and Jadeite have very much the same properties for carvers and
collectors, who refer to both as jade. But they are as different as Ca2(Mg,Fe)5Si8O22(OH)2 is
from NaAlSi2O6. That matters for scientific purposes, because chemical composition is conven-
tionally regarded as defining a substance’s identity. But for some purposes it may be irrelevant.
In psychology and in philosophy, where we are often interested in functions at different levels
of analysis, what counts as a natural kind is not so clear (de Sousa 1984).
Accordingly, some have denied that emotions form a natural kind (Rorty 1980; Griffiths
1997), while others have insisted that they do (Charland 1995). Who is right depends on why
you want to know. In the useful terms introduced by David Marr (1982), are we asking about
the computational, the algorithmic, or the implementation level? The biological perspective
favoured by Frijda invites us to take an interest both in what is the same, at the computational
level, and what is different at the level of algorithm and implementation. Like Aristotle’s dis-
tinction between matter and form, Marr’s levels are relative: a process implementing a certain
algorithm could itself define a task implemented by some lower level process. Relative to the
biological goals I began with, procedural memory is a mechanism of implementation; but
relative to the way it is effected in the brain, it can be thought of as a function requiring
an algorithm for its implementation. Some such multilevel complexity characterizes common
conceptions of emotion among psychologists. Klaus Scherer, for example, has defined an emo-
tion as ‘an episode of interrelated, synchronized changes in the states of all or most of the five
organismic subsystems in response to the evaluation of an external or internal stimulus event as
relevant to major concerns of the organism’ (Scherer 2005).1 Scherer’s characterization leaves
out an essential feature of all but the simplest of our affective responses: that they are pervaded
by memory at both the personal and the subpersonal levels. The emotions we have names for –
fear, anger, jealousy, sadness, and so forth – effectively rehearse paradigm scenarios that have a
narrative structure and call forth more or less pre-packaged responses. That typically involves
a narrowing of attention: emotions determine patterns of salience among objects of attention,
155
Ronald de Sousa
lines of inquiry, and inferential as well as practical strategies (de Sousa 1987: 196). Procedural
and semantic memory are always involved, and sometimes, when we are explicitly experiencing
an emotion targeting a past time, episodic memory is involved as well.
One consequence of these close links between emotions and memory is that the scripts and
scenarios in question have to be learned. The exploration of that aspect of emotion that links
it specifically to memory was highlighted by Freud, who thought it likely that the learning in
question takes place mostly in early childhood. Viewed in this perspective, an adult’s repertoire
of emotions consists largely in paradigm scenarios the dramatic structure of which was acquired
in infancy and rehearsed with only limited adjustments in later life. Insofar as this speculation is
correct, the construction of emotional episodes must include all three types of memory, includ-
ing the sort of retention most remote from standard episodic memory, where the subject has no
awareness of the original episode. Whatever their status as natural kinds, the close imbrication
of all three kinds of memory in relation to emotion seems to count against segregating conscious
episodic memory from the other ways in which the past influences the future.
A useful perspective on the shaping of our repertoire of emotions by our past is afforded by
looking at what we might mean by ‘semantic memory’ if we took it literally, as referring to our
knowledge of language. Our knowledge of the meaning of words requires us to know how to use
them. So this part of our semantic memory directly involves procedural memory. But how did
we acquire it? No two speakers of English had exactly the same sequence of experiences in the
first years of life. Each one of us, therefore, has become proficient in our native tongue in idiosyn-
cratic ways. Except for the odd word acquired on some memorable occasion, we generally have
lost all episodic recollection of the occasion on which we first learned it. Some misunderstandings
between native speakers of the same language can be expected as a result of generally undetected
differences between their idiolects. Subtle differences may arise between two speakers’ proto-
types (in the sense of Rosch and Lloyd 1979) for common words. We can speculate that the
specific circumstances in which we first learned to speak of ‘airplanes’, ‘apricot’, or ‘assignation’
influence the connotation of those words, if not their Fregean sense and reference. Resulting dif-
ferences between idiolects are often trivial. But a more poignant (fictional) example is provided
by Kundera’s ‘Dictionary of Misunderstood Words’, in which the two protagonists differ radically
in the valence they associate with a number of emotionally charged words (Kundera 1984: 96).
Some effects of past experience are confined to the subpersonal. Mere familiarity, even when
unconsciously induced, can determine preference (Zajonc 2000). Emotional and behavioural
responses can be influenced by irrelevant associations, set up by post-hypnotic suggestion. Thus
subjects who were caused to associate disgust with an arbitrary word, such as ‘take’ or ‘often’,
passed harsher moral judgements on the protagonist of a story when it included the innocent
word (Wheatley and Haidt 2005).
Ever since Bartlett’s experiments on the transmission of folk stories (Bartlett 1920), it has
become clear that episodic memories are not retrievals of unchanged memory traces. They are
imaginative constructions, produced on the basis of acquired ‘schemata’ similar to paradigm
scenarios for emotions. Allocating the scene to a specific time in the past requires an additional
element, which Bertrand Russell identified as a ‘specific feeling or sensation or complex of
sensations, different from expectation or bare assent in a way that makes the belief refer to the
past’ (Russell 1921: 186). As Jérôme Dokic has argued, however, that feeling is better inter-
preted as a subpersonal meta-cognitive feeling specific to episodic memory (Dokic 2014). It is
akin to a feeling or rightness or the feeling of knowing, and includes a ‘de se’ attribution (Perry
1979); that is, it is interpreted by the subject as referring to her own experience. Like other
meta-cognitive feelings, it is not necessarily conscious as such, neither is it necessarily veridical.
156
Memory and emotion
This can be illustrated by Penfield’s experiments with direct stimulation of the temporal lobes.
Subjects thought they identified ‘flashback’ memories; but further investigation established that
they were ‘reconstructions or inferences rather than actual memories . . . one patient, for exam-
ple, said upon stimulation that she suddenly saw herself as she had appeared in childbirth, and
that she felt as if she were reliving the experience’ (Loftus and Loftus 1980).
The schemata on the basis of which both memories and emotions are constructed make
use of specific experiences and cultural assumptions about ‘normal’ events and responses. This
becomes particularly clear in certain pathological cases, when patients confabulate elaborate
accounts of adventures they supposedly underwent, while actually being locked up in hospital.
Those confabulations clearly derive from beliefs about what someone might be expected to do,
rather than events actually experienced (Hirstein 2005).
Joëlle Proust has argued that feelings form a natural kind even if emotions don’t. They are not
necessarily affective, but typically are positively or negatively valenced. At the most elementary
level, they motivate something like approach or avoidance, and more generally function to
guide behaviour and monitor performance. To that end, ‘they carry non-conceptual informa-
tion about how much one’s present condition deviates from the expected condition. From a
functional viewpoint, they form a natural kind insofar as their function is to indicate a com-
parative outcome through a dedicated embodied experience’ (Proust 2015: 6). The monitoring
function, in particular, involves an important class of meta-cognitive feelings, such as a feeling of
rightness, a feeling of knowing, or a feeling of familiarity. These resemble emotions in that they
are sometimes valenced; they also both influence and result from appraisals. Unlike emotions,
however, they require neither a propositional object nor even any faculty of conceptualization.
Among those meta-cognitive or ‘epistemic feelings’ (de Sousa 2008), the feeling of famili-
arity is particularly intriguing, in that it appears to involve both a perceptual and an affective
component, subtended by different brain pathways. When dissociated, these result in strange
symptoms like those of Capgras syndrome.
In his original case description, Joseph Capgras described a patient who claimed that her hus-
band had been replaced by impostors, or ‘sosies’. When her delirium was particularly acute, she
asserted that between 1914 and 1918 over two thousand doubles of her daughter had appeared,
each one ‘neither completely different nor perfectly identical to the last’. ‘In short,’ Capgras
concluded, ‘Mme de Rio-Branco sees resemblance everywhere while failing everywhere to
see identity. So strictly speaking there is no problem with recognition; there is instead what we
might call an identification agnosia’ (Capgras 1923: 13).
Capgras’s own interpretation of the phenomenon sees it as stemming ‘not strictly speaking
from a sensory illusion but from an emotional judgment’ (Capgras 1923: 14). This posits an affec-
tive pre-condition on the recognition of close family or friends. More recent treatments of the
illusion have traced it more precisely to a disconnect between cognitive and affective associations.
Elisabeth Pacherie summed up this idea as follows:
On this model, face recognition involves two information-processing pathways: a

ventral visuo-semantic pathway, that constructs a visual image that encodes semantic
information about facial features and is responsible for overt recognition, and a dorsal
visuo-affective pathway responsible for covert autonomic recognition and for the spe-
cific affective response to familiar faces (the feeling of familiarity) . . . Capgras syndrome
might be a mirror image of prosopagnosia [the inability to recognize faces (see Sacks
2010)], with the affective pathway damaged but the visuo-semantic pathway intact.
(Pacherie 2008: 108)
157
Ronald de Sousa
The Capgras illusion, then, seems consistent with the view that ordinary recognition of inti-
mates is constructed on the basis of meta-cognitive feelings as well as other more complex
emotions. Further illustration of the inextricable collaboration of the personal and subpersonal
elements of memory and feeling is provided by post-traumatic disorders. Dissociated traumatic
memories sometimes appear to be transmuted into ‘somatoform’ or bodily symptoms. In these
cases, explicit memory and normal emotional response are replaced by apparently organic symp-
toms such as back pain, fatigue, headaches, or other unexplained pains. These frequently baffle
medical practitioners, because they do not admit of any organic diagnosis.
One hypothesis to explain this is that the somatoform symptoms result from a disturbed
inner representation of others’ negative attitude to the subject’s desire for closeness (Landa et al.
2012). This hypothesis was supported by interviews that followed the Relationship Anecdotes
Paradigm (RAP) to explore remembered episodes of emotionally significant interactions. The
RAP protocol can be regarded as operationalizing the Freudian concept of transference. By
measuring the extent to which the emotional needs of a subject have been met in the past,
it provides insight into some of the paradigm scenarios that govern a subject’s expectations
in social and intimate interactions (Luborsky and Crits-Christoph 1998). Landa found a high
correlation between somatoform symptom formation and an unmet need for closeness in past
interactions with significant persons (Landa et al. 2012: 414).
The RAP paradigm again illustrates the mutual imbrication of different levels of emotions
and memory. Nostalgia, the pleasure of recollection, and the pain of grief all provide fur-
ther illustrations of the narrative structure of both memory and emotion. In some sequels of
emotional trauma, these functions are disrupted in revealing ways. In addition to somatoform
symptoms, patients suffering from Post Traumatic Stress Disorder (PTSD) commonly exhibit
disruptions of perception, a decrease in attention coupled with increased arousal, and ‘altered
cognitive schemata and social apprehension’. The disruption of those schemata makes it impos-
sible for memories to cohere into a story of the sort normally associated with conscious episodic
memory (van der Kolk and Fisler 1995: 507). When memories lose their narrative structure,
then so does the associated emotion. Declarative memories of the traumatic event are missing:
‘No subject reported having a narrative for the traumatic event as their initial mode of aware-
ness (they claimed not having been able to tell a story about what had happened)’ (van der
Kolk and Fisler 1995: 517). Their memories took the form only of somatosensory flashback
experiences: ‘These flashbacks occurred in a variety of modalities: visual, olfactory, affective,
auditory, and kinesthetic, but initially these sensory modalities did not occur together.’ By
contrast, ‘when people have day-to-day, nontraumatic experiences, the sensory elements of
the experience are not registered separately in consciousness, but are automatically integrated
into the personal narrative’ (van der Kolk and Fisler 1995: 519). However, the mechanisms
responsible for serious pathology in those who have suffered complex traumas in crucial devel-
opmental phases of childhood operate in normal people as well. Anyone might worry about
the extent to which their current emotional responses might misperceive their present target,
because they rest on an unconscious identification of that present person with a figure of one’s
distant past (Freud 1915).
In short, memories of emotionally significant events affect our present emotional dispositions
in both strange and familiar ways. When I read and am moved by a poem, for example, I might
occasionally link the resulting emotion to a specific event in my own life: Ah yes, that is very like
the way I felt when . . . . But when that happens, I might well feel that the association is not so
much a help as a kind of pollution of the poetic experience. Nevertheless, the impact of poetic
language is highly personal. Emotions also affect semantic memory: I may respond intensely to
one Shakespeare sonnet and be left quite indifferent by another; yet there may be no explicit
158
Memory and emotion
memory triggered by the sonnet that moves me. If so, it seems reasonable to infer that some past
event that I cannot recall has influenced the words’ current connotations.
Any object identified by perception is categorized in terms of associations with past events.
These can be explicit, constituting episodic memories which are typically tinged with emotional
significance. There is evidence that unless an experience arouses some sort of affect, it will not
be transferred to long-term memory at all (Goor et al. 1982), and also that experiences with
negative valence are better remembered (Kensinger 2009). One can think of this as a mecha-
nism of relevance built-in by natural selection. If I witness something about which I could not
possibly care one way or another, there is no need to clutter my memory with it; if is pleasant,
remembering it is less important than if it is unpleasant.
2. What should we forget?

Sometimes, an experience is so unpleasant that one might want to forget it. Memory serves us in
many ways. But like other legacies of natural selection, it has its dark side. Sometimes it brings
pain or intrudes in disturbing ways. Should we then strive to forget?
Everyone has had experiences one would rather forget. We may be embarrassed to remem-
ber some faux pas, or wish away the memory of a disturbing scene. Nietzsche thought forgetting
was not a defect but a necessity: ‘there could be no happiness, no cheerfulness, no hope, no
pride, no present without forgetfulness’ (Nietzsche 1967: 35). Indeed, there is evidence that our
minds are automatically geared to forget unpleasant facts. In Bartlett’s original experiments on
the transformations of memory representations, unpleasant or shocking details, as well as those
that were merely unfamiliar in the context of a subject’s social environment, are frequently
omitted when people attempt to reproduce a story (Bartlett 1920: 36).
The protagonists of the movie The Eternal Sunshine of the Spotless Mind each decide to resort
to Lacuna, a technology that promises to erase all memories of a particular person. They do
this because their love affair is painful. That, at least, is the obvious motive. Troy Jollimore has
offered a more subtle reason: namely that one might want to forget a former lover to make
room for a new one. ‘The reality of one’s previous passion must be downplayed, minimized,
even altogether denied so that one may clear one’s slate in order to make room for a new and
genuine passion for someone else who is yet to come along’ (Jollimore 2009: 45). This rather
heartless form of romanticism nicely illustrates the way we construct our memories to fit ide-
ologies of emotion, which we interpret as laws of human nature. The very phenomenology of
being in love, in fact, induces a ‘denial of reality’ in the form of a conviction that one can only
genuinely love one person in a lifetime (Jollimore 2009: 41).
At the movie’s end, the protagonists have learned of the erasure of one another’s memories.
After some hesitation, in full consciousness of the likelihood of repeated disaster, they agree to start
over. This poses a conundrum similar to the one presented in Nietzsche’s thought experiment of
the Eternal Return (Nietzsche 1954: 101). In that famous passage, Nietzsche suggested that the
right attitude to an experience is to affirm it – to welcome the thought that this very moment will
recur again and again forever. From this and other texts, Jollimore extracts a complex thesis of
‘Nietzschean affirmation’, of which the most daring is that one cannot pick and choose among the
moments that have proved causally necessary and sufficient to bring about an existing situation.
In affirming a moment, one affirms one’s whole life. To do that, one ‘must accept that something
that has shameful or evil roots, or that will end in annihilation, failure, or pain, can nevertheless be
fully and wholeheartedly endorsed as good in the present moment’ ( Jollimore 2009: 51).
Affirmation, then, is an attitude that takes place in the present, and brings its object from the
past into the present. Implicit in the Nietzschean moral Jollimore draws from Eternal Sunshine,
159
Ronald de Sousa
is the thought that unpleasant memories should not be erased. We should, instead, ‘affirm’ the
errors in our past, as part of affirming the present and our resolve to live.
But there is an alternative to forgetting a past event, namely, regretting it. Is what one regrets
the same as that which one would rather forget? The question plays on a certain ambiguity in
the word ‘regret’. One can regret something good because it has ceased to be. This kind of
regret consists in wanting to retrieve something that is lost: it represents, in Janet Landman’s
striking phrase, ‘the persistence of the possible’ (Landman 1993). One can also regret something
bad in the sense of wishing it had never existed. Both forms of regret are factive, meaning that
one cannot logically regret, in either sense, something that never happened. But as Dorothea
Debus has shown, this does not mean that the emotion of regret (or any other past-directed
emotion) in itself constitutes a memory, either of the event itself or of the emotional response it
aroused at the time (Debus 2007). Rather, the present emotion is an attitude to the past event
and possibly also to the original emotion. It can distance itself from that past emotion or even
be antithetical to it, as when one feels ashamed at having been unkindly amused by someone’s
misfortune (Debus 2007: 762). Just as an ordinary memory is always a re-construction, so an
emotional memory is always a re-evaluation.
Can such a re-evaluation be correct or incorrect, rational or irrational? To answer this ques-
tion is to uncover an asymmetry between emotions of anticipation and emotions towards the
past. In the next section, I will suggest that while it is plausible to regard some future-oriented
emotions as more rational than others, it is difficult to find compelling reasons to ground norms
of rationality applicable to past-oriented attitudes. Specifically, I shall argue, there is no need for
‘global’ affirmation or regret.
3. Norms of regret
A certain tradition in philosophy insists on the gap between fact and value. Within any standard
emotion, there is a similar gap between the cognitive ground on which it is based and the atti-
tude it represents. The attitude is specific to the emotion’s formal object (thus fear is an attitude
essentially focused on the dangerous); but the instantiation of the formal object is never entailed
by the underlying facts of the situation (one might believe a situation dangerous without expe-
riencing fear). To be sure, it would seem ‘unintelligible’ for someone to regard the random
killing of a child with amusement rather than indignation, but the (in)appropriateness of either
attitude does not follow logically from the description of the facts. That is, at least, a logically
defensible position in line with Hume’s famous quip about the logical consistency of preferring
the destruction of the world to the pricking of my finger (Hume 1978: Bk II: iii, 2).
In practice, we regard some emotions as more rational than others, and view many situations
as calling unequivocally for one emotional attitude over others. That is the rationale for the
project of Rational-Emotive Behaviour Therapy (REBT) (Ellis and Joffe-Ellis 2011). REBT is
inspired by the Stoic motto that one should not care about anything we cannot change. Its main
goal is to achieve greater realism and rationality in behaviour, but it also aims to achieve axiologi-
cal rationality. Axiological rationality pertains to valuing, as strategic rationality pertains to action
and epistemic rationality to belief. Like those other modes of rationality, axiological rationality
is guided by meta-cognitive feelings of rightness, or doubt, or of knowing. Just as feelings of
pastness can stick to an imagined scenario, so a feeling of appropriateness can stick to an attitude.
When we are thinking of the future, it makes good sense to recommend some attitudes as
more rational than others. Here is an example inspired by Plato’s Philebus. Call it the ‘Philebus
principle’ (PP): ‘(PP) A pleasure of anticipation should be proportional in intensity to the anticipated
pleasure it relates to’ (de Sousa 2011: 13; original emphasis).
160
Memory and emotion
Although it might seem odd to think of Plato as a Darwinian, (PP) makes good biological
sense. The intensity of your pleasure of anticipation is likely to determine the tenacity of your pre-
sent motivation. If your present pleasure is uncorrelated with the future pleasure it motivates you
to pursue, then your present motivation will bear no settled relation to your future satisfaction.
Some attitudes to the future, then, seem more rational than others. And some emotional
responses to the past may be directly relevant to what we decide to do in the future. Complex
emotions, no less than the meta-cognitive feelings discussed by Joëlle Proust, serve to orient our
attention appropriately for what lies ahead. So do the memories of our immediate and distant
past. Again, memory and emotion work together. Forward-looking emotions prepare us for
imminent action. It is difficult to imagine what it would be like to plan for the future if you had
no attitudes, positive or negative, to various possible outcomes of actions. Merely being able to
discern possibilities requires a sense of the continuity of the immediate past – what was I about
to do? – with the present (what am I doing?) and the future (what am I now planning to do?).
Without episodic memory, the future, like the past, is a blank: ‘the lack of conscious aware-
ness of personal time encompasses both the past and the future’ (Tulving 1985: 4). The grim
imprisonment in the Now which afflicted amnesics such as Tulving’s patient K.C., Milner’s
patient H.M (Milner et al. 1968), or Clive Wearing (Wearing 2005), shows how meaningless
the immediate future is without an immediate past.
In sum, a memory may afford me a way of thinking of myself in the present; that in turn can
modify my intentions. But the memory’s influence on the future will likely be only a general
one. A memory is less likely than a forward-looking emotional attitude such as fear or anger to
trigger a specific action. When it does, that may serve to differentiate one specific emotion from
another: if I merely regret something that cannot be fixed, nothing follows about what I should
plan to do. By contrast, if what I feel is remorse, I may well be driven to some specific act of
contrition. Walking to Santiago de Compostela, perhaps.
Regret and remorse carry different implications for the future, though both are backward-
looking. In either case, however, like Joel and Clementine in the Eternal Sunshine movie, I might
be moved to want to get rid of them. That is just one way that a meta-level emotion will often
come to complicate the meaning of my past. In some cases, rather than walking to Santiago, I
may be moved to stamp out the superstitious attitude that caused me to feel I had sinned in the
first place.
How should we decide between those alternative responses? Can we justify any principle
of rationality pertaining to past-directed emotions in themselves? Or can such a principle be
derived only from the attitude’s relevance to the future? In some cases, it seems we can have
emotions that are purely backward-looking, and concern matters about which we can no longer
do anything. If we consider such backward-looking emotions on their own terms, without any
prudential motive deriving from their bearing on future behaviour, it does not seem that any
justification can be found for advocating one attitude over another.
Yet many poets as well as philosophers have expressed the opposite conviction. Consider, for
example, Dante’s expostulation: ‘There is no greater sorrow / Than to recall our time of joy /
In wretchedness’ (Dante Alighieri 2000: V-118–120). Against that, the French poet Alfred de
Musset accuses Dante of ‘insulting sorrow’, urging us rather to regard remembrance of happi-
ness as ‘truer than happiness’: ‘that fleeting instant was the whole of your life, not something
to regret!’; you should regret instead ‘your nights without hope and your days without light’
(Musset [1850] 1957: 409). Countless other texts in literature enjoin specific attitudes to the
past. When, for example, is it appropriate to feel regret?
I have in the past argued that we can rationally regret what we missed out on when getting
our first preference – providing that what we regret cannot be regarded as included in what
161
Ronald de Sousa
we did get. Only if the option that I didn’t choose belongs to another kind of value, as judged
by the qualia that made it the object of desire, can I reasonably regret it (de Sousa 1974). Tom
Hurka has attacked the value pluralism implicit in that argument. Hurka argued that a monistic
view of value is compatible with a diversity of qualitative experiences, some of which might be
desirable as such even though the value they conferred on the experience is the same (Hurka
1996). For my part, I still fail to see why I should regret some aspect of an event unless that
aspect was what I valued about it. Thus if two things differ only in the quantity of the same
value, it doesn’t seem to make sense to regret the lesser when I got the greater. But it now
seems to me that the dispute is idle: I no longer feel I can endorse the idea of rational regret in
either form.
Spinoza says that who regrets is ‘twice unhappy or twice impotent’ (Spinoza 1985: IV-54).
We are impotent to change the past; so an inability to control our unhappiness about it is
just another form of impotence. By contrast, Jay Wallace has recently argued that painful
retrospective emotions should be endorsed as ‘symptoms of valuing’ (Wallace 2013: 25).
Who is right?
Jay Wallace advocates ‘unconditional affirmation’ in something like the Nietzschean sense dis-
cussed in the section above. He construes affirmation as the contrary of regret. For Wallace, regret
is viewed as rational in virtue of its essential connection with valuing; but while one’s values
may dictate regret of specific actions now deemed wrong, ‘all-in’ regret should be precluded by
‘unconditional affirmation’ of everything that was causally responsible for my life as it is now.
Is Wallace right to recommend the choice of affirmation as opposed to a contrary attitude of
‘all-in regret’? I suggest that there is no rational answer. Wallace’s main argument rests on an
analogy between backward-looking emotions and intentions: ‘preferences regarding past states
of affairs could be thought of as a special kind of conditional commitment, where the relevant
condition is contrary to fact’ (Wallace 2013: 56). This is far-fetched. For it is not just by accident
but by necessity that those conditions will never be satisfied. The counterfactuals presupposed
by Wallace’s analogy become mind-boggling. Suppose for example that I regret the Tohoku
Tsunami. I must suppose
first, that the film of time might be unspooled, taking me back to March 11, 2011,
when the tsunami occurred . . . second, that there is something I might then have
done to bring it about that the tsunami did not lead to loss of life and property . . .
further . . . that there was some mechanism in place that could have been deployed . . .
and that it was in my power to activate the mechanism.
(Wallace 2013: 64)
In forming intentions for the future, we typically ignore the chaotic nature of life. In actuality,
we can never reliably foresee the long-term consequences of our choices. Precepts of pru-
dence are feeble but better than nothing. But when considering the past, our predicament is
much worse. About the past, choosing a preference requires us to make counterfactual supposi-
tions that admit of no non-arbitrary standards of correctness. All four of the counterfactuals in
Wallace’s thought experiment, for example, are pure fantasy. No cognitive basis for a rational
emotional response exists. Hence the hope of finding a rational solution to the question of
whether the past justifies all-in affirmation or regret is idle.
In short, nothing requires me to arrive at any ‘all-in’ verdict about the past. I can remem-
ber certain facts or aspects of events with pleasure, while deploring others. No requirement of
consistency can be defined for attitudes to the past.
162
Memory and emotion
4. Conclusion
Memory feeds our present emotions, the very structure of which are rooted in the past. And our
emotions give memories their point. Their mutual entanglement takes place at different levels of
analysis, and it is effected in a variety of ways, including mechanisms of monitoring and control
that work at the subpersonal micro-level. These may be shared or homologous with devices
that control simpler organisms. Other mechanisms consist in declarative memory and complex
emotions. Conversely, emotions both select and colour the view we take of our past. Our
emotions are composed of memories, and our memories are shaped by past and present emo-
tions. Memory, we might say, would be empty without emotion, and emotion blind without
memory. But however much we would like to set out rules governing attitudes to the past, I
conclude that any such rules would be arbitrary. Since it is a function of emotion to control the
relative salience of different modes of perception and information, our meta-emotional attitudes
cannot be justified in any non-circular way. To advocate a certain attitude to the past, whether
it be affirmation, or regret, is itself simply to be captured by just another arbitrary emotional
attitude. Nothing requires us to be choosing between totally unfathomable worlds.
Note
1 The ‘five organismic systems’ in question comprise characteristic neurophysiological processes, cognitive
content emerging from appraisals, conative dispositions or ‘action-tendencies’, subjective feelings, and
characteristic facial expressions.
Further reading
Dokic, Jérôme (2014) ‘Feeling the Past: A Two-Tiered Account of Episodic Memory’, Review of Philosophy
and Psychology, 5: 413–26.
Elaborates on the role of a meta-cognitive feeling of knowing in episodic memory.
Kensinger, Elizabeth A. and Daniel L. Schacter (2014) ‘Memory and emotion’, in Lewis, Michael, Haviland-
Jones, Jeannette and Feldman Barrett, Lisa (eds), Handbook of Emotion. 3rd edn. New York: Guilford Press,
pp. 601–17.
An up-to-date survey of research on the modes of influence of emotion on the character and quantity
of information remembered.
Proust, J. (2015) ‘The representational structure of feeling’, in T. Metzinger and J. Windt (eds), Open
MIND. Frankfurt am Main: MIND group.
Argues for a pervasive role of feeling, understood as involving a non-conceptual component affecting
motivation, in cognition and memory.
References
Bartlett, F. C. (1920) ‘Some experiments on the reproduction of folk-stories’, Folklore, 31(1): 30–47.
Capgras, J. (1923) ‘L’illusion des “sosies” dans un délire systématisé chronique’, Bulletin de la Société clinique de
médecine mentale, 11: 6–16. Retrieved from www.bium.univ-paris5.fr/histmed/medica/cote?epo1250.
Charland, L. C. (1995) ‘Emotion as a natural kind: Towards a computational foundation for emotion
theory’, Philosophical Psychology, 8(1): 59–84.
Dante, A. (2000) Inferno. R. Hollander and J. Hollander (eds.). New York: Anchor.
Debus, D. (2007) ‘Being emotional about the past: On the nature and role of past-directed emotions’,
Noûs 41(4): 758–78.
de Sousa, R. (1974) ‘The good and the true’, Mind, 83: 534–51.
163
Ronald de Sousa
de Sousa, R. (1984) ‘The natural shiftiness of Natural Kinds’, Canadian Journal of Philosophy, 14: 561–80.
de Sousa, R. (1987) The Rationality of Emotion. Cambridge, MA: MIT Press, a Bradford Book.
de Sousa, R. (2008) ‘Epistemic feelings’, in Brun, G., Do uo lu, U. and Kuenzle, D. (eds), Epistemology
and Emotions. Aldershot: Ashgate.
de Sousa, R. (2011) Emotional Truth. New York: Oxford University Press.
Dokic, J. (2014) ‘Feeling the past: A two-tiered account of episodic memory’, Review of Philosophy and
Psychology, 5: 413–26.
Ellis, A. and Joffe-Ellis, D. (2011) ‘Rational-emotive therapy’, in Corsini, R. J. and Wedding, D. (eds),
Rational Emotive Behavior Therapy. Washington, DC: American Psychological Association.
Freud, S. (1915) ‘Observations on transference love’, in Tyson, A. and Strachey, J. (eds), Standard Edition
of the Complete Psychological Works, vol. 12. London: Hogarth Press, pp. 157–73.
Frijda, N. (2016) ‘The evolutionary emergence of what we call “emotions”’, Cognition and Emotion, 30(4):
609–20.
Goor, P., Olivier, A., Quesney, L. F., Andermann, F. and Horowitz, S. (1982) ‘The role of the limbic
system in experimental phenomena of temporal lobe epilepsy’, Annals of Neurology, 12(2): 129–44.
Griffiths, P. E. (1997) What Emotions Really Are: The Problem of Psychological Categories. Chicago, IL:
University of Chicago Press.
Hirstein, W. (2005) Brain Fiction: Self-Deception and the Riddle of Confabulation. Cambridge, MA: MIT Press.
Hume, D. (1978) A Treatise of Human Nature, 2nd edn, revised with notes by L. A. Selby-Bigge;
P. H. Nidditch (ed.). Oxford: Oxford University Press.
Hurka, T. (1996) ‘Monism, pluralism, and rational regret’, Ethics, 106(3): 555–75.
Jollimore, T. (2009) ‘Miserably ever after: Forgetting, repeating and affirming love in Eternal Sunshine of the
Spotless Mind’, in Grau, C. (ed.), Eternal Sunshine of the Spotless Mind. New York: Routledge, pp. 31–60.
Kensinger, Elizabeth A. (2009) ‘Remembering the details: Effects of emotion’, Emotion Review, 1(2): 99–113.
Klein, S. B. (2015) ‘What memory is’, WIREs Cognitive Science, 6: 1–38.
van der Kolk, B. A. and Fisler, R. (1995) ‘Dissociation and the fragmentary nature of traumatic memories:
Overview and exploratory study’, Journal of Traumatic Stress, 8(4): 505–25.
Kundera, M. (1984) The Unbearable Lightness of Being. New York: Harper and Row.
Landa, A., Bossis, A. P., Boylan, L. S. and Wong, P. S. (2012) ‘Beyond the unexplainable pain: Relational
world of patients with somatization syndromes’, Journal of Nervous and Mental Disease, 200(5): 413–22.
Landman, J. (1993) Regret: The Persistence of the Possible. New York: Oxford University Press.
Loftus, E. F. and Loftus, G. R. (1980) ‘On the permanence of stored information in the human brain’,
American Psychologist, 35(5): 409–20.
Luborsky, L. and Crits-Christoph, P. (1998) Understanding Transference: The Core Conflictual Relationship
Theme Method, 2nd edn. Washington, DC: American Psychological Association.
Marr, D. (1982) Vision. San Francisco, CA: Freeman.
Michaelian, K. (2015) ‘Opening the doors of memory: Is declarative memory a natural kind?’ WIREs
Cognitive Science, 6: 475–82.
Milner, B., Coricin, S. and Teuber, H.-L. (1968) ‘Further analysis of the hippocampal amnesic syndrome:
14-year follow-up study of H.M.’, Neuropsychologia, 6: 215–34.
Musset, A. D. (1957/1850) ‘Souvenir’, in Poésies Complètes. Paris: Pléiade, pp. 404–9.
Nietzsche, F. (1954) The Portable Nietzsche. W. Kaufmann (ed., trans. and intro.). New York: Viking Press.
Nietzsche, F. (1967) On the Genealogy of Morality. M. Clark and A. J. Swensen (trans. and notes).
Indianapolis, IN: Hackett.
Pacherie, E. (2008) ‘Perception, emotions and delusions: The case of the Capgras delusion’, in Baynes, T.
and Fernández, J. (eds), Delusions and Self-Deception: Affective Influences on Belief Formation. Hove:
Psychology Press, pp. 107–26.
Perry, John (1979) ‘The problem of the essential indexical’, Noûs, 3(1): 3–21.
Proust, J. (2015) ‘The representational structure of feeling’, in Metzinger, T. and Wundt, J. (eds), Open
MIND. Frankfurt am Main: MIND group.
Proust, M. (1966) Swann’s Way. C. S. Moncrieff, and T. Kilmartin (trans.). New York: Vintage.
Rorty, A. (1980) ‘Explaining emotions’, in Rorty, A. (ed.), Explaining Emotions. Berkeley, CA: University
of California Press.
Rosch, E. and Lloyd, B. (eds) (1979) Cognition and Categorization. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Russell, B. (1921) The Analysis of Mind. London: Allen & Unwin.
164
Memory and emotion
Sacks, O. (2010) ‘Face-blind: Why are some of us terrible at recognizing faces?’ New Yorker, August 30:
43–9.
Scherer, K. R. (2005) ‘What are emotions? And how can they be measured?’ Social Science Information,
44(4): 695–729.
Spinoza, B. D. (1985) ‘Ethics’, in Curley, E. (ed. and trans.), The Collected Works of Spinoza. Princeton, NJ:
Princeton University Press.
Tulving, E. (1972) ‘Episodic and semantic memory’, in Tulving, E. and Donaldson, W. (eds), The
Organization of Memory. New York: Academic Press, pp. 381–402.
Tulving, E. (1985) ‘Memory and consciousness’, Canadian Psychology, 26(1): 1–12.
Varela, F., Maturana, H. and Uribe, R. (1974) ‘Autopoiesis: The organization of living systems’, BioSystems,
5: 187–96.
Wallace, R. J. (2013) The View from Here: On Affirmation, Attachment, and the Limits of Regret. New York:
Wearing, D. (2005) Forever Today: A Memoir of Love and Amnesia. London: Corgi.
Wheatley, T. and Haidt, J. (2005) ‘Hypnotic disgust makes moral judgments more severe’, Psychological
Science, 16(10): 780–84.
Zajonc, R. B. (2000) ‘Feeling and thinking: Closing the debate over the independence of affect’, in
Forgas, J. P. (ed.), Feeling and Thinking: The Role of Affect in Social Cognition. Cambridge: Cambridge
University Press, pp. 31–58.
165
PART IV
Memory and the self

13
MEMORY AND PERSONAL
IDENTITY
Shaun Nichols
1. Memory and the philosophy of personal identity

In modern philosophy, memory plays a central role in discussions of personal identity. On one
view, memory plays a role in what makes us the same person across time. On an alternative
view, memory instead serves as evidence of identity with a past person.
Memory as constitutive of personal identity

In his systematic treatment of personal identity, Locke pronounces that “as far as this conscious-
ness can be extended backwards to any past Action or Thought, so far reaches the Identity of
that Person; it is the same self now it was then.” This passage is often interpreted as promoting
a Memory Criterion of personal identity:
Memory Criterion: A at t2 is the same person as B at t1 if A at t2 can remember (from

the first-person perspective) an experience of B at t1.
Locke has a complex theory of memory (see Chapter 39), and there is controversy about
whether Locke is really committed to the Memory Criterion account of personal identity (see,
e.g., Atherton 1983; Behan 1979; Weinberg 2011). However, the Memory Criterion has played
a key role in theorizing about personal identity, and has been the inspiration for contemporary
theories of personal identity, so it is a natural starting point.
Objections to the memory criterion

The Memory Criterion has been challenged on several grounds, but I’ll focus on three
key objections. First is the problem of transitivity (see, e.g., Reid 2002: 276). Identity is
transitive – if A = B and B = C, then A = C. But memory, with its manifest vicissitudes, isn’t.
In Reid’s famous case, a General remembers capturing a flag as a Brave Officer, who in turn
169
Shaun Nichols
remembers being beaten as a child, but the General doesn’t remember the childhood beating.
In this case, the Memory Criterion yields the result that the General is the Officer, and the
Officer is the Child, but the General isn’t the Child. The foregoing failure of transitivity is a
linear case, since it proceeds by the gradual loss of memories across the lifespan. But we can
also get non-linear violations of transitivity. Imagine Mark in 1975 makes a game-winning hit
in Little League. In 1994, while inebriated in a sports bar, Mark boasts incessantly about his
game-winning hit in 1975. This direct memory makes boring-Mark1994 the same person as
Mark1975. However, boring-Mark1994 drinks so much that he blacks out and never recov-
ers any memories from that night. The next day, hungover-Mark still remembers Mark1975’s
hit. But he doesn’t remember anything from boring-Mark of the night before. Transitivity
fails here in a non-linear way: Mark1975 is boring-Mark, and Mark1975 is hungover-Mark,
but hungover-Mark isn’t the same person as boring-Mark.
A second familiar objection is circularity, raised by Butler (1736: 100). On this objec-
tion, if A can remember an experience of B, it follows trivially that A is the same person as B.
For if A wasn’t the same person as B, then at best he could only have a false memory of having an
experience of B. Thus, the objection goes, there is no substance to the proposal that memory
is sufficient for personal identity—memory requires identity in the first place. Butler calls this a
“wonderful mistake.”
A final concern is easy identity: for certain cases, the memory criterion makes it too easy to
be identical with a past person (e.g., Schechtman 2005: 12; Shoemaker 2015: 9–10). A single
memory of having an experience as a past person is sufficient for identity with that past person.
Thus according to the Memory Criterion, it seems that if I had a memory trace from Keith
Moon implanted in my brain such that I suddenly have a memory of blowing up a drum set on
The Smothers Brothers Comedy Hour, this would make me Keith Moon. But it’s very counterintui-
tive that this single psychological insertion would make me the same person as Keith Moon. I
probably couldn’t expect to be invited to the next reunion of The Who.
The Memory Criterion revisited

Philosophers have attempted to amend the simple Memory Criterion to address these problems.
For the transitivity problem, philosophers appeal to the fact that typically there are immedi-
ate memory connections from day to day, and one can think of these connections as chaining
together to secure a series of connected mental states. Grice (1941) develops the idea by using
the notion of the “total temporary” mental state, which includes all of the mental states in a
given person’s mind at a time. Within one person, such total states are linked into a temporal
series, he suggests, by memories. As a first pass, Grice suggests that in a single person, every
total state “will contain as an element a memory of some experience which is an element in the
temporally preceding member of the series” (1941: 342). Grice thus suggests, roughly, that if I
am the same person as someone from ten years ago, then there is a series of total states, such that
my total state now is linked by direct memory to the preceding state, which is itself linked to the
preceding state by direct memories, and this series stretches back for a decade. If such a temporal
series is sufficient for personal identity, then we can exclude linear transitivity violations like
the Brave Officer. For the temporal series only requires that there is a memory chain from the
General to the whipped boy—it doesn’t require that there is a direct memory.
For the circularity objection, philosophers have suggested that we can simply define a notion,
“quasi-memory,” that subsumes memory but does not presuppose identity (e.g., Shoemaker
1970). Parfit characterizes quasi-memory as follows:
170
Memory and personal identity
I have an accurate quasi-memory of a past experience if:

(1) I seem to remember having an experience;
(2) someone did have this experience;
(3) my apparent memory is causally dependent, in the right kind of way, on that past
experience.
(Parfit 1984: 220)
Obviously, there is a lot packed into this third condition on quasi-memory. But by drawing
on chains of quasi-memories, we might formulate an account of personal identity that draws
on something like the Memory Criterion without presupposing identity. Real memories can
then be defined as a proper subset of quasi-memories—a person’s real memories are quasi-
memories of her own past experiences (see, e.g., Shoemaker 1970: 271; Parfit 1984: 220; but
cf. Schechtman 1990).
As for the concern that the Memory Criterion allows for easy identity, theorists tend to stipu-
late that a single memory connection isn’t sufficient for identity (e.g., Parfit 1984: 206). One
option here would be to maintain that one must have a large set of direct quasi-memories to be
identical with a past person. That would mean that having a memory of blowing up a drum kit
on The Smothers Brothers isn’t enough to make me Keith Moon.
Although the problem of easy identity can be addressed simply by stipulating that there
need to be many direct memory connections at each link, the heirs to the Memory Criterion
have tended to make a more far-reaching revision to the view. They hold that memory is not
the only relevant connection for personal identity. The kinds of psychological connections
that matter to identity include beliefs, desires, and character traits (e.g., Parfit 1984: 204). This
expansion is intuitively plausible, since we do care about more than our memories, when
it comes to personal identity. In addition, by expanding the class of relevant psychological
connections, one might also address the non-linear failures of transitivity. Even though there
are no memory connections between boring-Mark and hungover-Mark, there can still be
many other kinds of direct connections between them, including beliefs, desires, and char-
acter traits.
The most prominent revised account that is heir to the Memory Criterion is the “Psychological
Continuity” view (described by Parfit 1984: 205ff.):
Psychological Continuity Criterion: A at t2 is the same person as B at t1 if A is

Psychologically Continuous with B.
For A at t2 to be psychologically continuous with B at t1 is for there to be “overlapping chains

of strong connectedness” between them (Parfit 1984: 206). And strong connectedness requires
that there be many direct connections. On Parfit’s proposal, strong connectedness from one day
to the next requires that the number of direct connects be “at least half the number in most
actual lives” (1984: 222). So, in order for me to be the same person from one day to the next,
I must enjoy strong connectedness. And I can retain personal identity across long stretches of
time, so long as this day-to-day connectedness forms an unbroken chain.
Although Psychological Continuity theorists allow that memory isn’t the only relevant
psychological state, memory still plays a key role. Indeed, the theory is plausibly driven by
the intuition that the Memory Criterion captures something important about personal iden-
tity. The revisions that we have seen largely assume that the Memory Criterion is basically
right and just needs some patches to fix the holes revealed by the objections. We solve the
171
Shaun Nichols
problem of transitivity by stipulating that overlapping chains suffice for identity; we solve
the problem of presupposition by defining a notion, quasi-memory, that doesn’t presuppose
identity, and we solve the problem of easy identity by maintaining that many connections are
required for identity.
Memory as evidence of personal identity

The previous theories take the psychological connections to constitute personal identity. It is in
virtue of the fact that I have many direct psychological connections, such as quasi-memories,
with a past individual that I am that past individual. The problems with the simple Memory
Criterion are largely addressed by patching the theory to insulate it from the objections. But
one might instead think, as Reid did, that it’s a colossal mistake to regard memory as constituting
personal identity. Rather than using memory to ground the metaphysics of personal identity,
one can take memory to have a key epistemic role. Having an apparent memory might provide
evidence of personal identity with a past individual.
The view that memory of experiences provides evidence of personal identity fits naturally
with animalist views, according to which the self persists as long as the organism does (e.g.,
Olson 1997). It’s plausible that experience memories very often reflect past experiences had by
the organism. I remember an experience of being on a plane last week, and that’s good evidence
that this organism was on a plane last week. Thus, if the self is the organism, then I have good
evidence that I—my self—was on a plane last week.
Memory can play a similar role if we adopt the view that the self is a simple substance (which
might be thought of as a soul). For instance, on Reid’s view, the self is a substance (rather than,
say, a process or a state) that has psychological states and executes decisions; the self is simple inso-
far as it has no parts (see Copenhaver 2014). On this view, the self persists as long as that simple
substance persists. If I remember the experience of being on a plane last week, and that experi-
ence was had by my simple substance, then that’s good evidence that my simple substance was
on a plane last week; thus if the self is the simple substance, it follows that I have good evidence
that I was on a plane last week. Note that if memory is taken as mere evidence of personal iden-
tity, then there is no need for the kind of strong connectedness demanded by the Psychological
Continuity theory. A single memory might provide evidence that this organism or this simple
substance graduated from high school decades ago.
2. Memory and judgments of psychological continuity

When people make judgments about personal identity, it’s no surprise that memory is critical.
Insofar as we are making judgments about a past self, we rely on memory. However, the role of
memory in judgments of personal identity is complex.
Psychologists distinguish between two fundamental types of long-term declarative memory.
Episodic memories are experience memories. For instance, I remember the tedious experience
of doing the laundry yesterday. Semantic memory, by contrast, is knowledge of facts, like the
knowledge that bats are mammals. One can have semantic memories about oneself of course. If
you know where you were born, this counts as semantic memory, unless you have the miracu-
lous ability to remember the experiences surrounding your birth (Tulving 1972).
As we’ve seen, psychological continuity depends on connections between psychological
features, including beliefs, character traits, and memories. When we make judgments of such
continuity, we must rely on knowledge of those psychological features. That is, insofar as we
172
track psychological continuity and connectedness, we must represent the psychological features
that are so connected. It’s no surprise that people remember their beliefs and their character
traits. But how is that information represented?
Mechanism
When participants are given a list of trait adjectives and asked whether they possess those traits,
people respond consistently over time, and their responses correlate with third-person reports
from friends and relatives (see, e.g., Gough 1960). One explanation for how one knows one’s
traits is that one recalls various experiences one has had and then uses those experiences as evi-
dence concerning one’s traits. So, for instance, if you ask whether I am timid, I might try to
recall various social occasions and consider how I reacted in those cases. Alternatively, I might
just have a store of knowledge for quick retrieval that includes information about my traits. Such
a store would plausibly be housed in semantic memory, along with other facts about myself, like
my birthday and my birthplace.
Evidence on amnesia patients indicates that knowledge of one’s traits is indeed at least partly
stored in semantic memory. Trait self-knowledge is retained even in cases of catastrophic amne-
sia. Of particular relevance, patients retain trait self-knowledge even when they have virtually
no episodic memory. Patient D.B. suffered anoxia following a heart attack, and this produced
profoundly impaired episodic memory. Yet his reports of his personality traits were consistent
across time and correlated with the ratings provided by his daughter (Klein et al. 2002). This
robustness of trait self-knowledge was also found in K.C., who suffered brain injury from a
motorcycle accident. K.C. apparently had a complete absence of episodic memory. When he
was asked to remember any event from his life, K.C. reported that it was just blank (Tulving
1985: 4). Despite the absence of episodic memory, K.C.’s reports of his traits correlated with
the reports offered by his mother (Tulving 1993). The fact that trait self-knowledge is retained
even in patients who lack episodic memory provides evidence that semantic memory includes
knowledge of one’s own traits.
Content
Character traits come in different varieties. Some traits concern temperament (e.g., shy); some
concern abilities (e.g., creative); some concern moral dispositions (e.g., kind). In addition, psy-
chological continuity includes features that aren’t character traits, including, of course, a bank of
episodic memories poised for retrieval. Are some of these features privileged over others when
it comes to judgments of personal identity? Presumably some memories are more important to
connectedness than others. My memories of the color of my previous car cannot be as important
to my identity as my memories of the birth of my first child. Are memories the most important
factor when compared to traits?
In several studies, memory has been pitted against moral character traits. There is some
reason to suspect that memories would be more important than moral traits. It has seemed to
many philosophers (e.g., Parfit 1984: 300) and psychologists (e.g., Nelson and Miller 1995)
that personal identity depends on our distinctive traits. And memories, unlike moral traits, are
distinguishing. There are lots of generous people. But only one guy remembers the birth
of my first child, from a father’s perspective. Despite the fact that memories are obviously
more distinctive than moral traits, across a wide range of studies, change in memory mattered
significantly less than change in morals (Prinz and Nichols forthcoming; Strohminger and
173
Shaun Nichols
Nichols 2014). For instance, when participants are asked to imagine the loss of a particular
trait and to assess the extent to which that change would make a change in personal identity,
participants judged changes in moral traits (e.g., honesty) to make more of a difference to
personal identity than changes in memories (e.g., loss of cherished memories of time with
parents). This basic effect held regardless of whether participants were considering questions
about their own personal identity or that of another person. In a subsequent study, fam-
ily members of patients with neurodegeneration (Alzheimer’s, amytotrophic lateral sclerosis
(ALS), and frontotemporal dementia) were asked about whether their loved one had changed
on a wide range of psychological features. Family members were also asked the extent to
which their loved one was the same person. The results indicated that change in moral
traits made an enormous difference to perceived identity, but amnesia did not matter at all
(Strohminger and Nichols 2015). This work is still in the earliest stages, but it suggests that
when people evaluate psychological continuity in assessing personal identity, memory plays
a subsidiary role to morality.
3. Episodic memory and sense of personal identity

Semantic memory contains knowledge of traits. However, it is episodic memory that is typically
invoked in philosophical discussions of personal identity.
Episodic memory as carrying conviction of personal identity

Episodic memory is typically characterized as carrying information about the what, where, and
when of an experience. I remember having breakfast, at my table, this morning at 7. In addition,
though, many philosophers and psychologists maintain that this form of memory also requires
that one remembers the experience as happening to oneself. The memory has to present the
experience as having happened to me. We find an expression of this in Reid:
The remembrance of a past event is necessarily accompanied with the conviction of

our own existence at the time the event happened. I cannot remember a thing that
happened a year ago, without a conviction as strong as memory can give, that I, the
same identical person who now remember that event, did then exist.
(Reid 1850: 202)
We find the view in William James, who wrote, “Memory requires more than the mere dat-
ing of a fact in the past. It must be dated in my past. In other words, I must think that I directly
experienced its occurrence” (1890/1950: 650). And it is a staple of contemporary theorizing
about episodic memory. Harlene Hayne and colleagues summarize the prevailing view about
episodic memory as holding that an episodic memory “is accompanied by conscious awareness
that the event happened to ‘me’ or will happen to ‘me’ that does not accompany retrieval of
other kinds of memories” (Hayne et al. 2011: 344).
While it is true that episodic memory can bring a palpable sense of personal identity, it’s
unclear whether that sense of identity is “necessarily accompanied” by experience memory.
First, it’s plausible that experience memories are present in animals that lack a concept of self.
For instance, there’s good reason to think rats can in fact form experience memories (see, e.g.,
Crystal et al. 2013). But there’s little reason to think that the rat’s memories involve an awareness
that the event happened to “me.” (And, indeed, neuroscientists who work on rat memory often
characterize episodic memory without explicit representation of self—e.g., Hasselmo 2011.)
174
If that’s right, then there must be some additional contribution that leads us humans to get the
sense of personal identity from experience memory.
Second, there are human cases in which experiences are remembered while the person disa-
vows a sense of identity with the experiencer. At age 43, R.B. was in a bicycle accident that
caused significant head trauma. After the injury, R.B. was able to recall episodes from his past,
but he reported that it didn’t seem like the experiences happened to him. For instance, he said
of one memory:
I could clearly recall a scene of me at the beach in New London with my family as a
child. But the feeling was that the scene was not my memory. As if I was looking at
a photo of someone else’s vacation.
In another case, he described a memory from when he was at school:
I can picture the scene perfectly clearly . . . studying with my friends in our study
lounge. I can “relive” it in the sense of re-running the experience of being there. But it
has the feeling of imagining, [as if] re-running an experience that my parents described
from their college days. It did not feel like it was something that really had been a part
of my life. Intellectually I suppose I never doubted that it was a part of my life . . . But
that in itself did not help change the feeling of ownership.
Thus, R.B. seemed able to retrieve memories of past experience, but those retrieved experi-
ence memories lacked the sense of having happened to him (Klein and Nichols 2012). Similar
reports occur in depersonalization disorder, one symptom of which is “The subjective feeling
of not being able to recall things (e.g., memory episodes) or having the feeling that the person
was not part of the episode” (Sierra and Berrios 2001: 631). For instance, one patient with this
symptom is described as follows: “When he recalled events in his life, he felt as though he was
‘not in them’” (Sierra and Berrios 2001: 631). Indeed, the depersonalization of memories forms
part of the standard measure of the disorder, as reflected in the following item: “I feel detached
from memories of things that have happened to me—as if I had not been involved in them”
(Sierra and Berrios 2000: 162).
Episodic memory and psychological continuity

The sense of personal identity generated by episodic memory does not depend on strong connect-
edness. Think about some experience you recall from your youth, like your first kiss. In my case,
I am psychologically very different from that adolescent, sufficiently different that strong connect-
edness doesn’t obtain. However, that doesn’t interfere with my sense of personal identity with
that past individual—the memory still gives me the distinct sense that *I* experienced that kiss.
Indeed, there is reason to think that the sense of personal identity can even obtain in the
absence of psychological continuity. This is perhaps most striking in cases of past-life beliefs
(White et al. 2016). Many people report having an episodic memory from a past life and that
this (apparent) memory convinces them that they existed in a past life, even while they don’t
seem to think that there is an overlapping chain of strong connections that stretches all the way
into that past life. In these cases, it seems, episodic memory leads to a belief in personal identity
even in the absence of a belief in psychological continuity.
Thus, past-life beliefs indicate that one can have the sense of personal identity even when
one doesn’t think that there is psychological continuity (defined in terms of overlapping chains
175
Shaun Nichols
of strong connectedness). Past-life beliefs plausibly are delusional—false memories. But someone
might have a memory that is not at all false, despite the absence of psychological continuity.
Consider H.M., who had much of his medial temporal lobes removed to alleviate his epileptic
seizures. The surgery alleviated his seizures, but it had catastrophic effects on H.M.’s mind. Most
strikingly, following his surgery, H.M. had profound anterograde amnesia—he could no longer
remember any new experiences for more than a couple of minutes. It’s plausible that pre-operative
H.M. was not strongly connected with post-operative H.M. In addition to his extensive antero-
grade amnesia, H.M. also suffered retrograde amnesia for several years leading up to the operation.
And after surgery, H.M. showed abnormally low emotional expression and lack of motivation.
Hence the person who went under surgery was not Strongly Connected with the person who
woke from surgery. Thus, as far as the Psychological Continuity theory goes, H.M. died on the
operating table and a new, deeply impaired, H.M. was born.
Despite the loss of Psychological Continuity, H.M. liked to recount a few episodes from his
youth. In his biography of H.M., Hilts writes:
[H.M.] tells of living in South Coventry, Connecticut, where he could go behind the
house to shoot guns. He had a rifle, “One with a scope!” he says, always enthusiastic
at this point in his story. “And I had handguns too! A .38 and a .32.”
(Hilts 1995: 138)
H.M.’s recollections presumably count as real memories. They reflect genuine experiences. Like
other episodic memories, the memories seem to carry the sense of personal identity. Witness
H.M.’s excitement—“I had handguns too!” But, as we’ve seen, H.M. doesn’t have Psychological
Continuity with the person who went under surgery. So, on the Psychological Continuity
account, pre- and post-surgery H.M. are different persons. As a result, when post-surgery H.M.
has a sense of personal identity with a boy playing with guns, it seems that we should say that
these memory shards present H.M. with an illusion of identity. This is a rather counterintuitive
result—a normal memory generates an illusion of identity. But the Psychological Continuity
theorist might just bite the bullet and accept that this is one of the substantive implications of the
account. This would also give the Psychological Continuity advocate another kind of response
to the charge of circularity. Recall that Butler charged the Memory Criterion with a “wonder-
ful mistake” insofar as memory presupposes identity, and this provoked the innovation of the
notion of “quasi-memory” as a way to rescue the Memory Criterion. But on the Psychological
Continuity view, it seems that memory in fact doesn’t presuppose genuine identity. There are
cases of normal memories in which a person remembers someone else’s memory. H.M.’s expe-
rience memory with the guns isn’t a false memory in any normal way. It’s just that memory, in
its normal functioning, delivers memories of experiences had by other selves.
Representations of self
As we saw (in Section 2), people have a rich knowledge of their traits. This knowledge is at least
partly stored in semantic memory. And some traits are regarded as more important to the self
than others—morals matter more than memory. However, the sense of personal identity given
by episodic memory suggests that, in such recollections, people are not representing the self as
a set of traits. Even if I know that I have a set of traits, when I recollect my past experience, the
recollection is not represented as a cluster of traits had that experience. When I recall the experience
of swimming last summer, I don’t represent this as a cluster of traits that is psychologically continuous
with the current traits had an experience of jumping into cold water. So although we do often think of
176
ourselves in terms of trait clusters, the representation of self in episodic memory is not given as
a trait cluster.
If the representation of self in episodic memory isn’t given as a set of traits, what is the repre-
sentation? Here, it’s useful to turn back to philosophy. We have a way of thinking of ourselves
in terms of an “essential indexical” rather than a description. To adapt an example from John
Perry, if Fred thinks Fred is about to be hit by a boulder, that won’t lead Fred to dodge unless he also
thinks I am Fred. That is, Fred won’t dodge unless he has a thought with the form I am about to
be hit by a boulder (Perry 1977: 494; Perry 1979). Second, it seems theoretically possible to wake
up with total amnesia, losing knowledge of all one’s traits and yet still sensibly wonder “where
am I?” (cf. Perry 1977). The amnesia case indicates that knowledge of traits is not necessary to
have these kinds of I-thoughts, and the boulder case indicates that knowledge of traits isn’t suf-
ficient to have these kinds of I-thoughts. This kind of description-independent representation
I is plausibly what we find implicated in episodic memory. That would explain why we can
get the sense of personal identity from episodic recollections even in the express absence of a
representation of trait continuity (Nichols 2014).
The prevailing view in contemporary memory theory holds that episodic prospection—
thinking about future experience—involves some of the same processes as recollecting past
experience (e.g., De Brigard et al. 2013; Suddendorf and Corballis 2007; Wheeler et al. 1997;
see Chapter 19). One bit of evidence for this view is that patients with deficits in episodic
memory typically show related deficits in episodic prospection (see, e.g., Szpunar 2010). In
addition, the phenomenology of episodic prospection is similar in important respects to the phe-
nomenology of episodic memory (e.g., D’Argembeau and Van der Linden 2004). It’s plausible
that one commonality between episodic recollection and episodic foresight is in how the self is
represented. When I remember an experience from my youth, my knowledge of the vast differ-
ence in traits doesn’t interfere with remembering the experience as happening to me. Similarly,
when I engage in episodic prospection and think about myself having a future experience, like
going to my child’s graduation, the fact that I will have different traits at that point does not
interfere with imagining myself having the experience. Just as episodic retrospection generates a
sense of identity that is oblivious to trait changes, so too episodic prospection seems to allow us
to imagine future experiences without any attention to trait differences.
This gives a psychological explanation for the natural reaction to a famous case from Bernard
Williams (Williams 1970). Williams says that if I think that I am going to have a complete break
in psychological continuity followed by torture, it’s still reasonable to fear the torture. Indeed,
most people seem to agree that in such a case, I will feel the pain (Nichols and Bruno 2011). In
this kind of thought experiment, Williams is inviting us to engage in episodic prospection. As a
result, the representation of self that is invoked is not the chunky representation of trait clusters,
but the description-independent representation, I. That representation can be projected into
future experience without any commitment to psychological continuity.
4. Conclusion
Many philosophers maintain that memory is part of what constitutes personal identity. It’s because
I retain memories and traits that I am the same person from day to day. A disruption in that con-
tinuity of memory and traits would mean a break in personal identity. Other philosophers deny
that memory is any part of what constitutes identity. Instead, memory might provide evidence
of identity with a past person. Episodic memory for past experiences seems to provide a sense of
personal identity that conforms to the latter view. Episodic memory can give one the sense of
having had a past experience even if the past experiencer is psychologically discontinuous with the
177
Shaun Nichols
present self. This doesn’t, of course, refute the view that personal identity is given by psychological
continuity. But it does suggest that the psychological continuity view is at odds with the sense of
personal identity delivered by memory.
Acknowledgment
Thanks to Sven Bernecker, Rebecca Copenhaver, Felipe De Brigard, and David Shoemaker for
very helpful comments on an earlier draft.
Related topics
• Locke and Reid on remembering
Further reading
Roach, R. (2006) “Defence of Quasi-Memory,” Philosophy, 81: 323–55.
Schechtman, M. (1996) The Constitution of Selves. Ithaca, NY: Cornell University Press.
Tulving, E. (1983) Elements of Episodic Memory. New York: Oxford University Press.
References
Atherton, Margaret (1983) “Locke’s Theory of Personal Identity,” Midwest Studies in Philosophy, 8: 273–94.
Behan, D. (1979) “Locke on Persons and Personal Identity,” Canadian Journal of Philosophy, 9: 53–75.
Butler, Joseph (1736) “Of Personal Identity,” in The Analogy of Religion, reprinted in John Perry (ed.),
(1975), Personal Identity. Berkeley, CA: University of California Press, pp. 99–105; page references are
to the reprinted version.
Copenhaver, Rebecca (2014) “Reid on Memory and Personal Identity,” The Stanford Encyclopedia of
Philosophy (Winter 2014 edn.), Edward N. Zalta (ed.), http://plato.stanford.edu/archives/win2014/
entries/reid-memory-identity/.
Crystal, J. D., Alford, W. T., Zhou, W., and Hohmann, A. G. (2013) “Source Memory in the Rat,”
Current Biology, 23(5): 387–91.
D’Argembeau, A. and Van der Linden, M. (2004) “Phenomenal Characteristics Associated with Projecting
Oneself Back into the Past and Forward into the Future: Influence of Valence and Temporal Distance,”
Consciousness and Cognition, 13: 844–58.
De Brigard, F., Addis, D., Ford, J., Schacter, D., and Giovanello, K. (2013) “Remembering What Could
Have Happened: Neural Correlates of Episodic Counterfactual Thinking,” Neuropsychologica, 51(12):
2401–14.
Gough, H. G. (1960) “The Adjective Check List as a Personality Assessment Research Technique,”
Psychological Reports, 6(1): 107–22.
Grice, H. P. (1941) “Personal Identity,” Mind, 50: 330–50.
Hayne, H., Gross, J., McNamee, S., Fitzgibbon, O., and Tustin, K. (2011) “Episodic Memory and Episodic
Foresight in 3- and 5-Year-Old Children,” Cognitive Development, 26(4): 343–55.
Hasselmo, M. (2011) How We Remember: Brain Mechanisms of Episodic Memory. Cambridge, MA: MIT Press.
Hilts, P. (1995) Memory’s Ghost. New York: Simon & Schuster.
James, W. (1890/1950) Principles of Psychology, vol. 1. Mineola, NY: Dover.
Klein, S. B. and Nichols, S. (2012) “Memory and the Sense of Personal Identity,” Mind, 121(483): 677–702.
Klein, S. B., Rozendale, K., and Cosmides, L. (2002) “A Social-Cognitive Neuroscience Analysis of the
Self,” Social Cognition, 20: 105–35.
Locke, J. (1690/2009) An Essay Concerning Human Understanding. New York: WLC Books.
Nelson, L. J. and Miller, D. T. (1995) “The Distinctiveness Effect in Social Categorization: You Are What
Makes You Unusual,” Psychological Science, 6(4): 246–9.
178
Nichols, S. (2014) “The Episodic Sense of Self,” in J. D’Arms and D. Jacobson (eds.), Moral Psychology and
Human Agency. Oxford: Oxford University Press.
Nichols, S. and Bruno, M. (2010) “Intuitions about Personal Identity: An Empirical Study,” Philosophical
Psychology, 23(3): 293–312.
Olson, E. (1997) The Human Animal: Personal Identity without Psychology. Oxford: Oxford University Press.
Parfit, D. (1984) Reasons and Persons. Oxford: Oxford University Press.
Perry, John (1975) Personal Identity. Berkeley, CA: University of California Press.
Perry, J. (1977) “Frege on Demonstratives,” The Philosophical Review, 86(4): 474–97.
Perry, J. (1979) “The Problem of the Essential Indexical,” Nous, 13(1): 3–21.
Prinz, J. and Nichols, S. (in press) “Diachronic Identity and the Moral Self,” in J. Kiverstein (ed.), Handbook
of the Social Mind. London: Routledge.
Reid, T. (2002) Essays on the Intellectual Powers of Man, D.R. Brookes (ed.). University Park: The
Pennsylvania State University Press.
Schechtman, Marya (1990) “Personhood and Personal Identity,” Journal of Philosophy, 87: 71–92.
Schechtman, Marya (2005) “Personal Identity and the Past,” Philosophy, Psychiatry, & Psychology, 12: 9–22.
Shoemaker, David (2016) “Personal Identity and Ethics,” The Stanford Encyclopedia of Philosophy, Edward
N. Zalta (ed.), forthcoming, http://plato.stanford.edu/archives/spr2016/entries/identity-ethics/.
Shoemaker, Sydney (1970) “Persons and Their Pasts,” American Philosophical Quarterly, 7: 269–85.
Sierra, M. and Berrios, G. E. (2000) “The Cambridge Depersonalisation Scale,” Psychiatry Research, 93(2):
153–64.
Sierra, M. and Berrios, G. E. (2001) “The Phenomenological Stability of Depersonalization: Comparing
the Old with the New,” Journal of Nervous and Mental Disease, 189(9): 629–36.
Strohminger, N. and Nichols, S. (2014) “The Essential Moral Self,” Cognition, 131(1): 159–71.
Strohminger, N. and Nichols, S. (2015) “Neurodegeneration and Identity,” Psychological Science, doi:
0956797615592381.
Suddendorf, T. and Corballis, M. (2007) “The Evolution of Foresight: What Is Mental Time Travel, and
Is It Unique to Humans?” Behavioral and Brain Sciences, 30: 299–313.
Szpunar, K. (2010) “Episodic Future Thought: An Emerging Concept,” Perspectives in Psychological Science,
5: 142–62.
Tulving, E. (1972) “Episodic and Semantic Memory,” in E. Tulving and W. Donaldson (eds.), Organisation
of Memory. New York: Academic Press.
Tulving, E. (1985) “Memory and Consciousness,” Canadian Psychology/Psychologie Canadienne, 26: 1–12.
Tulving, E. (1993) “Self-knowledge of an Amnesic Individual Is Represented Abstractly,” in T. K. Srull
and R. S. Wyer (eds.), Advances in Social Cognition (Vol. 5, pp. 1–49). Hillsdale, NJ: Erlbaum.
Weinberg, S. (2011) “Locke on Personal Identity,” Philosophy Compass, 6(6): 398–407.
Wheeler, M., Stuss, D., and Tulving, E. (1997) “Toward a Theory of Episodic Memory: The Frontal
Lobes and Autonoetic Consciousness,” Psychological Bulletin, 121: 331–54.
White, C., Kelly, B., and Nichols, S. (2016) “Remembering Past Lives: Intuitions about Memory and
Personal Identity in Reincarnation,” in H. Cruz and R. Nichols (eds.), The Cognitive Science of Religion
and Its Philosophical Implications. Routledge.
Williams, B. (1970) “The Self and the Future,” The Philosophical Review, 79: 161–80.
179
14
MEMORY AND
SELFCONSCIOUSNESS
1. The landscape
Philosophical questions about memory fall into (at least) three broad categories. Some are
fundamentally metaphysical, having to do with the nature of memory. In this category fall ques-
tions about how to understand the relation between memory and the past (should memory be
understood as a direct relation to the past, for example, or should we think instead in terms of
occurrent memory states that are causally derived from past events?). A second set of questions
is broadly epistemological. We can ask about how memory judgments are justified, and whether
memories create knowledge or simply preserve it. The third set of questions has to do with the
functional role of memories. How do memories interact with other mental states? What does the
capacity for memory contribute to cognitive and affective life? Issues of functional role can have
epistemological and metaphysical ramifications, as well as raising ethical questions (e.g., about
the role of shared memories in social life).
Thinking about the self and self-consciousness raises exactly the same categories of question
(see Bermúdez 2017a). Typical metaphysical questions have to do with the persistence condi-
tions for selves—are they bodily, psychological, or some combination of the two, for example?
Questions about the epistemology of the self are questions about self-consciousness or self-
awareness. Do self-conscious subjects have any type of privileged access to their own mental or
physical states, for example? Is self-consciousness a purely conceptual phenomenon, or does it
have a nonconceptual dimension? Exploring the functional role of self-consciousness involves
questions about how the capacity for self-consciousness depends upon (and/or is enabled by)
other cognitive and affective capacities, such as the ability to represent an objective world, or the
capacity to engage intellectually and socially with other self-conscious subjects.
This chapter focuses on areas of interdependence between memory and self-consciousness,
primarily in the dimensions of epistemology and functional role. In particular—
(1) the role of self-consciousness in memory;

(2) the role of memory in self-consciousness.
The most relevant type of memory for (1) and (2) is autobiographical memory. Section 2
of this chapter will clarify the nature of autobiographical memories and judgments based on
180
Memory and self-consciousness
autobiographical memory. In Sections 3 and 4, we will turn to an important epistemic prop-

erty of autobiographical memory judgments—their immunity to error through misidentifica-
tion relative to the first-person pronoun. Understanding this property is particularly important
for understanding the role of self-consciousness in memory. Section 5 explores the relation
between autobiographical memory and autobiographical narrative, approaching the issue
through a question posed by Bertrand Russell about the perceived pastness of autobiographical
memory.
2. What are autobiographical memories?

Psychologists studying memory typically classify memories in terms of three basic distinctions
(see also Chapter 1 of this volume).
The first distinction is between implicit memories and explicit memories. Implicit memory
can be seen at work in tasks where performance is improved through retained information
that falls below the threshold of consciousness (Roediger 1990). Priming experiments probe
implicit memory, using masked stimuli that are displayed for durations too short to be con-
sciously registered. Implicit memory is also a feature of some neuropsychological disorders,
such as amnesia, where patients are able to learn how to perform tasks even though they
deny ever having seen those tasks before (Brooks and Baddeley 1976). Explicit memories, in
contrast, are conscious. Subjects are both able to exploit retained information and aware that
they are so doing.
A second distinction is between procedural and declarative memory, which maps closely onto
the more familiar philosophical distinction between know-how and knowing-that. Procedural
memory underlies most cognitive and motor skills, such as talking or riding a horse. It is not
something that people are typically able to articulate, which is why language teaching and ath-
letic coaching are so difficult. Declarative memories, in contrast, involve retaining specific items
of information, as opposed to being able to deploy general skills or abilities.
The third standard distinction in the psychology of memory is between semantic memory
and episodic memory, two different categories of declarative memory. Semantic memory
is memories of particular facts. Being able to recall the year of the French Revolution is a
paradigm example of semantic memory. All that matters for semantic memory is that informa-
tion is recalled. There is no need to recall the circumstances in which the information was
acquired. In episodic memory, on the other hand, one remembers specific occurrences that
one experienced at first hand.
Autobiographical memories are explicit, declarative, and episodic. Normal subjects can typi-
cally recall considerable amounts of semantic information about themselves, much of it acquired
from others. But even when this semantic information concerns our personal history (informa-
tion about one’s place and time of birth, for example), it does not count as autobiographical in
the sense in which I am using the term.
Nonetheless, not all explicit, declarative, and episodic memories are properly described as
autobiographical. An example of a non-autobiographical episodic memory would be remem-
bering that Germany beat Argentina in the final of the 2014 World Cup in virtue of a recalled
image of the winning German players when the whistle went. This is not an autobiographical
memory because, although it originates in and recalls an episode in my personal history, it is not
a memory specifically about me and my past. So, we need to add a clause to the characterization
of autobiographical memories. In addition to being explicit, declarative, and episodic, autobio-
graphical memories are self-specifying.
181
The soccer game example sheds light on another dimension in thinking about autobiographical
memories. I could also, on the basis of the very same image, remember that I watched the game
on TV. That would qualify as an autobiographical episodic memory, because it is self-specifying
in the correct sense. So, a single memory image of a given episode can count in different contexts
as both an autobiographical and a non-autobiographical episodic memory.
This shows that we need to make an explicit distinction between memory images and
the accompanying memory judgments, rather than talking about memories simpliciter.
Autobiographical memory judgments have first-person contents—contents that would canoni-
cally be expressed using the first-person pronoun. They originate in, and are grounded by,
episodic memory images that preserve one’s original experience of the remembered event. In
many cases, those images could also give rise to, and ground, non-autobiographical memory
judgments. One of the principal topics for the following will be the relation between autobio-
graphical memory judgments and the memories on which they are based.
A final distinction. Some autobiographical memory judgments are explicitly recollective. Judgments
such as “I remember -ing” or “I remember that I -ed” make their source explicit. The fact
that the judgment is based on autobiographical memory is part of its content. Philosophical dis-
cussions of autobiographical memory have typically focused on explicitly recollective judgments.
But many, and quite probably most, autobiographical memory judgments are not explicitly
recollective. Ordinary past-tense judgments, of the form “I was -ing” or “I -ed,” count as
autobiographical memory judgments, provided that they are suitably derived from autobiographi-
cal memory images. It is not part of their content that they are based on memories (any more
than it is part of the judgment “my arms are folded” that it is based upon somatic proprioception).
As we will see, past-tense judgments pose significant challenges that are not present in explicitly
recollective autobiographical memory judgments.
3. Are autobiographical memory judgments immune to

error through misidentification?
Philosophers have explored different ways subjects might have privileged access to the contents
of their own minds. Debates have shifted from strong claims about indubitable or incorrigible
knowledge of our thoughts, feelings, and other states, to more subtle forms of privileged access.1
The most discussed notion in this area is immunity to error through misidentification relative
to the first-person pronoun. This concept originates in Wittgenstein’s Blue Book, but its role in
contemporary discussion is due to Sydney Shoemaker (1968) and Gareth Evans (1982).
Here is Wittgenstein’s statement of the distinction between “I-as-subject” and “I-as-object”:
There are two different cases in the use of the word “I” (or “my”) which I might call
“the use as object” and “the use as subject.” Examples of the first kind of use are these:
“My arm is broken,” “I have grown six inches,” “I have a bump on my forehead,” “The
wind blows my hair about.” Examples of the second kind are: “I see so-and-so,” “I try
to lift my arm,” “I think it will rain,” “I have a toothache.”
(Wittgenstein 1958: 66–7)
A certain type of error, Wittgenstein claims, is only possible when “I” is used “as object”:
One can point to the difference between these two categories by saying: The cases of
the first category involve the recognition of a particular person, and there is in these
cases the possibility of an error, or, as I should rather put it: The possibility of an error
182
has been provided for . . . It is possible that, say in an accident, I should feel a pain in
my arm, see an arm at my side, and think it is mine when really it is my neighbor’s.
And I could, looking into a mirror, mistake a bump on his forehead for one on mine.
On the other hand, there is no question of recognizing a person when I say I have
toothache. To ask “are you sure that it’s you who have pains?” would be nonsensical.
(Wittgenstein 1958: 67)
Something that does not emerge clearly in Wittgenstein’s discussion is that whether “I” is being
used as subject or object depends upon the grounds on which the assertion is based. Suppose
I say “I have a toothache” because I feel a pain in my tooth. I cannot feel a pain and then
wonder whether that pain is mine. So this is clearly a use of “I” as subject. But suppose I have
an unlocalized pain in my lower jaw and consult a neuroscientist who is able, using scanning
technology, to tell me that the pain is actually in my tooth. If I then say “I have a toothache,”
I would be using “I” as object because the possibility of error has entered the picture—my scan
might have got mixed up with someone else’s, for example, with the result that, although I am
feeling pain, it is not actually me whose tooth is aching. In this case, my judgment is susceptible
to error through misidentification, whereas in the first case, the judgment expressed with very
same words is immune to error through misidentification.
This relativity to grounds and evidence comes across clearly in the following formulation
of immunity to error through misidentification (henceforth: the immunity property), which
sticks close to Shoemaker’s original characterization, except by talking of grounded belief where
Shoemaker talks of knowledge:
To say that a statement “a is ϕ” is subject to error through misidentification relative to the first
person “I” means that the following is possible: The speaker believes that some particu-
lar thing is ϕ, with her belief grounded in an appropriate evidence base, but she makes
the mistake of asserting “I am ϕ” because, and only because, she mistakenly thinks that
she herself is the thing she believes to be ϕ.
(Bermúdez 2012: 125)
Judgments with the immunity property do not involve identifying a particular person as
oneself because the sources of information on which they are based are such that they can
only provide information about oneself. No inference is required to move from learning that
someone is F to the conclusion that one is oneself F. For that reason, such judgments are
widely held to be epistemically basic for self-knowledge.2
Do autobiographical memory judgments have the immunity property? We can approach this
question through debates about memory-based criteria for personal identity.
According to psychological accounts of personal identity, psychological continuity suffices
for continuity of self, so that there are no obstacles in principle to the same person consecutively
inhabiting a series of different bodies. Obviously memories, and autobiographical memories
in particular, are foundational for psychological continuity. However, Bishop Butler famously
argued that any memory-based criterion of personal identity is circular, because nothing can
count as a personal or autobiographical memory unless what is remembered was originally expe-
rienced or learnt by the very same person doing the remembering—personal identity cannot be
analyzed in terms of autobiographical memory because the latter presupposes the former.
In an attempt to circumvent the circularity objection, Sydney Shoemaker proposed an
impersonal notion of quasi-memory. It is, Shoemaker thinks, perfectly possible that one person’s
apparent memories might causally originate in someone else’s psychological life—we might
183
imagine a case of fission, for example, in which a psychological subject S splits into two “new”
subjects S1 and S2, neither of which is identical to the original subject S but each of which inher-
its all of S’s memories. It is just a contingent matter of fact that this sort of thing does not happen.
But Shoemaker concludes from the fact that it is even possible that we should not assume that
memory involves continuity of person. So he proposes to replace talk of remembering with talk
of quasi-remembering, defined as follows:
a kind of knowledge of past events such that someone having this sort of knowledge
of an event does involve there being a correspondence between his present cognitive
state and a past cognitive and sensory state that was of the event, but such that this
correspondence, although otherwise just like that which exists in memory, does not
necessarily involve that past state having been a past state of the very same person who
subsequently has the knowledge.
(Shoemaker 1970: 23–4)
Quasi-remembering, as I shall use the term, includes remembering as a special case. One way
of characterizing the difference between quasi-remembering and remembering is by saying that
the former is subject to a weaker previous awareness condition than the latter. Whereas some-
one’s claim to remember a past event implies that he himself was aware of the event at the time
of its occurrence, the claim to quasi-remember a past event implies only that someone or other
was aware of it (Shoemaker 1970: 24).
According to Shoemaker, Butler’s circularity objection can be circumvented by taking the
notion of quasi-memory as basic and defining memory in terms of it. We can say, for example,
that S remembers an event E just if S has a quasi-memory M of E and the causal chain connect-
ing M and E has not “branched” into two or more parallel chains (as in the fission process that
generates S1 and S2 from S). This definition does not presuppose continuity of person.
I am unconvinced that this maneuver is as effective as Shoemaker thinks. It is true that a
clause ruling out branching causal chains is suitably “impersonal” to feature in a non-circular
account of personal identity. But it is not general enough to exclude all the ways in which
a quasi-memory could fail to count as a real memory. Fission is just one way of producing
quasi-memories that are not genuine memories. But there are others. Anyone prepared to coun-
tenance the possibility of fission should have no conceptual problem with a scenario in which,
for example, some of a dying person’s memories are scanned and then transplanted into another
person’s brain, where they “fuse” with that person’s own memories. Such a scan/transplant sce-
nario would certainly involve a deviant causal chain (because memories are normally preserved
in ways that do not permit fusion). But here, unlike the fission case, spelling out why the causal
chain is deviant seems to require mentioning continuity of person.
In any event, Shoemaker takes the possibility of quasi-memory to entail that autobiographical
memory judgments cannot have the immunity property:
Whereas I can only remember an action from the inside only if it was my action, a world
in which there is quasi-remembering that is not remembering will be one in which it is not
true that any action one quasi-remembers from the inside is thereby an action he himself
did. So—assuming that ours may be such a world—if I quasi-remember an action from
the inside, and say on this basis that I did the action, my statement will be subject to error
through misidentification; it may be that my quasi-memory of the action is as accurate and
complete as it could be, but that I am mistaken in thinking that I am the person who did it.
(Shoemaker 1970: 27)
184
It is hard to dispute Shoemaker’s characterization of autobiographical memory judgments in an

environment where it is a live possibility that one might quasi-remember someone else’s actions
and experiences, because then it really would be open whether an apparent autobiographical
memory really reflected one’s own personal history. If such deviant quasi-memories are possible
then so too are errors of misidentification. But on the safe assumption that we do not (currently)
inhabit a world where deviant quasi-memories are live possibilities, this does not tell us anything
significant about autobiographical memory in our world. Certainly, if deviant quasi-memories are
even possible, then autobiographical memory judgments cannot be logically immune to error
through misidentification, where this would require there not being a possible world in which
that judgment could be made on the same grounds that it currently is and lack the immunity
property. But arguably no judgments have the immunity property in this strong sense, since
the immunity property is relative to the grounds on which the judgment is based and it seems
always possible to imagine a world with a route from grounds to judgment sufficiently deviant
to allow the possibility of misidentification. Our concern is with this world and other worlds
like it in not containing cases of fission and memory transfer. Within that sphere of possibility,
autobiographical memory judgments have the immunity property, albeit in a de facto sense.
Of course, Shoemaker makes the stronger claim that quasi-memory should be taken as fun-
damental in the order of analysis. A similar view has been proposed by Derek Parfit in the
context of defending a reductionist account of personal identity (an account that confronts
exactly the same circularity objection as do psychological continuity accounts of personal iden-
tity).3 But even if Shoemaker and Parfit are correct, then it is still not clear that autobiographical
memory judgments lack the immunity property. In a world without fission or artificial memory
creation, analyzing memory in terms of quasi-memory seems more notational than substantive.
To see this, it is helpful to look at Evans’s discussion of the immunity property. In Evans’s
terminology, judgments that lack the immunity property are identification-dependent, whereas
judgments with the immunity property are identification-free. Judgments are identification-free
when they are based on sources of information that can only provide information about the
self. Introspection yields this type of information about one’s own mental states and properties,
while proprioception and kinesthesis do so for bodily states and properties. Evans argues that
identification-dependent judgments should be analyzed in terms of a pair of judgments “I am µ”
and “µ is F,” where “µ” is some way of picking out or identifying a person. On pain of regress,
the judgment “I am µ” must either itself have the immunity property or be analyzable at some
point into a pair of judgments one of which has the immunity property (Evans 1982: 80–81).
One of the merits of this approach is that it offers a very clear way of articulating why judgments
with the immunity property are foundational.
Applying this back to Shoemaker and Parfit, analyzing autobiographical memory judgments in
terms of quasi-memory does nothing to show that they are identification-dependent and hence
need to be analyzed in terms of an explicit identification judgment in the way that Evans proposes.
Explicit identification judgments are required only when there is believed to be a risk of mis-
identification. Analyzing autobiographical memory judgments in terms of quasi-memory does not
create such a risk. For that, we would need to believe that deviant quasi-memories (in the sense
discussed three paragraphs above) are live possibilities, and all parties are agreed that they are not.
4. Explaining when and why autobiographical memory

judgments have the immunity property
So far we have been treating autobiographical memory judgments as a homogenous category.
Discussions of whether such judgments have the immunity property have typically focused
185
almost exclusively on what I earlier termed explicitly recollective memory judgments (judgments
of the form “I remembering -ing” or “I remember that I -ed”). The upshot of Section 3
was that explicitly recollective memory judgments do have the immunity property. But, as
was observed in Section 2, not all autobiographical memory judgments are explicitly recollec-
tive. Self-specifying information derived from autobiographical memory can be articulated in
past-tense judgments with the form “I was -ing” or “I -ed.” The contents of such past-tense
judgments does not reveal that they are based on autobiographical memory.
In Bermúdez (2012), I assumed that past-tense judgments based on autobiographical memory
uniformly have the immunity property. This was a mistake, as an anonymous referee subse-
quently pointed out in a different context.. Some past-tense judgments are susceptible to error
through misidentification, even when derived from autobiographical memories. We can illus-
trate this through a past-tense version of one of Wittgenstein’s examples from the Blue Book.
Suppose that I was involved in an accident and on waking up in hospital, I have a vivid memory
of feeling a pain in my arm and seeing an injured arm by my side with a large bruise. This is
plainly an autobiographical memory. It is suitably episodic and self-specifying. On the basis of
that memory, I form the judgment “My arm was bruised.” As it happens, however, the bruised
arm I saw was not my own and in fact the pain in my arm came from a fracture rather than a
bruise. This is clearly an error of misidentification.
Is there a principled way of identifying when past-tense judgments based on autobiographical
memory have the immunity property? The remainder of this section summarizes the account
proposed in Bermúdez (2013) and Bermúdez (2017b).
First, we need some terminology. Because autobiographical memories are all episodic mem-
ories, they must originate in some earlier experienced episode in the subject’s personal history.
Call that the experiential basis for the autobiographical memory. The experiential basis is causally
connected to the subsequent memory image and serves as the principal warrant for the eventual
judgment. At the time of the original experienced episode, the experiential basis could also have
served as warrant for the present-tense analog of the eventual autobiographical memory judgment.
This is the judgment that the agent either did or could have made by taking the experiential
basis at face value. So, in the accident example considered earlier in this section, my original
experience of feeling a pain in my arm and seeing a bruised arm is the experiential basis for my
eventual autobiographical memory judgment “I bruised my arm in the accident.” The present-
tense analog that I could have (and perhaps did) form at the time of the original experience is
the judgment “That’s my bruised arm.”
My proposal is that the immunity status of a past-tense judgment is grounded in its expe-
riential basis in a way that exactly maps how that experiential basis grounds the immunity
status of the present-tense analog. On this view, a past-tense judgment grounded in autobio-
graphical memory has the immunity property if and only if its present-tense analog has the
immunity property.
An initial reason to be sympathetic to this proposal is that the past-tense judgment and the
present-tense analog share the same warrant. Each is justified through the same experienced
episode, and one might plausibly think that the same features of that experienced episode will
settle the question of whether a judgment warranted by it has the immunity property irre-
spective of whether that judgment is in the present tense or the past tense. In the background
here is a general thesis about the semantics of tense. This is the thesis, most crisply articulated
in Prior’s tense logic, that tense functions adverbially, as an operator on sentences. On this
view, sentences are, as it were, intrinsically in the present tense. To put a sentence’s main
verb in the past or future tense is, in effect, to prefix the sentence with an operator either of
186
the form “It was the case that” or of the form “it will be the case that.” Complex tenses can
be constructed by iterating these two basic tenses. The future perfect (“it will have been the
case that ”) is modeled as “it will be the case that it was the case that ,” and the pluperfect
(“it had been the case that ”) is modeled as “it was the case that it was the case that .”4
Understanding tense adverbially in this way strongly suggests a fundamental identity of con-
tent between a past-tense judgment and its present-tense analog. If we assume that the past-tense
judgment is in the imperfect tense then it has the form “P( )” (where “P” is the “it was the case
that” operator). The embedded content is, of course, the content of the present-tense analog.
What we might term the non-adverbial content of the past-tense judgment just is the content of
the present-tense analog. If the past-tense judgment embeds the present-tense analog in this way,
it is eminently reasonable to think that either the two judgments will both have the immunity
property, or neither will. They differ only in the pastness operator and it is hard to see how that
can either remove the immunity property from an embedded judgment that possesses it, or add
the immunity property to a judgment that lacks it.5
The view that there is no difference in representational content between a past-tense judg-
ment and its present-tense analog sits very naturally with the view that memory is fundamentally
preservative. If memory is a tool for preserving knowledge, rather than creating it, then we
would expect there to be no intrinsic difference in representational content between an auto-
biographical memory judgment derived from an experienced episode in one’s personal history
and a present-tense judgment derived from that same episode.
In sum, explicitly recollective judgments all possess the immunity property. Judgments of
the form “I remember –ing” can of course be mistaken, but not because one misidentifies
oneself as the person whom one remembers –ing. Non-recollective past-tense memory judg-
ments do not invariably possess the immunity property. They possess it when, and only when,
the recalled experiences are such that they would have warranted a present-tense judgment that
would itself have had the immunity property. In this way, therefore, the immunity status of
past-tense memory judgments is inherited from epistemic features of the original experience.
5. Autobiographical memory and narrative

The discussion so far has focused primarily on the role of autobiographical memory in self-
consciousness, in particular on how and why autobiographical memory judgments have the
immunity property. We turn now to the opposite direction of fit—the role of self-consciousness
in autobiographical memory.
To introduce the issues here consider the following passages from Bertrand Russell’s
discussion of memory in The Analysis of Mind:
Suppose you ask me what I ate for breakfast this morning . . . The process of remem-
bering will consist of calling up images of my breakfast, which will come to me with
a feeling of belief such as distinguishes memory-images from mere imagination-
images . . . Memory-images and imagination-images do not differ in their intrinsic
qualities, so far as we can discover. They differ by the fact that the images that consti-
tute memories, unlike those that constitute imagination, are accompanied by a feeling
of belief which may be expressed in the words “this happened.” The mere occurrence
of images, without this feeling of belief constitutes imagination; it is the element of
belief that is distinctive in memory.
(Russell 1921: 175–6)
187
Memory demands (a) an image, (b) a belief in the past existence . . . The believing is
a specific feeling or sensation or complex of sensations, different from expectation or
bare assent in a way that makes the belief refer to the past; the reference to the past lies
in the belief-feeling, not in the content believed.
(Russell 1921: 186)6
It is plain from the context that Russell is discussing episodic memory, including autobiographical
memory. On Russell’s analysis, an autobiographical memory has two components: a memory-
image and what I will term an indicator of pastness. One does not have to accept Russell’s claim
that there are no intrinsic differences between memory-images and imagination-images to think
that there is an important question about how memory images are, as it were, referred to the past.
What is the indicator of pastness for autobiographical memory? Russell talks about a
“feeling of belief,” a “specific feeling or sensation, or complex of sensations” that can be expressed
as “This happened.” From a contemporary perspective, this seems confusing, eliding as it does
standard distinctions between the phenomenal and the doxastic (between sensations and beliefs).
On the face of it, there seem to be two different (but not necessarily exclusive) ways of thinking
about the indicator of pastness—in terms of sensation or in terms of belief. Russell, despite
his vocabulary, seems to be adopting the former approach, locating the indicator in feelings
of pastness that are attached to the memory-image but not part of its representational content.
It is plausible, though, that no purely sensational approach can do the job. Even if there
were a feeling of pastness that attaches to all and only episodic memories, it would need to be
interpreted in order to play a cognitive role. One way of thinking about this is in terms of judg-
ment, so that the feeling of pastness both gives rise to and justifies a judgment of pastness. This
sort of approach runs into familiar difficulties, however. As Davidson, McDowell, and others
have emphasized, it is far from obvious how “brute” sensations can provide justification or
warrant. This is one reason why other philosophers have explored the idea that perceptual and
sensational states might have nonconceptual content that is representational without requiring
concept possession or being propositionally structured.7
It is not immediately obvious how the notion of nonconceptual content can be applied in
this context, however. And in any case it would be hard to reconcile with the idea that the
putative feeling of pastness is not part of the content of the memory-image. It would be far more
plausible to think of the memory-image as (nonconceptually) representing the remembered
episode as having taken place in the past than to think of there being some additional meta-
representational state that represents the memory-image as referring to the past.8
In the remainder of this section, I will explore a rather different approach to the indicator
of pastness. To begin, suppose that there is a judgment of pastness: what would its content
be? Russell proposes “This happened.” This may well be adequate for non-autobiographical
episodic memories, but it leaves out the central feature of autobiographical episodic memories.
As we saw in Section 2, autobiographical memories are distinctive in being self-specifying. My
autobiographical memories don’t just represent events as having happened in the past. They
represent them as having happened in the past to me. In fact, we can go further. In autobio-
graphical memory, events are represented as having happened in my past.
This basic fact about autobiographical memory both reveals the role of self-consciousness in
autobiographical memory and marks a further and fundamental point of difference between auto-
biographical memory and non-autobiographical episodic memory. If autobiographical memory
presents events as having taken place in my past, then it actively exploits the ability to situate a
remembered episode within a personal history.9 This involves a sophisticated form of self-con-
sciousness—consciousness of oneself as a temporally extended agent with a distinctive and unique
188
personal history. It is a familiar point that human beings are narrative creatures, and that auto-
biographical memories are key elements in the raw materials from which personal narratives are
constructed. But we also need to think about the relation between narrative and autobiographical
memory in the opposite direction. Awareness of oneself as having a personal history seems to be
a precondition of having autobiographical memories. A creature cannot have autobiographical
memories, one might say, unless it can fit the remembered episodes into its autobiography.
Autobiographical memories can be specific or generic. Specific autobiographical memo-
ries are typically indexed to a certain point or period within the thinker’s autobiography. An
example of a generic autobiographical memory might be attending a concert at Symphony
Hall in St. Louis. I have vivid first-hand memories of entering the lobby and sitting in the
concert hall, for example, but they are a composite from many different episodes. But even
generic autobiographical memories are typically experienced as memories of events that took
place within a certain autobiographical time frame. An autobiographical time frame is a linearly
ordered sequence of events in the life of a person. This is in contrast to the cyclical perspective
on time that we find, for example, in a creature whose life is regulated by lunar cycles and the
seasons.10 An autobiographical narrative is more than just an ordered sequence of memories. It
can incorporate a thinker’s goals and aspirations, realized or otherwise. It can have built into
it information about the thinker’s personal history that is not derived from memory. And of
course it is likely to contain anticipations of the future, as well as plans.
So, there is no gap between autobiographically remembering an episode and locating that
episode within one’s own personal history. Non-autobiographical episodic memories are not
like that, however. Something like Russell’s “This happened” judgment is more appropri-
ate than “This happened to me” or (better) “This is part of my history.” One might think
that only language-users can construct and deploy autobiographical narratives. If that is cor-
rect, then only language-users can have autobiographical memories. But no such constraint
seems plausible for episodic memory. Episodic memories do not need to be integrated into
an autobiographical history, or even into any linearly ordered sequence. A creature whose
experience of time is cyclical rather than linear would not seem thereby to be disqualified
from episodic memory.
However, there may be broadly Kantian reasons for thinking that episodic memories do
have to be linearly ordered in creatures capable of autobiographical memory. Part of what it
is to possess an autobiographical narrative is, first, to be able to distinguish between one’s own
personal history and the history of a mind-independent external environment in which that
personal history unfolds, and, second, to be able to integrate them and understand how they fit
together. There is an analogy here with different ways of thinking about space. As Gareth Evans
(among others) has stressed, creatures with an objective conception of space are able, first, to dis-
tinguish their own, egocentric spatial route through the world from the non-egocentric spatial
framework through which they are moving, and, second, to integrate them and understand how
they fit together.11 If this analogy between temporal and spatial representation is correct, then
it would be plausible to think that (for creatures capable of autobiographical memory) episodes
remembered through episodic memory must be incorporated in and indexed to a (linear) his-
tory of the mind-independent external world.
I end this section by briefly suggesting some epistemological consequences of this narrative
conception of autobiographical memory. The narrative requirement is an important factor in
fixing the degree of confidence in apparent memories. The easier it is to fit a remembered epi-
sode within a personal history, the more likely one is to believe that it really happened. To that
extent considerations of coherence are fundamental in fixing a thinker’s degree of confidence
in an apparent autobiographical memory. This provides some support for coherentist theories
189
of memory, as proposed, e.g., in Bonjour (2002) and McGrath (2007). On coherentist theories
memory judgments are justified to the extent that they cohere with the thinker’s overall beliefs.
Coherentists typically do not pick out specific elements of the belief system as being more impor-
tant than others, but the discussion in this section suggests that coherence with the thinker’s
autobiographical narrative will be particularly important.
At the same time, though, these processes open the door for confabulation. As revealed by
a dispiritingly extensive empirical literature, it is all too easy to integrate fictitious memories
into a convincing autobiographical narrative. False memory syndrome is an extreme but well-
documented example (Mendez and Fras 2011). Equally compelling is the experimental work
in cognitive psychology on the (un-)reliability of eyewitness testimony, which has clear and
unwelcome consequences for how eyewitness testimony should be treated in courts of law
(Wells and Seelam 1995; Porter and Baker 2015). One lesson to draw is that coherence may
not be sufficient for autobiographical and episodic memories to be justified. In these cases,
coherence may need to be supplemented by externalist criteria such as reliability.
Notes
1 See Gertler (2011) for an overview of recent discussions.
2 For recent essays on the immunity property, see Prosser and Recanati (2012).
3 See Parfit (1984) and, for critical discussion, Burge (2003), Cassam (1997: Chapter 5), McDowell (1997),
and Peacocke (2014).
4 The sentence operator view of tense brings with it a number of general difficulties and challenges,
particularly with respect to the semantic value of the embedded sentence. See Evans (1985); Lewis
(1980/1998), and King (2007).
5 For another argument, see Bermúdez (2013).
6 These two passages represent the second of Russell’s two views of memory, the first being the theory
that memory is direct acquaintance of the past, proposed in Problems of Philosophy (Russell 1912). For
discussion of Russell’s views, see Pears (1975) and Baldwin (2010).
7 For further references and a review of the debate, see Bermúdez and Cahen (2015).
8 See Chapter 2 in this volume for further discussion of the phenomenology of memory.
9 Sven Bernecker asked whether it might be logically possible for there to be a mental life that contains
just one autobiographical memory.There seem to be no problems in principle with thinking of a single
episodic memory image. As indicated earlier, a single memory image can ground both an autobio-
graphical memory judgment and an episodic memory judgment. But moving from an episodic memory
image to an autobiographical memory judgment requires thinking of oneself as a temporally extended
entity. I have difficulty seeing how a creature whose mental life contains only one memory could think
of itself in that way. I recognize, though, that this falls short of an argument!
10 This general line of thought has its roots in Kant, particularly as interpreted by Strawson (Strawson
1959, 1975). See Campbell (1994) for extensive discussion of how a linear perspective on time contrasts
with a cyclical one, and how it is integral to self-consciousness. The papers in Hoerl and McCormack
(2001) are also very relevant.
11 See Evans (1982: Chapter 6) and, for related discussion, Campbell (1994), Cassam (2005), and Bermúdez
(2017b).
Related topics
• Taxonomy and the unity of memory
• The phenomenology of memory
References
Baldwin, T. 2010. Russell on memory. Principia 5: 187–208.
Bermúdez, J. L. 2012. Memory judgments and immunity to error through misidentification. Grazer
Philosophische Studien 84: 123–42.
190
Bermúdez, J. L. 2013. Immunity to error through misidentification and past-tense memory judgements.
Analysis 73: 211–20.
Bermúdez, J. L. 2017a. Self-consciousness. In The Blackwell Companion to Consciousness, eds. M. Welmans
and S. Schneider. Oxford: Wiley-Blackwell.
Bermúdez, J. L. 2017b. Understanding “I”: Thought and Language. New York: Oxford University Press.
Bermúdez, J. L. and A. Cahen. 2015. Mental content, nonconceptual. Stanford Encyclopedia of Philosophy
(plato.stanford.edu).
Bonjour, L. 2002. Epistemology. Lanham, MD: Rowman and Littlefield.
Brooks, D. N. and A. D. Baddeley. 1976. What can amnesic patients learn? Neuropsychologia 14: 111–22.
Burge, T. 2003. Memory and persons. Philosophical Review 112: 289–337.
Campbell, J. 1994. Past, Space, and Self. Cambridge, MA: MIT Press.
Cassam, Q. 1997. Self and World. Oxford: Oxford University Press.
Cassam, Q. 2005. Space and objective experience. In Thought, Reference, and Experience: Themes from the
Philosophy of Gareth Evans, ed. J. L. Bermúdez. Oxford: Clarendon Press.
Evans, G. 1982. The Varieties of Reference. Oxford: Oxford University Press.
Evans, G. 1985. Does tense logic rest on a mistake? In Collected Papers, ed. G. Evans. Oxford: Oxford
University Press.
Gertler, B. 2011. Self-Knowledge. New York: Routledge.
Hoerl, C. and T. McCormack. 2001. Time and Memory: Issues in Philosophy and Psychology. Oxford: Oxford
University Press.
King, J. C. 2007. The Nature and Structure of Content. Oxford: Oxford University Press.
Lewis, D. 1980/1998. Index, context, and content. In Papers in Philosophical Logic. Cambridge: Cambridge
University Press.
McDowell, J. 1997. Reductionism and the first person. In Reading Parfit, ed. J. Dancy. Oxford: Blackwell,
pp. 230–50.
McGrath, M. 2007. Memory and epistemic conservatism. Synthese 157: 1–24.
Mendez, M. F. and I. A. Fras. 2011. The false memory syndrome: Experimental studies and comparison to
confabulations. Medical Hypotheses 76: 492–6.
Parfit, D. 1984. Reasons and Persons. Oxford: Oxford University Press.
Peacocke, C. 2014. The Mirror of the World: Subjects, Consciousness, and Self-Consciousness. Oxford: Oxford
University Press.
Pears, D. 1975. Russell’s theories of memory. In Questions in the Philosophy of Mind. London: Duckworth.
Porter, S. B. and A. T. Baker. 2015. CSI (Crime Scene Induction): Creating false memories of committing
crime. Trends in Cognitive Sciences 19: 716–18.
Prior, A. N. 1968/2003. Papers on Time and Tense. Oxford: Oxford University Press.
Prosser, S. and F. Recanati, eds. 2012. Immunity to Error Through Misidentification: New Essays. Cambridge:
Roediger, H. L. 1990. Implicit memory: Retention without remembering. American Psychologist 45: 1043–56.
Russell, B. 1912. Problems of Philosophy. London: Williams and Norgate.
Russell, B. 1921. The Analysis of Mind. London: George Allen and Unwin.
Shoemaker, S. 1968. Self-reference and self-awareness. Journal of Philosophy 65: 555–67.
Shoemaker, S. 1970. Persons and their pasts. American Philosophical Quarterly 7: 269–85.
Strawson, P. F. 1959. Individuals: An Essay in Descriptive Metaphysics. London: Methuen.
Strawson, P. F. 1975. The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason. Vol. 18. Distributed
by Harper and Row, Barnes and Noble Import Division.
Wells, G. L. and E. P. Seelau. 1995. Eyewitness identification: Psychological research and legal policy on
lineups. Psychology, Public Policy, and Law 1: 765–91.
Wittgenstein, L. 1958. Preliminary Studies for the “Philosophical Investigations,” Generally Known as the Blue and
Brown Books. Oxford: Basil Blackwell.
191
15
MEMORY AND NARRATIVITY
Daniel D. Hutto
Although we can potentially narrate any specific event or recurring events in our lives, including
acts of remembering themselves, only one special sort of memory – autobiographical reverie – has
a strong claim for being indelibly narrative in nature. There is robust empirical and theoretical sup-
port for thinking that autobiographical remembering depends upon the mastery of socio-cultural
narrative practices and the exercise of narrative skills. In getting clear about why autobiographical
memory and narrativity may be inescapably bound together, a preparatory comparison with more
purely embodied forms of remembering proves instructive.
1. Memory’s mixed bag

Procedural memory is the most ubiquitous kind of remembering: it is in essence remembering
how to do something. We and other animals not only remember how but also remember who,
what and where in purely enactive and embodied ways – insofar as they require nothing more
than certain reinitiating a familiar pattern of prompted response, albeit with adjustments that are
dynamically sensitive to changes in circumstance and context (Sutton and Williamson 2014).
Purely embodied acts of remembering do not require representing any specific past hap-
pening or happenings, and especially not representing these as past happenings. These sorts of
memory are usefully characterized as embodied precisely because in an important sense such
remembering is a matter of re-enactment that does not involve representation (Casey 1987).
These forms of remembering, as exemplified by procedural memory, do ‘not store represen-
tations of external states of the world’ (Schacter and Tulving 1994: 26, see also Michaelian
2016: 26ff).
Here we find the great divide. Declarative memory – by contrast – is a different sort of ani-
mal altogether in that it does require contentful representation. Compare trying to recall and
relate, in words, what is involved in cooking a breakfast as opposed to demonstrating that one
remembers how to cook a breakfast by simply exercising the capacity by cooking one.
A crucial difference between these two types of memory is that the former does not involve
representing any of particular past occurrences whereas explicit representation is a necessity for
the latter. Non-declarative remembering differs starkly and fundamentally from acts of remem-
bering that could feature in autobiographical narratives in that the former does not inherently
involve any kind of contentful representation of particular happenings.
192
Memory and narrativity
Even within the class of contentful, declarative types of memory, some forms of remember-
ing are more clearly connected with narrativity than others. For example, pure semantic or
factual memory, the remembering of isolated facts – such as my date of birth – or so it seems at
first glance – involve narrativity, even if they involve acts of autobiographical memory that do.
It is the latter type of memory – autobiographical memory – that some hold can lay strong claim
to being inherently narrative.
2. The social interactionist theory of autobiographical memory

A well-established tradition in developmental psychology regards autobiographical memory
as shot through with narrativity. The principal architects of the Social Interactionist Theory,
or SIT, of autobiographical remembering hold that it is a distinctive kind of memory that
requires the development and exercise of socio-culturally acquired narrative capacities (Fivush
and Reese 1992; Fivush 1997; Fivush and Nelson 2004; Nelson 2003, 2007; Nelson and Fivush
2004; Fivush and Graci, Chapter 21, this volume).
Advocates of SIT claim that, as a matter of contingent fact, in our world ‘a specific kind of
memory emerges at the end of pre-school period’ (Nelson 2007: 185). In a nutshell, SIT’s big
idea can be captured in the following formula: with the emergence of ‘a new form of social skill’
comes the emergence of ‘a new type of memory’ (cf. Hoerl 2007: 622, 624).
According to SIT, emergence of autobiographical memory is socially and culturally mediated
in at least two respects. First, autobiographical memory only ‘emerges within social interactions
that focus on the telling and retelling of significant life events’ (Fivush et al. 2011: 322). Second,
in acts of autobiographical remembering, we actively draw on templates found in cultural arte-
facts that we encounter in the local environment. Hence such remembering is everywhere
‘modulated by . . . sociocultural models’ (Fivush et al. 2011: 322).
Narratives enter the story on both fronts. Defenders of SIT propose that the social interactions
that are responsible for making possible the first appearances of autobiographical memory, are of
a specific kind: they are necessarily linguistically mediated and narrative in form. Moreover, they
hold that the socio-cultural models that fuel such interactions derive from ambient narratives.
Putting these claims together yields the hypothesis that a special sort of social interaction, one
involving narrative practices, makes possible a special kind of cognition, autobiographical memory.
Hoerl (2007: 622) highlights the role that SIT thinks narratives are thought to play in making
autobiographical memory possible by presenting its two central claims as follows:
(A) Narratives are the vehicles for a distinct kind of social-communicative interaction.
(B) Narratives provide a distinct kind of cognitive framework or format for remembering events.1
With assumptions (A) and (B) in play, it is clear in what sense SIT claims that the appearance
of a unique kind of autobiographical memory depends upon the mastery of certain narra-
tive practices. Becoming a competent autobiographical narrator is not a built-in talent but an
achieved skill. The requisite narrative know-how is hard-won and emerges across various stages
of ontogenetic development, fostered by ‘the narrator’s mind-enabling and mind-extending
apprenticeship in storytelling’ (Herman 2013: 230). Children first gain and sharpen their skills
as narrators by consuming and producing narratives, where such narratives can be understood
as more or less transient cultural artefacts. Specifically, narratives are representational artefacts –
they depict a particular series of possible happenings, whether real or imagined.
It is well known that children’s first attempts at narrating are pre-linguistic; they typically
occur in acts of pretend play, usually when engaging with others, where the creation of such
193
Daniel D. Hutto
narratives is ‘accompanied by – rather than [achieved] solely through – language’ (Nelson

2003: 28). Yet while SIT recognizes that pretend play is important to the development of
the kinds of narrative skills needed for autobiographical remembering, it also regards the
basic narrative skills that pretend play affords as insufficient to enable such remembering. The
rudimentary narrative skills that are gained in pretend play need to be further honed, with the
active support of caregivers, if they are to eventually make autobiographical memory possible.
Proponents of SIT hold that mastering basic narrative skills in pretend play is not enough
because learning how to reminisce about the past requires dealing with others in mastering
narrative practices of a decidedly verbal variety.2 The claim is that only linguistic narratives
provide the right kind of public vehicle for sharing the requisite insights and experiences
(Nelson 2003: 29). Which insights? Critically, defenders of SIT hold that engaging with lin-
guistic narratives is necessary for forming an understanding of oneself that contrasts with, and
stands over and against, that of another and others – and hence for being able to represent one’s
past as belonging to one’s own personal history. SIT holds that social interactions which make
use of linguistic narratives as objects of shared attention are the means through which children
come to discover the existence of divergent perspectives. Thus verbal narratives are taken to
provide the medium for the child’s first communion with the idea that there are points of view
on things other than his or her own.3
Thus the major turning point in developing a capacity for autobiographical memory comes,
according to fans of SIT, when children are able to have conversations with adults that are
aimed at co-constructing narratives about the past. It is through this particular linguistic medium
that ‘children are being confronted with the fact that their memory is not the same as others’
(Fivush and Nelson 2006: 240). The following, representative exchange, in which M stands for
mother, and C for child, highlights key features of the process:
M: What other animals did we see at the circus?

C: A giraffe
M: No, we didn’t see a giraffe at the circus. Who did we see at the circus that looked
funny? Remember the rhinoceros?
< 4 turns focusing on seeing a rhinoceros >
M: What else did we see at the circus?
C: Um, giraffe.
M: No, we didn’t.
(Fivush and Nelson 2006: 239, reproduced from Reese et al. 1993)
This is a clear case not only in which a caregiver attempts to instil norms for giving a
correct account of the past but also through which the child’s representation of the past is co-
constructed. It is an instance in which ‘speakers are negotiating other people’s awareness of
the past events’ (Sutton 2003: 150). It highlights the ways in which, as the child learns what is
required for accurately representing the past, ‘the mother and child negotiate what happened,
how they felt about it, and their evaluative perspective of the event as a whole’ (Fivush and
Nelson 2006: 244; Fivush and Wang 2005).
Clearly, quite a lot changes in the pre-school and early school period that puts children in
a position to enter into and master the kinds of linguistically mediated conversations that are
apparently needed to scaffold their autobiographical narrative abilities. Such conversations with
others are crucial, according to SIT, in forging the capacity to remember the past autobiographi-
cally. These narratively based and focused social interactions are what enable the child to first
194
learn what making accurate claims about the past requires, as well as what it is to be a person
with a temporally extended existence that can feature in an autobiographical narrative.
There is strong evidence that the process of learning how to narrate one’s personal past is
neither straightforward nor easy for children. For example, when children of 2–4 years of age
are first learning to generate such narratives, they tend to appropriate elements of another’s life
story into the content of their own.4
There is also further evidence, which speaks in favour of SIT, that the ways in which adult
caregivers narrate events strongly influences ‘the richness and narrative organisation of children’s
memory talk’ (Harley and Reese 1999: 1338). Individual differences in maternal expositional
style – for example, using rich versus low levels of elaboration – strongly predicts how the chil-
dren of these mothers tend to narrate episodes of their pasts (Fivush and Fromhoff 1988; Fivush
et al. 2008).
Putting all of this together, it becomes clear why defenders of SIT hold that it is only after
children become accomplished narrators in ways that go significantly beyond their first efforts –
say, of narrating sequences of action in pretend play and dealing with such sequences in pic-
ture books – that they get into a position to ‘view the self as having a specific experiential
history that is different from others and thus [is] a specific personal past and a possible specific
future’ (Nelson 2003: 30).5
In sum, by SIT’s lights, children only come to be able to think about their pasts in autobio-
graphical terms towards the end of their pre-school years, because they have begun to master
narrative practices through which they ‘are exposed to an ever-widening circle of under-
standing people as temporally extended persons with temporally extended minds’ (Nelson and
Fivush 2006: 242).
And, SIT’s story does not end there. It draws on further evidence – mainly from studies on
Western populations – to defend the view that narrative skills become further refined during
adolescence (see Fivush et al. 2011 for an overview, and also Fivush and Buckner 2003; Reese
and Fivush 2008). It is during this later period of development that autobiographical narratives
become temporally well structured and begin to more fully respect local canons about how to
tell the story of one’s life across time. These late maturing autobiographical efforts exhibit a
much more stable personal frame of reference and are strongly influenced by the local narrative
forms and templates.
Developmental psychologists have been especially attracted to SIT because of its potential
to explain the well-documented phenomena of childhood, or infantile amnesia – the fact that
most of us are incapable of recalling events in our personal history that occurred before the
age of around 3 years (Schachtel 1947; Eisenberg 1985; Fivush et al. 1995; Harley and Reese
1999). Explanations in terms of SIT seek to account for our remarkable inability in this regard
by appealing to the fact ‘something dramatic changes in memory during the years from two to
five that makes remembering specific events from one’s past life both feasible and of value for
the individual’ (Nelson 2007: 185).6
That something dramatic happens during this period is hard to deny, but it provokes the
questions: what exactly changes? How dramatic is the change? And how precisely should we
understand it?
Read conservatively, the evidence accumulated by proponents of SIT might be thought to
show, minimally, that the mastery of narrative practices significantly adds to existing cognitive
capacities, enhancing – albeit in important respects – memory capacities that are already in
place. So construed, social interactive narrative practices, though important as scaffolds, would
not be responsible for making autobiographical memory possible per se. Instead, the mastery
195
Daniel D. Hutto
of narrative practices puts children in a position to enhance and improve their recall of facts by
constructively correcting or filling in the ‘gaps’ in the contents of their memory (Barnier and
Sutton 2008: 179).
On a weak reading of SIT, mastery of narrative practices help children to recover, add to,
improve and cement whatever content memories they already have, even if those contents
are only at first sketchy or only seen through a glass darkly. In line with this, learning how to
narrate the past along with others may help children to remember things more accurately or
clearly and help make memories stick and become more durable, but the narrative capacities
would not make it possible to have memories with autobiographical content. On this construal,
memories with autobiographical content would be initially formed and stored, even though
they would be hard to faithfully recover until they are embedded in narratives. Such a weak
reading is consistent with the SIT assumption that acquiring a capacity for narrating the past is
something individuals can only normally gain by interacting socially with others. This weaker,
enhancement reading of SIT, therefore, sponsors a more modest understanding and explana-
tion of childhood amnesia, according to which ‘the reason why adults cannot remember any
particular events that happened during their early childhood is that their own memory, at the
time, was not geared up for retaining memories of such particular events’ (Hoerl 2007: 625,
emphasis added).
Yet SIT is more frequently advanced as a more radical claim – viz. that the mastery of
narrative practices makes possible a wholly new, and unprecedented, kind of memory (Nelson
2007: 184; Fivush et al. 2011: 322). Unlike the enhancement view, the strong reading of SIT
holds that prior to gaining the capacity to construct meaningful narratives about the past there is
no possibility of representing the past autobiographically at all. Fivush and Nelson advocate this
stronger reading when they write:
until children begin talking with others about what they have experienced in the past,
or about the experiences of others at some other time, young children do not represent
themselves in past lives or project themselves into possible futures . . . parent-guided reminisc-
ing, specifically about internal states, structures children’s developing understanding of
their own and other’s past and allows for [/makes possible] children’s construction of
a temporally extended understanding of self and other.
(Fivush and Nelson 2006: 235, emphasis added)
Pressing for the strong reading of SIT, Fivush and Nelson maintain that ‘talk about the past, in
particular, is essential for children’s developing understanding that memory is a representation of
a past event’ (2006: 240, emphasis added). What this makes clear is that the strong reading of
SIT differs from the enhancement reading precisely in claiming that prior to mastering the rele-
vant narrative practices, there is no properly autobiographical content to personal memories. Strong
SIT claims that it is the mastery of narrative practices that provides children with the very first
opportunity to meaningfully recall the past in a properly autobiographical manner at all. Thus, if
strong SIT turns out to be true, it follows that there is no pre-existing autobiographical content
to our memories to be enhanced prior to our learning to narrate the past along with others.
3. Narrative fragments and the puzzle of pure episodic remembering

Can SIT account for non-narrative, and ex hypothesi non-autobiographical, kinds of episodic
memory? The question is pressing since humans – at least – are apparently capable of hav-
ing some kind of episodic memories before their mastery of the relevant narrative skills and
196
independently of their exercises of them. Yet since SIT claims that such skills must be brought
to bear if there is to be properly autobiographical remembering, it seems that there are capaci-
ties to remember things about our personal past in ways that (a) do not reduce the kinds of
non-declarative, non-narrative embodied forms of remembering (described in Section 1 of this
chapter), but which nevertheless fall short of (b) sophisticated, narratively based sorts of auto-
biographical remembering.
The urgent need for proponents of SIT to address this challenge is made evident if we
consider Nelson’s various formulations of the relations between autobiographical and episodic
memory (for a discussion, see Hoerl 2007: 623ff). On the one hand, Nelson speaks as if auto-
biographical memory were a sub-type of episodic memory – one that is marked out from other
kinds of episodic memory in that it makes special reference to the self. Hence, for this reason,
she cleanly distinguishes autobiographical memory from pure kinds of episodic memory (see
Nelson 2007: 186). This way of formulating the relation between autobiographical and epi-
sodic memory is problematic for the strong SIT she advocates since, if true, it would entail that
autobiographical memory isn’t truly a distinctive kind of memory at all, but only a special sub-
variety of episodic memory, albeit one that possesses a particular kind of content.
On the other hand, in places Nelson also writes as if the two notions were equivalent,
such as when she says: ‘episodic (and thereby autobiographical) memory’ (Nelson 2007: 184).
Equating these notions is even more problematic for SIT as long as it is accepted that episodic
memory is already present early on in development, prior to the development of narrative
practices. For in that case, the simple equation of the two notions would entail that references
to the self would have to be possible prior to developing the relevant socially scaffolded capaci-
ties that are needed to narrate the past. That would render SIT, in either version, manifestly
false. Or it would require, implausibly, holding that ‘episodic memory is not available in the
early years of life’ (Nelson 2007: 187). Thus clarifying the claims of SIT and evaluating their
prospects of being true requires getting clear about how episodic memory and autobiographical
memory might be related.
On one hand, helpfully neutral, construal episodic memory ‘refers, roughly, to the form of
memory responsible for allowing us to revisit specific episodes or events from the personal past’
(Michaelian 2016: 5). Episodic memory is often taken to be a form of declarative memory that
involves representing the past in explicit and consciously accessible ways (Michaelian 2016: 27).
However, a virtue of the neutral formulation of episodic memory supplied above is that it does
not require that all such memory is declarative in these respects. Nor, importantly – for reasons
that will become clear shortly – does it demand that episodic memory involves any explicit
representation of the self.
How then might we understand pure episodic remembering? A number of theorists have
proposed that episodic memory is simply a form of re-creative or simulative imagining that
enables us to construct and entertain possible episodes (De Brigard 2014; Michaelian 2016).7
Inspired by scientific developments around this topic, it has been claimed that there is no
intrinsic difference between remembering and imagining. Starkly put, the simulation theory of
episodic memory holds that ‘to remember, it turns out, is just to imagine the past’ (Michaelian
2016: 14, 120). According to this simulation theory, at its core, episodic remembering is really a
matter of generating ‘self-centered mental simulations about possible events that we think may
happen or may have happened to ourselves’ (De Brigard 2014: 173).
There is convincing empirical support for the hypothesis that there is a common cognitive
basis for acts of memory and imagination (Szpunar et al. 2007; Schacter et al. 2007; Schacter
and Addis 2009). Novel scientific work on mental time travel has repeatedly confirmed the
existence of some strong similarities in the patterns of neural activity associated with the sorts of
197
Daniel D. Hutto
cognitive procedures employed in thinking about our past and imagining our possible futures.
Acts of recall in which specific events or episodes are re-experienced – e.g., when we remem-
ber what it was like for us to enter a particular classroom for the first time – can be understood
in terms of acts of re-enactment imagination that involve neural reuse and reactivation. The
simulationist theory gains general support from the finding that the brain often re-uses its neural
apparatus to do various distinct kinds of cognitive work (Anderson 2010, 2014).
In line with this simulation theory, De Brigard hypothesizes that episodic remembering is
made possible by a particular operation of a larger cognitive system – an operation that enables
us to entertain hypothetical thoughts about possible happenings (De Brigard 2014: 177). Thus
he offers an account of episodic memory as ‘an integral part of a larger system that supports not
only thinking of what was the case and what potentially could be the case, but also what could have
been the case’ (De Brigard 2014: 158; original emphasis).
Drawing on insights from the popular Predictive Coding or Processing paradigm for think-
ing about cognition, De Brigard proposes that this sort of remembering consists in the optimal
reconstruction of a previous experience, where optimal reconstruction is understood as ‘a
retrieval process probabilistically constrained both by schematic knowledge and the frequency
of prior encounters with the target memory’ (De Brigard 2014: 171). He is attracted to this view
because of empirical findings that shows many ordinary memory distortions present events in
ways that are coherent and plausible (De Brigard 2014: 173).8 The claim is that at least when we
set our imaginations to the task of remembering the past we constrain them to generate plausible
episodes of what may or may not have happened – possibilities that are appropriate to the situ-
ation in question (De Brigard 2014: 179).
Gerrans identifies the larger system in question with the Default Mode Network (or
DMN), and he claims that in such cases the DMN provides imaginative simulations that can
serve as narrative elements (2014: 13). Indeed, Gerrans suspects that the DMN ‘evolved to
produce narrative fragments’ (Gerrans 2014: 17).
All of this is quite consistent with the suggestion that, when not constrained by tasks of
remembering the past or predicting the future, the DMN drifts into the free play of the imag-
ination witnessed in daydreaming, dreaming and delusion. Or, in other words, ‘When not
organised for problem solving the DMN reverts to the screensaver mode we experience as
daydreaming or mindwandering. In this mode there is no overarching goal to provide narrative
structure’ (Gerrans 2014: 62).
For our purposes, what is important is that even in episodic remembering the imaginative
episodes that constitute the narrative elements or fragments of which Gerrans speaks ‘are not
always assembled into full-scale narratives, they may remain fragmentary and episodic, but their
cognitive nature is to be the building block of a story assembled from subjective experiences’
(Gerrans 2014: 13).
Autobiographical memory proper is something more than pure episodic recall, where the
former is characterized as going beyond ‘recalling the who, what, where, and when of an event,
to include memory of how this event occurred as it did, what it means, and why it is important’
(Fivush et al. 2011: 322).
One distinctive characteristic of autobiographical memory is the particular way accounts of
the self get bound up in the recall of past happenings. Prior to being able to construct autobio-
graphical narratives, insofar as children are capable of a kind of episodic remembering, ‘there is
no distinct self associated with the contents of memory’ (Nelson 2007: 186). Nelson explains,
even though such a claim may seem counterintuitive to those who assume that all experience
necessarily entails having a sense of self, in case of pure episodic remembering there is no neces-
sity and no grounds for supposing that the self appears as a content of the remembering since
198
‘there is no contrast; there are no contents that are independent of the activities and interests of
the self, therefore the self is implicit in all, but not “self-evident”’ (Nelson 2007: 186).9
All in all, if we endorse the simulation vision of pure episodic memory, it is possible to hold
that prior to the development of autobiographical memory, younger children do have memo-
ries about novel events, ‘although their memory for the components of such experiences may
be scattered and fragmented, reflecting the character of their experience in terms of the limits of
short-term memory for connecting segments of extended experience’ (Nelson 2007: 186). And
to say this is consistent with holding that in their very early careers children are not capable of
‘re-experiencing a specific episode in the sense of thinking “I was there”; rather the data can be
explained simply in terms of recalling “things that happened”’ (Nelson 2007: 186).
Combining the simulative pure episodic remembering with SIT’s account of autobiographical
remembering can explain the fact that ‘when an adult does remember something from earlier than
age three, it is usually a fragment or a brief scene, not a full meaningful episode’ (Nelson 2007:
185). Given all of this, it should be unsurprising then that a major reason that we ‘talk about the
shared past with partners, friends, family, or colleagues in order to facilitate or tap what may be
only fragmentary, partial, or shrouded in our own memories’ (Barnier and Sutton 2008: 179).
Allying the strong SIT of autobiographical memory with a simulationist theory of episodic
memory has the further virtue of opening up the possibility of developing an even more fun-
damental defence of the idea that autobiographical remembering depends on scaffolded social
interactions. For it might be argued that it is only through the process of mastering linguistically
mediated practices that one encounters the kind of cognitive friction needed to learn how to
make contentful claims and, thus, for getting things right or wrong – about the past, or any
other topic – at all. If so, mastery of special socio-cultural practices may be required for more
fundamental reasons than merely to enable children to form representations about their personal past
which have properly autobiographical content. More pivotally, mastery of such practices might
be necessary to enable them to make any truth-evaluable, contentful claims about the past at all.
This thesis about the link between linguistic practice and contentful representation is far from
uncontroversial.10 Yet – if it turns out to be true – it would advance the fortunes of a strong
reading of SIT since it would entail that only individuals who acquire the ability to fashion auto-
biographical narratives about their past could have any properly meaningful contentful thoughts
about their memories at all.
Nor is it necessarily at odds with the assumption that purely episodic remembering is
grounded in simulative imaginings. Many theorists who defend a simulation theory of imagin-
ing hold that even though simulative imaginings make a bona fide cognitive difference, they are
nevertheless unlike familiar cognitive attitudes in that they lack inherent correctness or congru-
ence conditions.11 Thus even if we assume that episodic remembering is a kind of simulative
imagining, there need be no conflict in thinking that making truth-evaluable claims about one’s
personal past is a capacity that requires a special kind of social immersion and support.
4. The roles and functions of narrative remembering

A strong reading of SIT gains in credibility when considered against the backdrop of new,
empirically motivated thinking about the function of autobiographical memory. Combined,
scientific findings using various paradigms reveal the great extent to which our memories are
easily and powerfully susceptible to outside influences.
De Brigard, who provides for an extensive review and philosophical analysis of the impor-
tance of that empirical literature, holds that unless we are prepared to accept the fact that
misremembering is utterly pervasive (and often unnoticed), we are best advised to suppose that
199
Daniel D. Hutto
the main work of autobiographical remembering is something other than faithful represen-
tational reproduction. Thus, he holds ‘it is a mistake to think of memory as [a] system that is
uniquely – or even primarily – dedicated to reproducing the contents of previous experiences’
(De Brigard 2014: 177). In the light of the accumulated evidence it seems clear that ‘autobio-
graphical memory does not reproduce specific past events with precision’ (Campbell 2006: 363).
Looking afresh at what autobiographical memory does for us, these authors argue, is prefer-
able to supposing that it reliably fails to perform a biologically basic function which, given how
we know it actually operates, is ‘almost impossible to achieve’ (Campbell 2006: 365).
What alternative function, if not accurate representational reproduction, might autobi-
ographical memory have? There are grounds for thinking that it plays important roles in
establishing social cohesion in ways that are primarily about generating and reshaping self-
narratives in ways that help to regulate expectations and steer future planning and actions,
both collective and individual. The value of shared remembering, such as in the retelling of
family stories, typically provides something ‘besides or beyond accuracy’ (Barnier and Sutton
2008: 179). In such contexts, remembering together is less to do with representing the past
accurately and more to do with forging and maintaining ‘intimate and longstanding relationships’
(Barnier and Sutton 2008: 181).
Engaging in narrative practices, those that involve giving accounts in rich storied content,
has been shown to correlate positively with mental health. A number of findings demonstrate
that people ‘who are able to narrate the emotional events of their lives in more self-reflective
ways show better physical and psychological health’ (Fivush et al. 2010: 46; see also Fivush et al.
2003). In particular, it has been found that:
individuals whose narratives include more causal explanatory language (words such as
because, thus, and understand) and more emotional language (the inclusion of both posi-
tive and negative emotion words) subsequently show lower anxiety, lower depression,
higher sense of well-being, and higher immune system functioning than individuals
who use less of this kind of language.
(Fivush and McDermott-Sales 2006: 126)
Other research reveals that choice of narrative, but not necessarily its accuracy, is important to
our well-being, showing that ‘How we remember the stressful events of our lives has an impact
on our ability to cope’ (Fivush and McDermott-Sales 2006: 125; McDermott Sales et al. 2005).
This sort of coping is not a matter of recovering details of past happenings accurately so much
as a forward-looking basis for dealing better with other, similarly stressful future happenings.
These observations chime well with Campbell’s claim that autobiographical remembering
may be an activity that aims at shaping ‘the possibilities for how we go on’ (Campbell 2006: 372).
Indeed, over-emphasis on the idea that memory is fundamentally about representational fidelity
has surely contributed to little attention having been paid to the affective dimensions and social
benefits of autobiographical memory and remembering together.
Yet even if we abandon the one-sided picture of the primary function of memory as accu-
rately representing past happenings, none of these observations should lead us to suppose that
autobiographical remembering never involves attempting to get at and establish the truth about
what happened in the past faithfully.
Getting claims about the past right is, undoubtedly, one of the central things we hope to
achieve in our acts of autobiographical remembering. Nor is there any essential conflict between
our attempts to seeking the truth about the past via autobiographical remembering and the fact
that such remembering takes a narrative format. For some narratives aim at conveying truths.
200
Taking everything into account, the lesson we should derive from reflection on the col-
lected sum of empirical findings is that remembering is open to the influences of others. We
are not entitled to conclude, however, that such influences need always be wholly negative.
Remembering with, or under the influence of, others can cut both ways to produce both ‘ben-
eficial and troubling outcomes’ (Barnier and Sutton 2008: 179). On the one hand, the evidence
clearly shows that autobiographical remembering is vulnerable to misinformation, distortion and
corruption. Yet, on the other hand, it is surely possible that reminiscing with the support of
others can – at least under the right circumstances – enable us to ‘successfully renegotiate the
emotional significance of a shared past experience, to arrive at both a more accurate picture of
the past and a more fruitful conception of current self and other’ (Barnier and Sutton 2008: 179).
This rethink of the functions of autobiographical memory fit with SIT, even in its strong-
est version. For – as noted at the end of the previous section – it is possible that we are only
able to raise questions about the truth or accuracy of our memories once we have learned the
ropes of a particular kind of claim-making narrative practice. If so, it will not just be because
children, until they have mastered the relevant narrative practices, cannot form representations
with autobiographical content but, more fundamentally, because until they have mastered such
practices, children are not in a position to make claims that represent their pasts in ways that can
be true or false, full stop.
It may be that getting at the truth of past happenings is not fundamentally unlike getting
about the truth of things in any other domain. Seeking the truth about our past is a sophisticated
business, one we are only able to conduct at all because of our familiarity with the norms of
this peculiar social enterprise. Typically, success in this endeavour requires the care and effort of
many people. Yet even when conducted alone, we must call on socially instituted norms and
procedures since, ‘One of the demands of recollection is that we are prepared to be critically
attentive to the concepts, narratives, feelings, and self-conceptions through which we experi-
ence the past’ (Campbell 2006: 374).12 Consequently, just as a strong variant of SIT would have
it, there is reason to think that being in a position to make claims about the past, let alone to
determine their veracity, is not a solo feat that can be performed by individuals by dint of their
unenculturated, biologically basic cognitive hardware alone.
5. Conclusion
All told, even though it may require us to adjust some long-held philosophical intuitions
about the character and functions of autobiographical and episodic memory, there are com-
pelling empirical and theoretical reasons for supposing that the exercise of narrative skills may
well be necessary in enabling us to remember our personal pasts in contentful ways. Even if
this proves true, this fact would be far from the whole story about memory. Nevertheless, the
possibility is an important one that deserves greater philosophical attention.
Notes
1 SIT has considerable empirical backing of a wealth of findings ‘accumulated in psychological experi-
ments over more than 100 years’ (Nelson 2007: 185). At a minimum, it is supported by ‘research on
parent–child reminiscing that extends over more than a 20-year period’ (Fivush and Nelson 2006: 235).
2 Or, as Fivush and Nelson put it, ‘the developmental “mechanism” requires linguistically mediated social
interaction’ (2006: 236).
3 In defending this idea Nelson identifies ‘three essential components of narrative [that] . . . remain weakly
or non-existent in most of the narrative productions of the 3- to 5-year-old preschooler: temporal per-
spective, the mental as well as physical perspective of self and of different others, and essential cultural
knowledge of the unexperienced world’ (2003: 28).
201
Daniel D. Hutto
4 Nelson discusses this phenomenon with reference to work by Miller et al. (1990) that provides
‘observations of appropriation by a child from a parent’s account. For example, one small boy who
was told by his mother in a cautionary way about having fallen off a stool when she was young retold
this story later as his “when he fell off the stool”. Miller also reports children in preschool freely
appropriating’ (Nelson 2003: 31).
5 One barrier to an early onset of this capacity is that ‘at 2 years of age the child has only a few of the rudi-
ments that enter into narrative’ (Nelson 2003: 27). At the early stages in the child’s career ‘language is being
learned and used but it is not yet a vehicle for conveying the representation of narrative’ (Nelson 2003: 27).
6 Developmental psychologists reliably report that, ‘the period from 3 to 5 years appears to be one espe-
cially crucial phase of transition in the development of children’s conceptions of persons’ (Richner and
Nicolopoulou 2001: 398).
7 That memory is fundamentally creative and reconstructive is not a new idea. As Schacter and colleagues
remind us, ‘the general idea that memory is a constructive process . . . rather than a literal replay of the
past, dates to the pioneering work of Bartlett (1932), and has been developed by a variety of investiga-
tors who have demonstrated the occurrence of memory distortions and theorized about their basis’
(Schacter et al. 2012: 681).
8 Citing research by Pezdek and colleagues (1997), De Brigard reports that ‘Most ordinary cases of mis-
remembering have an air of plausibility to them’ (2014: 163).
9 As such, acquiring a sense of self as a person with a particular past is ‘dependent upon language used
to exchange views about self and other, primarily through narratives but also through commentary on
the self by others, as well as on their own feelings, thoughts and expectations of what might happen’
(Nelson 2003: 33). Thus acquiring a sense of self of the sort needed for autobiographical remembering
is something that happens ‘during the critical years when the child can enter fully into the linguistic
world but is not yet a participant in formal schooling’ (Nelson 2003: 22).
10 Various philosophers have defended the thesis that contentful thought and talk depends on mastery of
special linguistic practices on purely conceptual grounds (see, e.g. Davidson 1984; Price 2013). For a
naturalistically motivated defence of this idea, see Hutto and Satne (2015).
11 For example, Gerrans maintains that ‘qua simulations imaginative states do not have congruence condi-
tions’ (2014: 105; see also p. 18). Langland-Hassan tells us, ‘Much of what has been said about sensory
imagination conflicts with the idea that imaginings have substantive correctness (or veridicality, or
accuracy) conditions at all’ (2015: 665). Michaelian holds that reconstructive memory retrieval trades in
representations with sensory content which are ‘nonpropositional, in the sense that it cannot be evalu-
ated for truth and falsity in a binary manner’ (Michaelian 2016: 53). For a more detailed discussion of
the question of imaginative content, see Medina (2013) and Hutto (2015).
12 For Campbell, recollection is as a complex social activity, one in which ‘taking care to get the past
right involves at least our implicit accountability to others’ (2006: 377). In this sense, determining
‘memory’s faithfulness to the past is, in many cases, a complex epistemological/ethical achievement’
(Campbell 2006: 363).
Related topics
• Habit memory
• Memory and truth
Bibliography
Anderson, M. 2010. Neural reuse: A fundamental organizational principle of the brain. Behavioral and Brain
Sciences. 33: 245–313.
Barnier. A. J. and Sutton, J. 2008. From individual to collective memory: Theoretical and empirical
perspectives. Memory. 16(3): 177–82.
Bartlett, F. C. 1932. Remembering: A study in experimental and social psychology. Cambridge: Cambridge
University Press.
Campbell, S. 2006. Our faithfulness to the past: Reconstructing memory value. Philosophical Psychology.
19(3): 361–80.
Casey, E. 1987. Remembering: A phenomenological study. Bloomington, IN: Indiana University Press.
202
Davidson, D. 1984. Inquiries into truth and interpretation. Oxford: Clarendon Press.
De Brigard, F. 2014. Is memory for remembering? Recollection as a form of episodic hypothetical thinking.
Synthese. 191(2): 155–85.
Eisenberg, A. R. 1985. Learning to describe past experiences in conversation. Discourse Processes. 8: 177–204.
Fivush, R. 1997. Event memory in early childhood, in N. Cowan (ed.), The development of memory in child-
hood (pp. 139–162). London: University College Press.
Fivush, R. and Buckner, J. P. 2003. Constructing gender and identity through autobiographical narra-
tives, in R. Fivush and C. A. Haden (eds.), Autobiographical memory and the construction of a narrative self:
Developmental and cultural perspectives (pp. 149–68). Mahwah, NJ: Erlbaum.
Fivush, R. and Fromhoff, F. 1988. Style and structure in mother–child conversations about the past.
Discourse Processes. 11: 337–55.
Fivush R. and McDermott-Sales, J. 2006. Coping, attachment, and mother–child narratives of stressful
events. Merrill-Palmer Quarterly. 52(1): 125–50.
Fivush, R. and Nelson, K. 2004. Culture and language in the emergence of autobiographical memory.
Psychological Science. 15: 573–77.
Fivush, R. and Nelson, K. 2006. Parent–child reminiscing locates the self in the past. British Journal of
Developmental Psychology. 24: 235–51.
Fivush, R. and Reese, E. 1992. The social construction of autobiographical memory, in M. A. Conway,
D. Rubin, H. Spinnler and W. Wagenaar (eds.), Theoretical perspectives on autobiographical memory
(pp. 115–134). Dordrecht: Kluwer Academic.
Fivush, R. and Wang, Q. 2005. Emotion talk in mother–child conversation of the shared past: The effects
of culture, gender, and event valence. Journal of Cognition and Development. 6: 489–507.
Fivush, R., Bohanek, J. G. and Zaman, W. 2010. Personal and intergenerational narratives in relation to
adolescents’ well-being, in T. Habermas (ed.), The development of autobiographical reasoning in adolescence
and beyond: New directions for child and adolescent development. Wiley: 131, 45–57.
Fivush, R., Haden, C. and Adam, S. 1995. Structure and coherence of preschoolers’ personal narratives
over time: Implications for childhood amnesia. Journal of Experimental Child Psychology. 60: 32–56.
Fivush, R., McDermott Sales, J. and Bohanek, J. G. 2008. Meaning making in mothers’ and children’s
narratives of emotional events. Memory. 16(6): 579–94.
Fivush, R., Habermas, T., Waters, T. E. A. and Zaman, W. 2011. The making of autobiographical mem-
ory: Intersections of culture, narratives and identity. International Journal of Psychology. 46(5): 321–45.
Fivush, R., Berlin, L., Sales, J. M., Mennuti-Washburn, J. and Cassidy, J. 2003. Functions of parent–child
reminiscing across negative events. Memory. 11: 179–92.
Gerrans P. 2014. The measure of madness. Cambridge, MA: MIT Press.
Harley, K. and Reese, E. 1999. Origins of autobiographical memory. Developmental Psychology. 35:
1338–48.
Herman, D. 2013. Storytelling and the sciences of the mind. Cambridge, MA: MIT Press.
Hoerl, C. 2007. Episodic memory, autobiographical memory, narrative: On three key notions in current
approaches to memory development. Philosophical Psychology. 20(5): 621–40.
Hoerl, C. and McCormack, T. 2005. Joint reminiscing as joint attention to the past, in N. Eilan, C. Hoerl,
T. McCormack and J. Roessler (eds.), Joint attention: Communication and other minds: Issues in philosophy
and psychology (pp. 260–286). Oxford: Oxford University Press.
Hutto, D. D. 2015. Overly enactive imagination? Radically re-imagining imagining. Southern Journal of
Philosophy. 53(S1): 68–89.
Hutto, D. D. and Satne, G. 2015. The natural origins of content. Philosophia. 43(3): 521–36.
Langland-Hassan, P. 2015. Imaginative attitudes. Philosophy and Phenomenological Research. 90(3): 664–86.
McDermott Sales, J., Fivush, R., Parker J. and Bahrick, L. 2005. Stressing memory: Long-term rela-
tions among children’s stress, recall and psychological outcome following Hurricane Andrew. Journal of
Cognition and Development. 6(4): 529–45.
Medina, J. 2013. An enactivist approach to the imagination: Embodied enactments and fictional emotions.
American Philosophical Quarterly. 50(3): 317–35.
Michaelian, K. 2011. Is memory a natural kind? Memory Studies. 4(2): 170–89.
Michaelian, K. 2016. Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge,
MA: MIT Press.
Miller, P. J., Potts, R., Fung, H., Hoogstra, L. and Mintz, J. 1990. Narrative practices and the social con-
struction of the self in childhood. American Ethnologist. 17: 292–311.
Moran, R. 1994. The expression of feeling in imagination. Philosophical Review. 103(1): 75–106.
203
Daniel D. Hutto
Nelson, K. 2003. Narrative and the emergence of a consciousness of self, in G. D. Fireman, T. E. J. McVay
and O. Flanagan (eds.), Narrative and consciousness (pp. 17–36). Oxford: Oxford University Press.
Nelson, K. 2007. Young minds in social worlds: Experience, meaning and memory. Cambridge, MA: Harvard
University Press.
Nelson, K. and Fivush, R. 2004. The emergence of autobiographical memory: A social cultural develop-
mental theory. Psychological Review. 111: 486–511.
Pezdek, K., Blandon-Gitlin, I. and Gabbay, P. 2006. Imagination and memory: Does imagining implausible
events lead to false autobiographical memories? Psychonomic Bulletin and Review. 13(5): 764–9.
Pezdek, K., Finger, K. and Hodge, D. 1997. Planting false childhood memories: The role of event plausibility.
Psychological Science. 8: 437–41.
Price, H. 2013. Expressivism, pragmatism and representationalism. Cambridge: Cambridge University Press.
Reese, E. and Fivush, R. 2008. Collective memory across the lifespan. Memory. 16: 201–12.
Reese, E., Haden, C. A. and Fivush, R. 1993. Mother–child conversations about the past: Relationships
of style and memory over time. Cognitive Development. 8: 403–30.
Richner, E. S. and Nicolopoulou, A. 2001. The narrative construction of differing conceptions of the
person in the development of young children’s social understanding. Early Education and Development.
12: 393–432.
Roediger III, H. L. 1980. Memory metaphors in cognitive psychology. Memory and Cognition. 8(3): 231–46.
Schachtel, E. 1947. On memory and childhood amnesia. Psychiatry. 10: 1–26.
Schacter, D. L. and Addis, D. R. 2009. On the nature of medial temporal lobe contributions to the con-
structive simulation of future events. Philosophical Transaction of the Royal Society B: Biological Sciences.
364: 1245–53.
Schacter, D. L., Addis, D. R. and Buckner, R. L. 2007. Remembering the past to imagine the future: The
prospective brain. Nature Reviews Neuroscience. 8: 657–61.
Schacter, D. L., Addis, D. R., Hassabis, D., Martin, V. C., Spreng, R. N. and Szpunar, K. K. 2012. The
future of memory: Remembering, imagining, and the brain. Neuron. 76(4): 677–94.
Schacter, D. L. and Tulving, E. 1994. What are the memory systems of 1994?, in D. L. Schacter and
E. Tulving (eds.), Memory systems (pp. 1–38). Cambridge, MA: MIT Press.
Sutton, J. 2003. Truth in memory: The humanities and the cognitive sciences, in I. McCalman and Anne
McGrath (eds.), Proof and truth: The humanist as expert. Canberra: Australian Academy of the Humanities.
Sutton, J. 2010. Observer perspective and acentred memory: Some puzzles about point of view in personal
memory. Philosophical Studies. 148: 27–37.
Sutton, J. and Williamson, K. 2014. Embodied remembering, in L. Shapiro (ed.), The Routledge handbook
of embodied cognition. London: Routledge.
Szpunar, K. K., Watson, J. M. and McDermott, K. B. 2007. Neural substrates of envisioning the future.
Proceedings of the National Academy of Science USA. 104: 642–47.
Thompson, C. P., Skowronski, J. J., Larsen, S. F. and Betz, A. L. 1996. Autobiographical memory: Remembering
what and remembering when. Mahwah, NJ: Erlbaum.
Tulving, E. 1972. Episodic and semantic memory, in E. Tulving and W. Donaldson (eds.), Organization of
memory (pp. 381–402). New York: Academic Press.
Wade, K. A., Garry, M., Read, D. and Lindsay, D. S. 2002. A picture is worth a thousand lies: Using false
photographs to create false childhood memories. Psychonomic Bulletin and Review. 9(3): 597–603.
204
PART V
Memory and time

16
MEMORY AND THE
CONCEPT OF TIME
Christoph Hoerl
In The Problems of Philosophy, Bertrand Russell writes of a form of memory the essence of which
is constituted by ‘having immediately before the mind an object which is recognized as past’
(Russell 1912: 66), and adds the following:
But for the fact of memory in this sense, we should not know that there ever was a
past at all, nor should we be able to understand the word ‘past’, any more than a man
born blind can understand the word ‘light’.
(Russell 1912: 66)
A similar passage can already be found earlier in the same book:
It is obvious that we often remember what we have seen or heard or had otherwise
present to our senses, and that in such cases we are still immediately aware of what we
remember, in spite of the fact that it appears as past and not as present. This immediate
knowledge by memory is the source of all our knowledge concerning the past: with-
out it, there could be no knowledge of the past by inference, since we should never
know that there was anything past to be inferred.
(Russell 1912: 26)
It is plausible to interpret these remarks as being concerned with the particular type of memory
that psychologists, and increasingly also philosophers, refer to as episodic memory – that is, the
distinctive capacity we have for recollecting particular past events that we have personally expe-
rienced.1 Thus interpreted, Russell’s claim is that the possession of episodic memory is necessary
for a grasp of the concept of the past, and, by extension, a full understanding of the concept of
time. Understanding what it is for there to have been a past, and for events to have happened in
the past, depends on the possession of episodic memory. This is the claim I wish to investigate
in this chapter. I refer to it as the Dependency Thesis.2
207
Christoph Hoerl
1. The idea of episodic memory as a generative capacity

Suppose the Dependency Thesis is true. If the possession of episodic memory is indeed
necessary for a grasp of the concept of the past, how should we think of the connection
between the two? There is a certain conception of the general nature of memory that, at least
prima facie, seems to be in tension with the thought that memory could play a role in our grasp
of a particular concept. That is the conception of memory, in general, as being preservative, i.e.
as preserving knowledge that has been acquired by other means, rather than itself generating or
being a source of knowledge.
As applied to semantic memory, the idea that memory is preservative in this sense possesses a
great deal of plausibility. This is the kind of memory that simply involves the retention of knowl-
edge of propositions such as ‘Paris is the capital of France’, or ‘Caesar crossed the Rubicon’.
In order to acquire and retain such knowledge, the subject must already be able to grasp the
concepts that figure in the relevant propositions; otherwise she won’t understand and be able to
commit them to memory in the first place. (Note that this applies also to the contribution the
past tense form of the verb makes to the meaning of the proposition in the second example.)
Thus, semantic memory, due to the way in which it simply preserves propositional knowledge,
doesn’t seem to be the kind of psychological faculty that could explain how we come to have
some specific concept in the first place. On these grounds, one might think that if episodic
memory, too, was purely preservative, it also couldn’t play any role in our grasp of concepts,
including the concept of the past.
Obviously, one possible way in which one might react to this perceived tension is by simply
denying that episodic memory is (purely) preservative, and instead arguing that there is also an
epistemically generative aspect to it. There are in fact a number of theories of episodic memory
that effectively imply that recollecting an event in episodic memory makes available to the sub-
ject some kind of information that goes beyond (or is at least different from) the information
that was available to her when she experienced the event.3 Here, for instance, is Gareth Evans
writing about the possibility of a form of memory which ensures
the subject’s possession of a non-conceptual informational state, whose content corre-

sponds in a certain respect with that of some earlier state of the subject (a perceptual
state); although its content differs from that of the antecedent perceptual state in that,
if the subject is in the memory state, it seems to him that such-and-such was the case.
(Evans 1982: 239; original emphasis)
Does adopting a conception of episodic memory as epistemically generative help in fleshing out
the Dependency Thesis? The traditional way in which something like the general idea sketched
by Evans has usually been construed is in terms of what is sometimes referred to as the empiricist
theory of memory. According to this theory, there is supposedly something about the state in
which the subject finds herself when she remembers – a specific feature of the phenomenology
of being in that state – that indicates to her that that state represents a past event. Prominent
examples of this type of approach come from Hume (1739–40), who appeals to a difference in
‘force and vivacity’ between perception and memory, or Russell (1921), who appeals to the idea
that feelings of ‘familiarity’ and ‘pastness’ accompany memories, which can act as an indicator
that we are remembering something that happened in the past. (See also Chapters 13 and 24
of the present volume on views of this type.) Quite apart from a host of well-known problems
that the empiricist theory of memory faces, though,4 such a theory also in fact seems to be of
208
Memory and the concept of time
no help when it comes to trying to flesh out the Dependency Thesis, i.e. the idea that memory
might have a role to play in our very grasp of the concept of the past.5 As David Pears writes,
discussing in particular Russell’s theory:
No doubt, familiarity is a felt property of certain images. But it suggests the proposi-
tion, ‘I have experienced something like this before’ only to someone who already
possesses the concept ‘past’. Presumably, the same is true of whatever property Russell
means by the ‘pastness’ of an image. In general, such properties of images get their
names from the propositions which they suggest, and they suggest propositions con-
taining the concept ‘past’ only to those who have already learned the meaning of the
word ‘past’ at closer quarters.
(Pears 1975: 238)
Thus, the empiricist theory of memory does not provide a suitable framework for fleshing out
the Dependency Thesis. Rather than explaining what enables a subject to grasp the concept of
the past, it presupposes a grasp of the concept of the past on the part of the subject, which she
can employ in inferring from some present feature of her mental state that it represents some
past occurrence.
Having said that, it is doubtful anyway whether Evans, in writing the passage I quoted
before, was thinking of an account of episodic memory along empiricist lines, in the sense
just sketched. Immediately after it, he explicitly distances himself from the idea of memory as
involving ‘free-floating images whose reference to the past is read into them by reasoning on
the part of the subject’ (Evans 1982: 239), which can be read as an attempt to distinguish the
account he is putting forward from an account of the type that is the target of Pears’s criticism.
Instead, it seems, we are to envisage the ‘informational system’ directly putting the subject into
a state of it seeming to her that such-and-such was the case, without this requiring any inference
on her part.
This gives us an alternative way in which episodic memory might be conceived of as being
epistemically generative, in the sense of making available to the subject information that differs
from the information available to her when she experienced the remembered event. Yet, on
closer inspection, it is far from obvious how we are to conceive of the idea of a non-conceptual
informational state with a past-tensed content. Note in particular that it is important not to
conflate this idea with the idea of states involved in information processing that are sensitive to
time. The idea of the latter possesses a great deal of plausibility – indeed, postulating such states
seems indispensable to explaining timing abilities possessed by both humans and animals. Work
in psychology seems to demonstrate the existence of a number of timing mechanisms that make
us sensitive, say, to the length of temporal intervals. But the existence of such mechanisms
alone, even on the assumption that they play some role in episodic memory, is not enough to
provide for the idea of information processing simply putting the subject into a state of it seem-
ing to her that such-and-such was the case.6 Christopher Peacocke, who holds a view according to
which such timing mechanisms do play a crucial role in episodic memory, acknowledges this
point when he raises the question as to what the difference is ‘between being merely responsive
to the temporal interval which has elapsed since a particular event occurred, and taking it as
a temporal interval’. And he goes on to say, ‘In the latter case, the event at the far end of the
interval is assigned a position in one’s history, and, correlatively, in the history of the world’
(Peacocke 2001: 364). That is to say, taking an interval that has elapsed as a temporal interval
that has elapsed requires, beyond some capacity to be causally sensitive to its length, also the
209
Christoph Hoerl
ability to locate the event at the start of that interval in the past. As such, it seems once again
to presuppose an existing understanding of the concept of the past, rather than being able to
explain the existence of such an understanding.
Setting aside these difficulties with particular ways of construing the idea of episodic memory
as epistemically generative, there is also a more general issue with trying to press this idea into
service in attempting to flesh out the Dependency Thesis. I said earlier that one reason one
might have for thinking that episodic memory is generative in this context is in order to account
for the difference between episodic and semantic memory, and for how the former, in contrast
to the latter can have a role to play in our grasp of concepts, specifically the concept of the past.
Yet, as G.E.M. Anscombe effectively argues, if the capacity to retain beliefs about the past in
semantic memory can’t explain the subject’s ability to grasp what a thing’s having taken place in
the past consists of in the first place, it is in fact of no more use trying to explain this ability by
appealing to the idea that one simply finds oneself with such beliefs as the result of the workings
of memory. As she puts it:
If the question is ‘What does it mean to say that such and such happened?’ one is not
helped to answer it by saying ‘It did happen’; nor is one helped by saying ‘I have the
idea of its having happened, without being told and as a witness’. For the question is
‘What is the idea?’
(Anscombe 1981: 109; original emphasis)
Thus – somewhat paradoxically, perhaps – attempts to account for the difference between
episodic and semantic memory by arguing that the former is not purely preservative can actu-
ally end up diminishing the difference between the two forms of memory that is of potential
relevance to the Dependency Thesis. They do so if they end up construing both episodic and
semantic memory as putting a subject in a position to know, at a certain time, that a certain fact
about the past obtains, whilst telling us nothing about how the subject can come to grasp what
it is for facts of the relevant type to obtain in the first place.7
2. Episodic memory and grasp of time as a domain

Adopting a conception of episodic memory as epistemically generative, thus, provides at least no
straightforward strategy for fleshing out the Dependency Thesis (and may also come with other
problems). It is therefore worth asking whether there is perhaps a way of construing episodic
memory, too, as essentially preservative – albeit in a different way from semantic memory – that
is nevertheless compatible with the idea that it has an essential role to play in our grasp of the
concept of the past, or what I have called the Dependency Thesis?
I started this chapter with some quotations from Russell’s The Problems of Philosophy illustrating
the Dependency Thesis. Russell’s discussion of memory in The Problems of Philosophy is set
within the wider context of a distinction he draws between two forms of knowledge: know-
ledge by acquaintance and knowledge by description. More specifically, the thought that there is a
form of memory that involves ‘having immediately before the mind an object which is recog-
nized as past’ (Russell 1912: 66) – which he singles out as the one at issue in the Dependency
Thesis – turns, for him, on the idea that this form of memory involves knowledge by acquaint-
ance, rather than just knowledge by description.
This suggests an alternative way of conceiving of what episodic memory consists in, and
how it differs from semantic memory. Moreover, as M.G.F. Martin (2001) has pointed out, it
allows us to draw a distinction between episodic memory and semantic memory, whilst at the
210
same time conceiving of both of them as preservative. What makes episodic memory distinctive,
as Martin puts it, is that it preserves the distinctive kind of ‘cognitive contact’ with events that
perception provides us with. Perceptual experience, the thought is, does not just typically furnish
us with certain items of propositional knowledge, it also puts us in a relation of acquaintance
with the experienced events and states of affairs themselves (in a way that, e.g. just reading about
them doesn’t), and this acquaintance with the particular events or states of affairs we witnessed
is retained or preserved in episodic memory. In this way, we arrive at an account that lets us see
how the knowledge involved in episodically recollecting an event ‘simply is the knowledge I
had as an eyewitness, maintained in being’, as Michael Dummett (1993: 414f.) puts it, though
knowledge of a different sort than that maintained in being through semantic memory.
What becomes of the Dependency Thesis, if it is viewed against the background of this kind
of view of episodic memory? Why might one think that memory, thus conceived, has a central
role to play in our grasp of the concept of the past? I think here it might help to contrast two
different pictures of what grasping the concept of the past comes to in the first place, and the
potentially different role they might assign memory in grasp of such a concept.
For an anti-realist about the past, understanding a sentence in the past tense is essentially a
matter of knowing how it could be verified, i.e. recognizing what would count as (present or
future) evidence for its truth. The anti-realist is motivated by the thought that any account of
the meaning of linguistic expressions must render intelligible how we can acquire a grasp of
that meaning, and manifest such a grasp in our behaviour. She then answers this question by
construing acquiring such a grasp as a matter of learning to endorse or reject sentences in which
the relevant expressions figure in response to certain pieces of evidence, and manifesting such a
grasp as a matter of demonstrating this sensitivity to the relevant evidence.
Memory is likely to play a prominent role in the anti-realist’s account of our grasp of the
meaning of the past tense, as a paradigmatic example of something that puts us in a position to
verify sentences in the past tense. Yet, this is not to say that anti-realism provides a congenial
context for framing the Dependency Thesis. For it is at least not obvious what grounds the anti-
realist might have for making the stronger claim that a grasp of the meaning of the past tense
requires possession of memories, as we obviously think that there can be other forms of evidence
for the truth or falsity of past-tensed statements that a subject can learn to recognize.8
What the anti-realist about the past denies, and what realists about the past, by contrast, assert,
is that we have a grasp of the meaning of the past tense that allows for the idea of truths about
the past that we are in no position now, or at any point in the future, to establish as such. As
John McDowell explains the point:
the conception which the realist claims the right to ascribe is the conception of a kind
of circumstance. He claims the right to ascribe it on the basis of behaviour constru-
able as a response to some instances of the kind, in spite of the admitted fact that other
instances, on his view, are incapable of eliciting any response from the possessor of the
conception.
(McDowell 1978: 139; original emphases)
So the dispute between the anti-realist and the realist turns on the kind of grasp of the meaning
of the past tense we can ascribe to ourselves. In particular, crucial to the realist’s picture is the
thought, as we might put it, of time as a domain in which events can be located, irrespectively of
what evidence of their occurrence there is at other locations within the same domain. As John
Campbell (1997: 157) also puts it, the realist’s picture is one according to which ‘there is a single
temporal reality onto which all one’s various temporal perspectives face’, making it possible for
211
Christoph Hoerl
some of those temporal perspectives to reveal aspects of that reality that are inaccessible from
other temporal perspectives.9
It is because the realist operates with this particular conception of the past that there is also, I
believe, a reason for thinking that episodic memory has a distinctive role to play in the realist’s
picture, assuming a more central significance in her account of our grasp of the past tense than
the anti-realist’s picture allows for. Moreover, the particular conception of episodic memory
that appears to fit best with this role is precisely the one articulated at the beginning of this
section, of episodic memory as retained acquaintance with past events. The thought, in short,
would be that episodic memory, by preserving the relation of acquaintance with past events that
our own experience of them furnished us with, can ground what I referred to as a grasp of time
as a domain in which these events can be located alongside those that are now present, and also
alongside others of which we may have no cognizance.
Annette Baier can be seen to articulate a view along those lines when she writes that ‘it is
not an unimportant conceptual truth that memory is of times, while knowledge is of facts, and
that times cannot be discontinuous, as known facts can be disjoint’ (Baier 1976: 220). As the
context makes clear, the type of memory she has in mind here is, more specifically, episodic
memory, which she explicitly contrasts with ‘the sort of human memory which is just informa-
tion storage, just a capacity to regurgitate input’ (Baier 1976: 219). The point she can be seen
to be making is that an important dimension of the contrast between the two types of memory
(which I have been framing in terms of the contrast between episodic memory and semantic
memory) is that, whereas the latter presupposes a separate understanding of whatever informa-
tion is retained, and cannot explain such an understanding, the former can itself contribute to, or
be a manifestation of, understanding. Specifically, because of the way episodic memory allows
us to retain acquaintance with events we once witnessed, it can ground an understanding of a
‘single temporal reality’, as Campbell puts it, that affords different temporal perspectives on that
reality at different times.
It is in this way, then, that episodic memory might be thought to have a distinctive, and
essential, role to play in our grasp of the concept of the past, as the Dependency Thesis main-
tains. As I have sought to bring out, crucial to this view of the role of episodic memory in our
grasp of the concept of the past is the idea that episodically recollecting a past event is more than
just a matter of retrieving information about that event; in episodic memory, the thought is, we
are still in cognitive contact with past events themselves, and this is why episodic memory can
ground an understanding of the reality events happening at different times are all part of.
3. Is episodic memory necessary for our grasp of the concept of the past?
So far in this chapter, I have simply taken for granted something like Russell’s view that,
without episodic memory, ‘we should not know that there ever was a past at all, nor should
we be able to understand the word “past”, any more than a man born blind can understand
the word “light”’ (Russell 1912: 66). Instead, what I have looked at is the question as to how
exactly, on such a view, we should think of the role of episodic memory in our grasp of the
concept of the past.
Yet, one might of course question whether it is in fact true that a grasp of the concept of
the past requires episodic memory. One challenge to this claim comes from empirical studies
of the temporal reasoning abilities of amnesic patient KC, who has been described as having
no episodic memories whatsoever, but who can still describe features of time and use temporal
vocabulary in the correct way. As Carl Craver explains, KC
212
defines the future as ‘events that haven’t happened yet’ and the past as ‘events that
have already happened.’ He believes that it is possible to change the future, ‘by doing
different things,’ and that what happened in the past influences what happens in the
future. He believes that once an event is past it will always stay in the past, and that
it is not possible for someone to undo a murder at some time after the murder has
occurred.
(Craver 2012: 465)
I will conclude with some remarks about how the case of KC might be thought to bear on what
I have called the Dependency Thesis, i.e. the claim that episodic memory is necessary for a grasp
of the concept of the past.
First, note that there is a weaker and a stronger reading of the Dependency Thesis. On the
weaker reading, it expresses something like a developmental claim, or what we could call the
Developmental Dependency Thesis: without having episodic memories, the thought would
be, one cannot acquire a concept of the past, even though it is subsequently possible to retain
a grasp of that concept whilst losing the capacity to recollect particular past events in episodic
memory.10 One indication that Russell himself possibly had only this weaker reading in mind
in The Problems of Philosophy is that he compares the impossibility of grasping the concept of
the past without episodic memory to a congenitally blind person’s inability to understand the
word ‘light’, which suggests that he thought of both of these cases primarily in terms of the
acquisition of concepts.
Any evidence of a preserved grasp of temporal concepts in KC is of course perfectly consist-
ent with the Developmental Dependency Thesis. What is of more interest in studying amnesic
patients such as KC is whether a stronger thesis can also be sustained, which we might call
the Constitutive Dependency Thesis. According to this Constitutive Dependency Thesis, it
is not just the case that acquiring the concept of the past requires possessing episodic memories
at the time, retaining a full grasp of that concept also depends on retaining the ability to recall
past events in episodic memory. Here, too, the analogy Russell draws with the concept ‘light’
may be instructive. As plausible as it is that it is possible for a blind person to understand that
concept if they were previously sighted, it is arguably nevertheless also the case that their con-
tinuing to grasp that concept requires that at least some abilities connected to their previous
visual experiences remain intact. For instance, a continued grasp of the concept ‘light’ may
be thought to require the continued ability to call to mind or imagine experiences of light,
colours, etc.11 Otherwise, all that might remain is an ability to recognize certain sentences
about light as true, but knowledge of the phenomenon itself will be lost. In a similar vein, the
defender of the Constitutive Dependency Thesis can argue that there is a specific cognitive
capacity that needs to be preserved for a retained grasp of the concept of the past, and that this
capacity is episodic memory.
It is this second reading of the Dependency Thesis that the case of KC clearly puts some
pressure on. What a defender of the Constitutive Dependency Thesis would have to argue in
response is that, whilst KC may still be able to talk about various aspects of time, he has never-
theless lost a proper understanding of what time itself is. Obviously, though, a bland assertion
of this claim is unlikely to convince anybody, so the question is whether there is any evidence
at all to suggest that KC’s time-related reasoning abilities are impaired in this way. There is one
particular limitation in KC’s abilities that Craver and his colleagues did find, which might point
in such a direction. One specific conversation they report is concerned with KC’s understanding
of the notion of regret:
213
Christoph Hoerl
SR: What does it mean to regret something?

KC: Something you don’t like doing or wish that you hadn’t done.
SR: Can you name some things a person might regret?
KC: If someone lost a large sum of money.
SR: Do you have any regrets?
KC: I don’t think so.
SR: What do you regret most about your life?
KC: Nothing – I can’t think of anything.
SR: What are some things a person might do if they feel regretful about something?
KC: Try to make it right [. . .]
SR: Do you know Richard Nixon?
KC: Yes.
SR: Do you think he has any regrets in life?
KC: I don’t think so.
SR: Do you think your mother has regrets?
KC: No.
SR: What are some things that they might regret?
KC: Nothing.
SR: Can you describe how these people feel ‘inside’?
KC: Mad at themselves.
(Craver et al. 2014: 193–4)
Craver and colleagues (2014) interpret KC’s responses here as a sign that his ‘command of the
semantics of regret persists’, but it is at least possible to question how conclusive the evidence for
this is. For instance, losing money is only an occasion to feel regret if it happened through a fault
of one’s own, and people can also get mad at themselves out of frustration over their own limita-
tions. So these responses could be the result of semantic associations without full understanding.
More importantly, though, if KC’s understanding of the nature of regret is indeed perfectly
intact, it seems difficult to explain his apparently complete inability to think of anything he or
someone else he knows might regret. The latter seems particularly surprising in light of the fact
that, despite not being able to episodically recollect any of the events in question, he retains
knowledge of his own ‘thrill-seeking’ lifestyle prior to the onset of his amnesia (Rosenbaum
et al. 2005: 993), including the motorcycle accident that caused the brain damage responsible
for the amnesia, in which he rode his motorcycle off a highway exit ramp.
What is it about the capacity to feel regret, specifically, that might explain the particular dif-
ficulty KC has in figuring out which of the events in his or someone else’s life that he knows
about might be occasions for regret? In the context of arguing that KC’s ability to reason about
time remains fully intact Craver at one point suggests that what may be able to underpin a
preserved grasp of temporal concepts, despite complete loss of episodic memory, is retained
knowledge of the fact ‘that the past uniquely constrains the future’ (Craver 2012: 465). It does
indeed seem difficult to deny that KC is aware that there are certain truths about the past, and
that those truths have implications for what is the case now and for what can be done now and
in the future. However, it is at least arguable that paradigmatic cases of regret actually involve
the ability to reason about time in a way that is somewhat more sophisticated than this. What
is at issue in such cases is that, at the time when the individual made a certain choice that
subsequently led to a certain outcome, another course of action was also open to her, which
would or might have led to a more positive outcome. Representing this state of affairs requires
the ability to think about the particular time when the choice was made in what is sometimes
214
referred to as an ‘event-independent’ way (McCormack and Hoerl 2008). That is to say, it

requires being able to make the time when a certain event happened an object of thought in
its own right, as a time when a different event might have happened instead, rather than just
recognizing the existence of certain past tense truths and their implications for the present
(on this issue, see also Hoerl and McCormack 2016).
If it is this particular temporal reasoning ability that KC has difficulty with on account of
his amnesia – which would be one way of explaining his inability to think of anything he or
someone familiar to him might regret – then this might be one way of making concrete a sense
in which he has lost a full grasp of the concept of the past. More specifically, in line with what I
said in Section 2 of this chapter, KC might be described as lacking a properly realist conception
of the past. There I emphasized the particular role played within that conception by the idea of
time as a domain in which events can be located. As I also put it, following Campbell, the realist’s
picture is one on which there is a single temporal reality affording different temporal perspec-
tives on it at different times. Something like this picture, I take it, is also presupposed by the
capacity for event-independent thought about time, in which we latch on to a particular point
in time in thought, but think of it as a point in time at which various things could have hap-
pened. In other words, in such thinking, the past is not just thought of as something we know
certain truths of, in the form of past-tensed statements about certain events for which we still
have evidence. It is also thought of as a different region of time in its own right, each moment
of which was once present in the same way a different moment is now present, and which could
have contained a different history of events. If it is episodic memory that underpins – not just
developmentally, but constitutively – our ability to think of other times in this way, and the
alternative temporal perspectives on reality they afford, KC’s inability to think of anything he
might regret might also, at the same time, be a sign of a more fundamental loss of an important
part of our everyday understanding of time.
4. Conclusion
At one point, Craver and his colleagues write: ‘If KC is trapped in the present [because of his
total loss of episodic memory], he is trapped there with an awareness of his past, present and
future’ (Craver et al. 2014: 193). One way of characterizing the issue at stake in what I have
called the Dependency Thesis is by asking just what such an ‘awareness of the past and future’
can amount to in the absence of episodic memory – in KC, and even more so in someone who
never possessed any episodic memories in the first place (if it is the Developmental Dependency
Thesis that is in question).
There is of course a trivial sense in which a person without episodic memories necessarily
lacks one type of awareness of the past that those with episodic memories can enjoy, insofar as
the latter, but not the former, can consciously recollect particular past events they have wit-
nessed. On the other hand, cases such as that of KC (and other amnesiac patients that have been
studied) also seem to make it undeniable that a lack of episodic memory need to be no bar to
acquiring and retaining knowledge of truths couched in the past tense, and of truths about the
nature of time in general. Is there any more to be said, beyond these two relatively uncontrover-
sial observations, that would still leave room for substantive debate over the truth or falsity of the
Dependency Thesis? One thing I have sought to bring out in this chapter is how realism about
the past, as discussed in the philosophical literature, might be seen to provide for a way of giving
substance to the possibility of a profound sense in which a person may lack a genuine under-
standing of what it is for an event to have happened, even though they can recognize certain
sentences in the past tense as being true. As I have argued, of particular relevance in this context
215
Christoph Hoerl
is the realist’s idea that such an understanding also involves, as John Campbell (1997: 157) puts
it, a grasp of the ‘single temporal reality onto which all one’s various temporal perspectives face’.
What is at stake in the Dependency Thesis, ultimately, is what furnishes us with such a grasp, if
it isn’t the retained acquaintance with events as viewed from such other temporal perspectives
that episodic memory provides for.
Notes
1 The current usage of the term in psychology is due to Tulving (1972, 1985). Philosophers who make
use of the term include Martin (2001), Soteriou (2008) and Michaelian (2016).
2 See also Wittgenstein (1958, vol. II, pt xii), Dummett (1993), Hoerl (1999) and Debus (2013) for similar
claims.
3 Locke (1690: II, 10, §2) portrays memory as being generative when he defines it as ‘a power [possessed
by the mind] in many cases to revive perceptions which it has once had, with this additional percep-
tion annexed to them, that it has had them before’. For more recent examples, see Dokic (2001) and
Fernández (2016). There are at least two senses in which memory could be thought to be epistemically
generative. It might make available to the subject content or information not already available through the
original experience or learning episode, or it might be able to produce positive epistemic qualities such
as justification (on the latter, see, e.g. Lackey 2005). In the current context, when I speak of the idea that
episodic memory is epistemically generative, I have in mind specifically the former.
4 For critiques of the empiricist theory of memory, see, e.g. Holland (1954), Urmson (1971) and Hoerl
(2014).
5 Note also in this context that the version of the empiricist theory of memory advocated in Russell
(1921) differs quite radically from that informing the avowals of the Dependency Thesis in Russell
(1912) that I quoted at the beginning of this chapter.
6 This issue is discussed in more detail in McCormack (2001).
7 If the argument in this section is along the right lines, it plausibly also carries over to other accounts of
episodic recollection according to which it involves, for instance, being in a state that carries the infor-
mation that that state itself originates in a prior perceptual experience, as Fernández (2016) argues.
8 Dummett, who provides a classic discussion of anti-realism about the past in Dummett (1968), else-
where also argues for a version of the Dependency Thesis (Dummett 1993: esp. pp. 420ff.). It is
difficult to work out what exactly his argument is meant to come to, though, because he seems to
run it together with the separate argument that our general reliance on memory is not something for
which we can demand a justification without thereby making it impossible for us to acquire any sort
of extended body of knowledge (on this topic, see also Hoerl 2013). This difficulty stems from the
fact that Dummett does not distinguish between episodic memory and semantic memory (to which
the latter argument also applies).
9 See here also Peacocke’s (2001) related characterization of the realist’s position as one on which past-
tensed statements or thoughts are ‘equi-categorical’ with their present-tensed counterparts.
10 On this way of reading the claim, see McCormack (1999), which is a critical discussion of Hoerl (1999),
in which the stronger claim is being put forward.
11 Compare here the ‘ability hypothesis’ put forward in discussions of our grasp of phenomenal concepts
(Lewis 1990; Nemirow 1990), according to which knowledge of ‘what it is like’ to an have experience
of a certain kind consists in the ability to imagine having an experience of that kind, which cannot be
acquired through testimony, but only through actually having an experience of the relevant kind one-
self. This also entails that, once acquired, such knowledge can be lost again if the relevant imaginative
abilities are lost.
Related topics
• Bertrand Russell
• Ludwig Wittgenstein
216
Further reading
Russell’s account, in The Problems of Philosophy (Oxford: Oxford University Press, 1912), of the relation
between memory and the concept of the past, is influenced by Chapters 15 and 16 of W. James, The
Principles of Psychology, Vol. 1 (London: Macmillan, 1890). The first of these seeks to locate ‘the original
of our experience of pastness, from whence we get the meaning of the term’ (ibid.: 605), in the percep-
tion of time; the second, however, argues that conceiving the past, in such a way that we can ‘mentally
project’ (ibid.: 643) events into it, requires memory. An alternative account on which the origins of the
concept of the past lie in perception can also be found in A. J. Ayer, The Problem of Knowledge (London:
Macmillan, 1956). Attempts to ground grasp of the concept of the past in perception are criticized in
C. Hoerl, ‘The Perception of Time and the Notion of a Point of View’, European Journal of Philosophy
6 (1998): 156–71.
References
Anscombe, G. E. M. (1981). The reality of the past. The collected philosophical papers of G. E. M. Anscombe,
Vol. 2: Metaphysics and the philosophy of mind (pp. 103–19). Oxford: Basil Blackwell.
Baier, A. (1976). Mixing memory and desire. American Philosophical Quarterly, 13(3): 213–20.
Campbell, J. (1997). The realism of memory. In R. G. Heck (ed.), Language, thought, and logic: Essays in
honour of Michael Dummett (pp. 157–82): Oxford: Oxford University Press.
Craver, C. F. (2012). A preliminary case for amnesic selves: Toward a clinical moral psychology. Social
Cognition, 30(4): 449–73.
Craver, C. F., Kwan, D., Steindam, C. and Rosenbaum, R. S. (2014). Individuals with episodic amnesia
are not stuck in time. Neuropsychologia, 57: 191–5.
Debus, D. (2013). Thinking about the past and experiencing the past. Mind & Language, 28(1): 20–54.
Dokic, J. (2001). Is memory purely preservative? In C. Hoerl and T. McCormack (eds.), Time and memory:
Issues in philosophy and psychology (pp. 213–32). Oxford: Oxford University Press.
Dummett, M. (1968). The reality of the past. Proceedings of the Aristotelian Society, 69: 239–58.
Dummett, M. (1993). Testimony and memory. In M. Dummett, The seas of language (pp. 411–28). Oxford:
Evans, G. (1982). The varieties of reference, J. McDowell (ed.). Oxford: Clarendon Press.
Fernández, J. (2016). Epistemic generation in memory. Philosophy and Phenomenological Research, 92(3):
620–44.
Hoerl, C. (1999). Memory, amnesia, and the past. Mind and Language, 14(2): 227–51.
Hoerl, C. (2013). Memory and knowledge. In H. Pashler (ed.), Encyclopedia of the mind (Vol. 2,
pp. 489–92). London: SAGE.
Hoerl, C. (2014). Remembering events and remembering looks. Review of Philosophy and Psychology, 5(3):
351–72.
Hoerl, C. and McCormack, T. (2016). Making decisions about the future: Regret and the cognitive func-
tion of episodic memory. In K. Michaelian, S. Klein and K. Szpunar (eds.), Seeing the future: Theoretical
perspectives on future-oriented mental time travel (pp. 241–66). Oxford: Oxford University Press.
Holland, R. F. (1954). The empiricist theory of memory. Mind, 63(252): 464–86.
Hume, D. (1739–40). A treatise of human nature, L. A. Selby-Bigge (ed.); 2nd edn. rev. P. H. Nidditch.
Oxford: Clarendon Press, 1975.
Lackey, J. (2005). Memory as a generative epistemic source. Philosophy and Phenomenological Research, 70(3):
636–58.
Lewis, D. (1990). What experience teaches. In W. G. Lycan (ed.), Mind and cognition (pp. 29–57). Oxford:
Blackwell.
Locke, J. (1690). An essay concerning human understanding, P. H. Nidditch (ed.). Oxford: Clarendon Press,
1975.
Martin, M. G. F. (2001). Out of the past: Episodic recall as retained acquaintance. In C. Hoerl and
T. McCormack (eds.), Time and memory: Issues in philosophy and psychology (pp. 257–84). Oxford:
McCormack, T. (1999). Temporal concepts and episodic memory: A response to Hoerl. Mind & Language,
14(2): 252–62.
McCormack, T. (2001). Attributing episodic memory to animals and children. In C. Hoerl and T. McCormack
(eds.), Time and memory: Issues in philosophy and psychology. Oxford: Oxford University Press.
217
Christoph Hoerl
McCormack, T. and Hoerl, C. (2008). Temporal decentering and the development of temporal concepts.
Language Learning, 58(Suppl. 1): 89–113.
McDowell, J. (1978). On ‘The reality of the past’. In C. Hookway and P. Pettit (eds.), Action and interpretation:
Studies in the philosophy of the social sciences (pp. 127–44). Cambridge: Cambridge University Press.
MA: MIT Press.
Nemirow, L. (1990). Physicalism and the cognitive role of acquaintance. In W. G. Lycan (ed.), Mind and
cognition. Oxford: Blackwell.
Peacocke, C. (2001). Understanding the past tense. In C. Hoerl and T. McCormack (eds.), Time and
memory: Issues in philosophy and psychology. Oxford: Clarendon Press.
Pears, D. F. (1975). Russell’s theories of memory, in D. F. Pears, Questions in the philosophy of mind (pp.
224–50). London: Duckworth.
Rosenbaum, R. S., Köhler, S., Schacter, D. L., Moscovitch, M., Westmacott, R., Black, S. E., Gao, F.,
and Tulving, E. (2005). The case of KC: Contributions of a memory-impaired person to memory
theory. Neuropsychologia, 43(7): 989–1021.
Russell, B. (1912). The problems of philosophy. Oxford: Oxford University Press.
Russell, B. (1921). The analysis of mind. London: George Allen & Unwin.
Soteriou, M. (2008). The epistemological role of episodic recollection. Philosophy and Phenomenological
Research, 77(2): 472–92.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving and W. Donaldson (eds.), Organization
of memory. New York: Academic Press.
Tulving, E. (1985). Elements of episodic memory. Oxford: Oxford University Press.
Urmson, J. O. (1971). Memory and imagination. Mind, 80(1): 70–92.
Wittgenstein, L. (1958). Philosophical investigations, G. E. M. Anscombe (trans.); G. E. M. Anscombe and
R. Rhees (eds.). Oxford: Basil Blackwell.
218
17
MEMORY AND THE
METAPHYSICS OF TIME
Robin Le Poidevin
1. Introduction
To understand the nature of memory is to understand the nature of time itself. That rather bold
thesis is the subject of the following discussion. The idea that we can simply read off the mind-
independent nature of reality from our mental representations of it does seem rather doubtful, of
course. But a rather more promising line of inquiry is to see whether some positions in debates
about time’s true nature sit rather better than do others with some widespread beliefs about
remembering. So that is how this discussion will proceed, by first introducing a metaphysical
debate about time, and then exploring the alleged connection with memory. The aspects of
memory that we will be particularly concerned with are its role in structuring our perceptual
experience, and its role in providing knowledge of our experiential pasts. And the metaphysical
views of the nature of time those aspects (arguably) connect with concern the passage of time,
the reality of the past, and the basis of time’s direction.
2. Why do we experience time as passing? The role of

memory in temporal awareness
The most obvious aspect of time that experience affords is its passage: what we are now
perceiving tends to be fleeting, and soon consigned to the past. But does time really pass?
According to the A-theory of time, it does. By ‘really’ here is meant that it is a feature of time
itself, independently of us or our beliefs, that it passes. What is it for time to pass? Intuitively,
we think of it as present events ceasing to be present, and receding into the past. Or we might
think of events as ceasing to exist altogether when they cease to be present. Quite how the
idea is expressed varies from one version of the A-theory to another, but what they have in
common is the thought that the distinction between past, present and future is an objective,
mind-independent one. Remove all sentient life from the world, and there would still be a
fact of the matter as to what was happening now. And that state of affairs does not remain fixed.
It is the nature of presentness to be transient: whatever is now present will cease to be so. At
that point, it may cease to be part of reality altogether (as ‘presentists’ hold). Or it may simply
become a different part of reality, one that stands in a relation of precedence to the present (as
‘growing block’ theorists hold). The question of the reality of the past is one we shall examine
219
Robin Le Poidevin
more closely in the next section. The key beliefs unifying A-theorists are that there is a privi-
leged time which is present, that whether or not something is present is mind-independent,
and that what is present does not remain so. This is what it is for time to pass in reality.
Intuitive though those beliefs are, they are taken to be false by the B-theory of time. For the
B-theorist, once you remove all sentient life from the world, there is no longer any question as
to what is going on now: the term simply has no application. Terms like ‘now’, ‘then’, ‘last year’,
‘tomorrow’, ‘past’, ‘present’ and ‘future’ – tensed terms, in short – simply reflect our perspec-
tive on events. We correctly judge it to be raining now, or in the present, when a (typically,
local) episode of rain is simultaneous with our judgement: mind-independent presentness does
not come into it. The B-theory conceives of time as consisting of a series of times (or events,
or states of affairs) related by precedence or simultaneity. That event E1 precedes E2 in the series
is a mind-independent fact.1 That E2 is present and E1 past is a perspectival matter, true from
the perspective of someone who has a mental representation of the two events. In short, ‘now’
for the B-theorist is the temporal analogue of ‘here’. However, the relations of earlier than and
simultaneous with are not mind-dependent:2 these relations, which order events in time, are
entirely objective. As the B-theorist might put it: there is order, but no passage.
Why is it the A-theory that has the most immediate appeal, indeed that almost seems to be
stating the obvious, whereas the B-theory appears to be revisionary, doing violence to our pre-
theoretical conceptions? Both sides typically admit that it is ordinary experience that seems to
press on us the notion of time passing. Indeed, so pervasive is this sense of passage supposed to
be that it constitutes what we might call the argument from experience for time’s real passage. There
is, however, no one aspect of experience that is thought responsible for (apparently) intimating
passage. Here are three ways in which experience could be thought to do so:
• We are aware that things used to be different from the way they are now.
• We directly perceive change and succession.
• Our awareness of things as being present or as past does not appear to us as merely perspectival,
but as objective, or absolute.
If it is accepted that these features are what lead us to believe in passage, then the onus is on the
B-theorist to suggest a mechanism underlying these aspects of experience, a mechanism which
does not involve actual passage (as opposed to its representation). Alternatively, the B-theorist
may deny that experience does have the feature in question. The first two of these, at least, will
hardly be challenged by the B-theorist, who will nevertheless insist that they fall some way short
of constituting direct awareness of a real passage of time. The third is, perhaps, a different matter.
The first surely just involves a comparison between current awareness and memory, the discrep-
ancy between which does not have to be explained in tensed terms. Instead, that is, of reporting
the facts in this way: it is the case that p and it was the case that Not-p, we can say simply that p
obtains at one time but does not obtain at an earlier time. This is inferential knowledge. Perception
of change and succession, in contrast, is arguably more direct, and non-inferential: we just see one
horse pass the finishing line after the other; we don’t need to infer it from a perception/memory
contrast. Nevertheless, memory may be involved in a less obvious way. The perception of the
first horse crossing the line is not simultaneous with the perception of the second horse doing so.
But the perception of the first is still retained, and colours the perception of the second, in such
a way as to give rise to the experience of one event as following the other. The retention of per-
ceptual information is a form of memory, even if it does not present itself as a memory. So here
the perception of succession could be explained simply by a causal connection between memory
and perception, a process that need not involve passage.3 This does not necessarily rule out other
220
Memory and the metaphysics of time
mechanisms of change perception. The visual perception of motion, for example, may be a direct
result of changes in retinal stimulation, without the need for some information-retaining process
(Gregory 1958). But again, there is no necessity to appeal to real passage here.
The third alleged feature of experience is a little more controversial. It is certainly true that
perception and memory both give rise to tensed beliefs: present- and past-tensed beliefs respec-
tively. It is also true that these tensed beliefs are essential to action: only if I believe that I am
now in a minefield will I take appropriate action; the tenseless belief that I am in a minefield at
12 noon on 20 January 2016 will not by itself have the same effect (Perry 1979). But does my
awareness that it is, or was, the case that p present itself to me as non-perspectival? Experience
by itself might be neutral on that point, and only reflection on experience gives rise to a further
belief that time’s passage is objective. Moreover, there is a case for saying that, by confronting
us with different temporal perspectives on events, memory actually intimates the perspectival
character of temporal experience (Percival 1994). This is particularly true of what is known as
episodic memory: memory of our own past experiences. For that memory both presents me with
the content of a past experience, an experience of something as happening now, while simulta-
neously providing the knowledge that the experience in question is past. To resolve what would
otherwise be a contradiction (the event both is present and is wholly past), we need to view the
two tenses as different perspectives.
None of this shows directly that time does not in fact pass. What it does show, however, is
that appeal to memory can provide the B-theorist with the resources with which to undermine
the argument from experience. Experience may give rise to the belief in time’s passage, but if
we can explain the character of that experience without appeal to such a notion, experience no
longer tells decisively in favour of the A-theory.4
3. Do we remember a real past? Realism and

the epistemology of memory
One motivation for the A-theory, less directly connected with experience than the features just
mentioned, comes from the thought that there is an ontological difference between past and
present, that whereas the present enjoys full concrete reality, the past is no longer part of that
reality – and the future is not yet part of reality. This is presentism. Like the A-theory in general,
it arguably corresponds to our ordinary conception of time. But if we are drawn to some version
of the correspondence theory of truth, or more broadly the view that (as it is sometimes put)
‘truth supervenes on being’ (Bigelow 1988), then there is immediately a question over what
the truth-makers are for our memory beliefs. I believe, on the basis of memory, that the river
burst its banks last week, flooding the village. What, if not past reality, makes that true, or false?
The only (concrete) reality available to supply those truth-makers is present reality. So a natural
solution to the difficulty is to take the present causal traces of the past to be the truth-makers of our
beliefs about the past.5
An early (mid-twentieth-century) version of this approach had an epistemological focus: the
relevant truth-makers are what we would count as evidence for past-tensed truths. So my memory
of last week’s flood is made true by the observable remains of that flood: the broken-down wall,
the sodden sandbags outside houses, the quantity of debris in the road, etc. The motivation for
this view is essentially a verificationist one: that the meaningfulness of our statements (and the
coherence of the corresponding beliefs) depends on our knowing how to go about verifying, or
at least confirming, their truth. Those epistemological concerns apply quite generally, of course,
and not just to the past. But the resulting ‘anti-realism’ about the past (the name Dummett gives
to this brand of verificationism – see, e.g. Dummett 1969) has something in common with its
221
Robin Le Poidevin
ontological counterpart, namely that the notion of ‘past fact’ is problematic. If we are not driven
by verificationist concerns, then we need not worry about the accessibility of evidence for past-
tensed truths: the present causal traces of the past need not be restricted to those that are in
principle available to us. But to guarantee that all statements/beliefs about the past have a determi-
nate truth-value by virtue of what is now the case, we would need to commit to determinism. That
is, the belief that it was the case at some specified date that p is true if and only it is necessitated
by the current total state of the universe together with the laws of nature. Without this, present
causal traces (whether accessible or hidden) will be insufficient to deliver determinate truth-values
for past-tensed statements: they could only be rendered more or less probably true or false.
This view that memories have present truth-makers does not sit well, however, with how
we understand the epistemology of memory. Memory provides us with knowledge of the past
(though see the final section of this chapter for a crucial qualification of this remark). But there
is something indirect about it. It is only a route to knowledge because the original experience on
which it is based was a route to knowledge. This feature John Campbell (1994) calls the ‘step-
wise’ character of memory, and as he points out, it is a feature memory shares with testimony.6
Someone’s testimony only provides us with knowledge because the testifier had an independent
route to that knowledge. The belief we form just on the basis of testimony is only as secure as
the belief on which the testimony was based. Suppose someone witnesses what they take to be
a meteor shower, and tells their credulous friend the following morning that there had been a
meteor shower last night. If what was witnessed was actually a distant firework display, then the
friend’s belief, formed on the basis of ill-grounded testimony, cannot count as knowledge, even
if as a matter of fact there was a meteor shower last night. Now, as we might put it, memories
are originally formed on the basis of the testimony of the senses. If that testimony, that is, the
original experience, was unreliable, then the memory inherits that unreliability. There is a close
connection, then, between what grounds the memory belief, and what grounds the veridicality
of the original experience. Indeed, for some cases, we might even think of the relevant relation
between the two as one of identity, which is as close a connection as one could have. Suppose
the original experience was one of an explosion, giving rise to the belief that an explosion
occurred. Sometime later, the observer remembers that experience, again forming the belief
that an explosion occurred. Is it not plausible to say that the experience belief and the memory
belief have the same grounds, namely the explosion itself?
We might capture the relationship between a memory and the original experience whose
content it preserves in the following ways: both the experience and the memory are tracking the
same facts, the same bit of reality, through changing perspectives on that reality. It is as if we are
keeping a moving object in view: the perspective on that object changes, but the object does
not. We might also put the point in terms of truth-makers: what made the original experience
veridical also makes true the later memory (Le Poidevin 2007: Chapter 4).
Now put these remarks on the epistemology of memory in the context of anti-realism
about the past. The truth of the original experience belief is grounded in the evidence available
at the time of that belief. But the memory belief is grounded, not in that evidence, but rather
in the evidence available at the time of the memory, and this may be very different. Compare see-
ing the raging torrent of water causing havoc in the village, with inspecting the remains of the
devastation in the now-dry village a week later. The two sets of grounds, one for the original
experience and the other for the later memory, are quite separate, and indeed independent of
each other. It now looks as if the experience and the memory are answerable to two independent
sets of data. So what has become of the stepwise character of memory?
That, essentially, is Campbell’s challenge to anti-realism about the past, for which the notion
of evidence is key.7 What if we move to non-verificationist presentism, and the ontological
222
thesis that the past is unreal? On this view,8 the truth-makers of past-tensed beliefs are, insofar
as they are concretely real, must be present facts, facts which do not necessarily have to be
accessible as evidence. Again, it looks as if there is going to be a disparity between the two
truth-makers. The truth-maker for the original experiential belief was a present fact (at the time
of the experience). But by the time of the memory, that fact – for the presentist – has gone
out of existence, and what provides the truth-maker for the memory belief is a fact that is now
present. So, again, it seems that the two beliefs are answerable to different facts. We can no
longer preserve the intuitive thought that memory and experience are tracking the same reality
through changing perspectives. Instead, we have to say that they are tracking a changing reality.
But it might be said that, although the truth-makers for the experience and the truth-maker
for the memory are indeed different facts, they are not independent facts, and this is a significant
difference between presentism and anti-realism about the past. Recall that the presentist approach
to the truth-makers of past-tensed beliefs can only guarantee a determinate truth-value for those
beliefs by a commitment to determinism. So the two facts, one past, the other present, are bound
together by a necessary connection. Past facts are no longer real, but that they once obtained is
necessitated by present fact (given the laws of nature). However, the idea that the presentist really
is binding together facts from different times is an illusory one. Rather, we have a connection
between two propositions which are true now, one a present-tensed statement concerning the
present state of the universe, and one a past-tensed proposition concerning how things were.
Why is this not enough? Because of a further aspect of memory. We said that memory
provides us with knowledge of the past. How does it do that? Not simply by providing true past-
tensed beliefs. Those might turn out to be accidentally true. It also matters how the memory was
acquired: its causal history. What we find in cases of genuine knowledge (or at least, of knowl-
edge of concrete things, as opposed to mathematical or logical knowledge) is a close connection
between the truth-makers of the knowledge belief and the causal story behind its acquisition. But,
according to the presentist (at least in the version we have been considering), the truth-makers of
the memory are in the present. But then they can be playing no part in the causal history of the
memory. That story involves facts now past, and so no longer available as truth-makers.
Presentism is often put forward, with some justification, as the natural view of time, the one
that we are drawn to before encountering the philosophical debates surrounding it. But once
we put it in the context of temporal experience, and in particular what we understand about
the structure of memory, it starts to look quite revisionary. It disrupts the connection between
memory and the original experience that give rise to the memory. And the very notion of mem-
ory as having a causal history, one we want to appeal to in explaining how memory provides
us with knowledge of the past, is thrown into confusion by presentism. For causation is, surely,
a transtemporal relation: it spans times. Yet, on the presentist picture, there is only one time: the
present. So there aren’t enough relata for the causal relation to relate!9
4. What can memory tell us about the arrow of time? Remembering

and the causal theory of time order
Along with the passage of time, and the (un)reality of the past, the fact that time appears to have
a direction is one of the most-debated topics in the metaphysics of time. Even if we do not iden-
tify the direction of time with its passage, but take a strictly B-theoretic view, it would surely be
a step too far to treat time as exactly akin to a spatial dimension. There is no intrinsic direction
to space, only directions in space which we conventionally mark out. In contrast, the difference
between earlier and later is not, surely, a purely conventional one. But what is it that grounds
the direction, or ‘arrow’ of time?
223
Robin Le Poidevin
A popular, but decidedly unsatisfactory account is in terms of entropy. According to the

Second Law of Thermodynamics, energy becomes increasingly degraded over time. That is, it
tends to become less organised, and more dissipated in the form of heat. ‘Entropy’ is the name
given to the degree to which energy has been degraded in this way. The more degraded and
randomly distributed, the greater the entropy. Even though we may organise local packets
of energy in order to do useful work (by winding up a clockwork mechanism, for example),
thus bringing about a local decrease in entropy, we do so only at the cost of a global increase.
Now we could say that the arrow of time consists precisely in the fact that entropy increases.
But here we face a dilemma. If we say that entropy increases over time, we presuppose time’s
direction, since ‘increases’ is a temporal term: the increase of entropy aligns itself in the earlier
to later direction, not vice versa. So the increase in entropy does not explain time’s direction,
but is just one more puzzling asymmetry to explain. We could get around that problem by
defining the earlier-to-later direction in terms of entropy: a time is later than another by virtue
of the fact that entropy is greater at that time. But now we risk saying something that is just
false, for we now rule out by fiat the possibility of global entropy decreasing, however briefly.
Insofar as the Second Law of Thermodynamics is to do with the statistical probability of energy
distributions, however, such a global decrease in entropy, though perhaps vanishingly unlikely,
is not actually impossible.
If we are to attempt to reduce the direction of time to some other direction, then a more
promising candidate is the direction of causality, that is, the direction from causes to effects. Let
us for the time being work with the simplest version of this theory: x is earlier than y if and only
if x is a cause of y. Causally unrelated items are therefore also temporally unrelated. That will
no doubt seem far too strong a commitment, but for now let us see how this account of time’s
arrow fits with some fairly basic aspects of memory.
As we have already said, memory provides us (or can provide us) with knowledge of the
past. Moreover, in the case of episodic memory, it provides us with singular knowledge of the
past: that is, it enables us to have a singular thought about a particular past experience we had.
You may remember, for example, not simply that you once visited St Paul’s Cathedral, but also
remember actually doing so, looking down Ludgate Hill at the enormous dome, perhaps also
having an image in your mind at the time an image of the cathedral during the Blitz surrounded
by smoke and flames. Finally, memory presents us with the past as the past: it represents its con-
tent as something that happened earlier than the present time. We have already noted the role
memory plays in our awareness of succession, of one event succeeding another. And though we
may be mistaken in the actual order of external events, we cannot similarly be mistaken when
memory presents an experience of ours as earlier than the present memory of it. All this is fairly
uncontroversial. But it still needs explaining.
What explains, then, the content of memory? What makes this memory a memory of that
event? The most influential answer to this question has been a causal one: the original event is
the original cause of the memory (Martin and Deutscher 1966).10 And the causal connection is
mediated by the original experience, which thus transmits its content to the memory. That the
memory constitutes knowledge of the event is guaranteed (as we noted above) by the coinci-
dence of cause and truth-maker, insofar as the event plays both roles. But what now of the fact
that memory also provides us with knowledge that the experience is earlier than that memory?
This is where the causal account of time order comes in. We have already appealed to a causal
connection between experience (E) and memory (M) to explain their shared content. But now
that very causal connection also guarantees the truth of the belief that E is earlier than M, and so
once again we have the truth-maker for the belief that M succeeds E as part of the causal story
behind the acquisition of that belief. So the belief that E is earlier than M is not only true, but
224
also constitutes knowledge. Causal connections between event, experience and memory thus
explain the salient features of episodic memory.
Two objections might be raised against this account. First, it might be suggested that we
do not need anything as strong as a reductive theory of the ‘earlier than’ relation to guarantee
the truth of the belief that experience precedes the memory of the experience. All we need is
the fact that causes invariably precede their effects. But unless we opt for a reductive account,
either from time to causation, or from causation to time, what guarantee is there that causes
do invariably precede their effects? It looks just like a brute fact, and while we are compelled
to accept some brute facts, the coincidence of time order with causal order does not look like
one. But now (to develop the objection a little further), it might be asked why, if we have to
opt for a reductive account, we should aim to reduce time to causation, rather than the other
way around? It might, that is, be constitutive of the causal relation, that causes are the earlier
relata.11 But this is not the only asymmetry in causal relations. If it were, we would not suppose
the asymmetry of causation to be particularly significant, but would take it simply as marking
something like a grammatical distinction between ‘cause’ and ‘effect’. However, causes are sup-
posed to explain their effects in a way in which effects do not explain their cases (Mellor 1995: 60).
Causes make their effects much more probable than effects make their causes. And, relatedly,
effects are supposed to be counterfactually dependent on their causes, not the other way around.
How could mere precedence explain those other asymmetries?
The other kind of objection to the account of memory offered above is that it is vulnerable
(of course) to the various objections to the causal account of time order. To identify time order
with causal order, as in the rather basic account offered above, is, it will be said, too strong, for
(i) we can conceive of causally isolated systems which share a common time series; (ii) causation
is a two-place relation, whereas temporal precedence is a three-place relation, the third relatum
being the inertial frame, for (as the Special Theory of Relativity allegedly teaches us), whether
one event is simultaneous with, or earlier than, another, is relative to inertial frame. Causality is
not so relative, ergo they are different relations. QED. We can, however, accommodate both
objections by a modification of the strong causal account: x is earlier than y if only if x is simul-
taneous with a cause of some event which is simultaneous with y. So x and y can stand in the
relation of precedence even in the absence of a causal connection. The frame-relativity of sim-
ultaneity will then imply that some relations of precedence will vary from frame to frame. Some,
but not all. For when x and y are causally related, then the fact that x precedes y is something
that does not vary from frame to frame. We can thus define time order by means of causal order,
without identifying time relations with causal relations. Perhaps something of the simplicity of
the strong reduction is lost, but what is preserved is the insight that a world without causality is
a world devoid of temporal direction.
5. A questionable model for memory?

The foregoing discussion has been an attempt to connect issues in the metaphysics of time
with what was described as ‘widespread beliefs about remembering’. The beliefs in question
were that our memories provide us with knowledge of our past, by preserving a reliable con-
nection with that past, a connection constituted by causal links, both between the memory
and the original experience that led to the memory, and between the experience and what
the experience was of.
If we take that characterisation to be a statement of necessary conditions for something to
count as remembering, it corresponds to the ‘epistemic theory of memory’: to remember at
t2 that p obtained at t1 is to know that p obtained at t1 because one knew at t1 that p obtained
225
Robin Le Poidevin
then. But, influential though that theory has been, it is not, as Sven Bernecker (2007, 2010) has
shown, beyond criticism. Bernecker argues that there are counter-examples both to the present
knowledge condition (that remembering that p entails that one now knows that p) and to the
past knowledge condition (that remembering that p entails that one knew then that p). As a
counter-example to the past knowledge condition: one might now remember seeing the Loch
Ness Monster, even though at the time of the perception, one had good reasons (now discred-
ited) for thinking that one was hallucinating. The past knowledge condition is not satisfied here
because knowledge entails justification, and one was not justified in believing at the time that
one was seeing the Loch Ness Monster. And as a counterexample to the present knowledge
condition: one might remember that one’s sister wore a blue blouse even though one now has
evidence that the blouse was in fact green, undermining both justification and belief that one
did in fact see a blue blouse.12
If, then, the epistemic theory of memory is vulnerable to objections such as these, does this
not undermine the previous attempt to use features of memory to motivate certain views of
time? Were we, after all, operating with a questionable model? The arguments canvassed in the
previous sections, however, need not rest on anything as strong as the epistemic theory, which
is intended as a conceptual analysis of memory. It can be conceded that not all cases of remem-
bering constitute knowledge of the past, and that remembering does not entail having had past
knowledge. It is, however, important that paradigm cases of remembering exhibit the features
identified above. Memory is at least capable of providing a reliable link to our experiential
past, and where it does so, the reliability of that part of our present mental state that constitutes
the memory (in distinction to any other beliefs we might have acquired by the time of the
memory) inherits its reliability from that of the original experience. Nevertheless, by chipping
away at the epistemic theory, the various objections to it might prompt a revisionary account of
memory, leaving the door open for the possibility of jettisoning altogether the idea of memory
as providing a means to preserve reliable connections with the past. And that radical step might
be motivated further by a commitment to the metaphysical views of time targeted above, and
perhaps in particular to those brands of the A-theory that posit the unreality of the past. Even
the causal theory of memory has not been free from criticism (see, e.g. Squires 1969). But the
point of this discussion was not to offer knock-down arguments for theories of time, but rather to
display the interesting ways in which they link with views on memory. Quite which dialectical
direction one should move in when exploring them I leave to the reader.
Acknowledgement
Many thanks to Sven Bernecker for very helpful comments on an earlier version of this chapter.
Notes
1 By ‘fact’, here and throughout this chapter, I mean a part of reality, not some true statement. Facts are
the items that make true statements true.
2 It might be thought that the Special Theory of Relativity implies that simultaneity is not objective. But,
on the standard interpretation of the Special Theory, simultaneity is held to be relative to an inertial frame,
and frame-relativity is not the same as mind-dependence, or subjectivity.That two events are simultaneous
with respect to a frame is itself entirely objective and mind-independent.
3 See, e.g. Mellor (1998: Chapters 10 and 11).
4 See Le Poidevin (2007: Chapter 5) for further elaboration of this.
5 This is one version of presentism, and arguably the one which meets the obligation to provide an expla-
nation for the determinate truth-value of past-tensed statements head on. Not all presentists, however,
concede the need for presently existing truth-makers. For some varieties of presentism, see Bourne
226
(2006), and for a defence of a form of presentism which sidesteps a potentially problematic truth-maker
ontology, see Tallant and Ingram (2015).
6 See Campbell (1994: 233). This analogy between memory and testimony is also drawn by Audi (1994:
410), and Fricker (2006: 603–4). I am grateful to Sven Bernecker for drawing my attention to these
references.
7 See especially, Campbell (1994: 239f.).
8 Or at least, the version of it that we are now considering. See Note 5.
9 For presentist responses to this challenge, see Bourne (2006: Chapter 4).
10 For a detailed discussion of this classic paper, criticisms of the causal theory of memory, and how the
causal theorist might respond to those criticisms, see Dorothea Debus’s contribution to this volume
(Debus 2017).
11 For Hume, temporal priority is precisely what introduces the asymmetry in what would otherwise be
a symmetric relation, the other key aspects of causality, on his account, being contiguity and constant
conjunction. See Hume (1739–49/1978: 173).
12 See Bernecker (2007: 144–5; 2010: Chapter 3).
Further reading
Connections between features of memory and debates over the nature of time are discussed in Campbell
(1994) (specifically in relation to the reality of the past), Mellor (1998) (in relation to the direction of time)
and Le Poidevin (2007) (in relation to the passage of time).
References
Audi, Robert (1997) ‘The place of testimony in the fabric of knowledge and justification’, American
Philosophical Quarterly, Vol. 34, pp. 405–22.
Bernecker, Sven (2007) ‘Remembering without knowing’, Australasian Journal of Philosophy, Vol. 85: 137–56.
Bernecker, Sven (2010) Memory: A Philosophical Study, Oxford: Oxford University Press.
Bigelow, John (1988) The Reality of Numbers, Oxford: Oxford University Press.
Bourne, Craig (2006) A Future for Presentism, Oxford: Oxford University Press.
Campbell, John (1994) Past, Space and Self, Cambridge, MA: MIT Press.
Debus, Dorothea (2017) ‘Memory causation’, in Sven Bernecker and Kourken Michaelian, eds., The
Routledge Handbook of the Philosophy of Memory, London: Routledge.
Dummett, Michael (1969) ‘The reality of the past’, reprinted in Truth and Other Enigmas, London:
Duckworth, 1978, pp. 358–74.
Fricker, Elizabeth (2006) ‘Second-hand knowledge’, Philosophy and Phenomenological Research, Vol. 73:
592–618.
Gregory, R. L. (1958) ‘Eye movements and the stability of the visual world’, Nature, Vol. 182: 1214–16.
Hume, David (1739–40/1978) A Treatise of Human Nature, ed. L. A. Selby-Bigge; 2nd edn. rev. P. H. Nidditch,
Oxford: Clarendon Press.
Le Poidevin, Robin (2007) The Images of Time: An Essay on Temporal Representation, Oxford: Oxford
University Press.
Martin, C. B. and Deutscher, M. (1966) ‘Remembering’, Philosophical Review, Vol. 75: 161–96.
Mellor, D. H. (1995) The Facts of Causation, London: Routledge.
Mellor, D. H. (1998) Real Time II, London: Routledge.
Percival, Philip (1994) ‘Absolute truth’, Proceedings of the Aristotelian Society, Vol. 94: 189–214.
Perry, John (1979) ‘The problem of the essential indexical’, Noûs, Vol. 13: 3–21.
Squires, Roger (1969) ‘Memory unchained’, Philosophical Review, Vol. 78: 178–9.
Tallant, Jonathan and Ingram, David (2015) ‘Nefarious presentism’, Philosophical Quarterly, Vol. 65: 355–71.
227
18
MEMORY AS MENTAL TIME
TRAVEL
1. Introduction: continuism and discontinuism

about mental time travel
When Tulving (1972) first introduced the term, he defined episodic memory essentially as a
specialized store devoted to information about the ‘what’, ‘when’, and ‘where’ of experienced
past events. Episodic memory thus contrasted both with nondeclarative memory, devoted in
part to skills and habits, and, within the category of declarative memory, with semantic memory,
devoted to general facts. This definition was broadly compatible with traditional analyses of what
philosophers had referred to as recollective, experiential, or personal memory (Brewer 1996),
including the popular causal theory (Martin and Deutscher 1966; Bernecker 2010). But semantic
memory, too, is capable of storing information about the what, when, and where of events, and
accumulating evidence of a tight relationship between the ability to remember the past and the
ability to imagine the future subsequently led most psychologists (including Tulving 2002) to
redefine episodic memory as a form of mental time travel (MTT) in which the subject imaginatively
re-experiences past events, just as, in future-oriented mental time travel (FMTT) he imagina-
tively ‘pre-experiences’ future events (Michaelian et al. 2016; De Brigard, Chapter 10, this
volume). It is unclear whether this new definition of episodic memory as mental time travel is
compatible with traditional philosophical analyses, which assume that there is a deep difference
between remembering the past and imagining the future. Empirical research within the MTT
framework has revealed a wealth of commonalities between episodic memory and FMTT,
leaving no doubt that there is some sort of tight relationship between them. But is it really the
case, as the framework suggests, that the only important difference between episodic memory
and FMTT is constituted by their distinct temporal orientations?
Those to whom we will refer as continuists answer this question in the affirmative, main-
taining that, aside from their distinct temporal orientations, there is no fundamental difference
between episodic memory and FMTT – that there is a single general faculty of mental time
travel (Suddendorf and Corballis 2007). Those to whom we will refer as discontinuists answer
in the negative, arguing that ‘[episodic memories] of past events and [episodic imaginations]
of future events are ultimately mental occurrences of two different kinds’ (Debus 2014).1 The
debate between continuists and discontinuists is ongoing, and this chapter will not attempt to
settle it. The aim of the chapter is, more modestly, to survey the evidence and arguments to
228
Memory as mental time travel
which continuists and discontinuists have appealed. Section 2 describes plausible versions of
continuism and discontinuism. Sections 3 and 4 provide a brief overview of the psychological
evidence for each view. Section 5 provides a more detailed review of philosophical arguments
for discontinuism, and Section 6 sets out continuist critiques of those arguments. Finally,
Section 7 discusses one promising discontinuist response to continuist critiques.
2. Varieties of continuism and discontinuism

A degree of conceptual ambiguity is an inevitable feature of any new theoretical framework,
and it has not always been clear exactly what is at stake in debates over the continuity or
discontinuity of episodic memory and FMTT (Klein 2016). It is crucial, in particular, for
parties to the debate to be explicit about whether they hold episodic memory and FMTT
to be different/the same in kind or merely in degree. An extreme continuist, for example,
might maintain that their distinct temporal orientations constitute literally the only difference
between episodic memory and FMTT, but there is also room for a view which acknowledges
the existence of differences of degree while insisting that episodic memory and FMTT are
the same in kind. Thus (anticipating some of the evidence reviewed in Section 4 below) a
moderate continuist might grant, for example, that imagining future events is somewhat more
cognitively effortful than remembering past events and that it involves a correspondingly
stronger activation of certain brain regions, but argue that what is at issue in both remembering
the past and imagining the future is nevertheless a single process, carried out by a single system.
In light of the evidence for differences of degree between episodic memory and FMTT, most
continuists endorse something like this moderate view.
Extreme and moderate forms of discontinuism are likewise available. An extreme discon-
tinuist would maintain that episodic memory and FMTT have nothing at all in common with
each other, but there is also room for a view which acknowledges the existence of similarities of
degree while insisting that episodic memory and FMTT are different in kind. Thus (anticipating
the evidence reviewed in Section 3), a moderate discontinuist might grant that episodic memory
and FMTT involve constructive episodic simulation of events and that constructive episodic
simulation, in turn, depends on the brain’s core network but argue that the necessity of a causal
link with past experience underwrites a difference in kind between the process of remembering
the past and that of imagining the future.
In light of the evidence for quantitative similarities between episodic memory and FMTT,
most discontinuists endorse something like this moderate view. But the fact that both con-
tinuists and discontinuists agree that there are some similarities and some differences between
episodic memory and FMTT should not lead us to underestimate the extent of the disagree-
ment between them: for the latter, there is a difference in kind between what we do when we
remember the past and what we do when we imagine the future; for the former, there is only
a difference of degree.
3. Empirical evidence for continuism

A detailed review of the empirical evidence for each of these views would be out of place
here, but this section will review some of the highlights of the evidence for continuism, and
the following section will do the same thing for discontinuism.2
Perhaps the most impressive evidence for continuism comes from imaging studies, which
have demonstrated that strongly overlapping regions of the brain are involved in both episodic
229
memory and FMTT, supporting the claim that a core (or default) network, whose major com-
ponents include medial prefrontal regions, posterior regions in the medial and lateral parietal
cortex, the lateral temporal cortex and the medial temporal lobe, constitutes the neural basis
for both forms of MTT (Addis et al. 2007; Schacter et al. 2007). This claim, in turn, aligns
with the constructive episodic simulation hypothesis (Schacter and Addis 2007a, b), which, in
a refinement of Bartlett’s insights on the constructive character of remembering (Bartlett 1932;
Wagoner 2016), sees both remembering the past and imagining the future as products of a con-
structive, simulational process in which traces of disparate past experiences are linked together
into coherent representations of particular events, the difference between them being simply
that, in FMTT, the process simulates possible future events, whereas, in episodic memory, it
simulates past events. While there is debate about how, exactly, to characterize the process
in question,3 the constructive episodic simulation hypothesis comes close to capturing the
dominant view among psychologists and neuroscientists on the relationship between episodic
memory and FMTT.
Imaging evidence is reinforced by evidence from a variety of other sources. Studies of MTT
in memory-impaired patients have found that deficits in the ability to remember one’s past are
strongly correlated with deficits in the ability to imagine one’s future (e.g. Klein et al. 2002;
Rosenbaum et al. 2005; Hassabis et al. 2007). Similarly, patients suffering from depression dis-
play parallel tendencies to remember the past and to imagine the future in overly general ways
(Williams et al. 1996), and the capacities to remember past episodes and to imagine future
episodes emerge in development at roughly the same age (Suddendorf and Busby 2005; Atance
2008; Perner et al. 2010; Suddendorf 2010; Viard et al. 2012). Further support for continuism
comes from studies of phenomenological similarities between episodic memory and FMTT.
Level of detail and intensity of experience vary with temporal distance in a similar manner in
both forms of MTT (D’Argembeau and Van der Linden 2004, 2006; D’Argembeau et al. 2011;
Addis et al. 2011; Schacter et al. 2012), in line with Tulving’s (1985) claim that the same phe-
nomenology continuism comes from studies of phenomenological similarities between episodic
memory and FMTT. Level of detail and intensity of experience vary with temporal distance.
And studies of autobiographical memory have found that episodic memory and FMTT are
organized in a similar fashion, in the sense that autobiographical memories and autobiographical
future events are embedded in the same narrative structures (Rathbone et al. 2011).
4. Empirical evidence for discontinuism

Psychology and neuroscience have thus provided persuasive evidence for continuism. But they
have also provided evidence for discontinuism, and, after an initial wave of enthusiasm for
extreme forms of continuism, there has been a tendency to adopt more qualified views (Schacter
et al. 2012). Remembered past events, for example, are associated with richer and more vivid
sensory and contextual detail than imagined future events (D’Argembeau and Van der Linden
2004, 2006; Addis et al. 2010) or imagined past events (Addis et al. 2010; De Brigard and
Giovanello 2012), and the emotional valence of remembered and imagined episodes displays a
similar discrepancy, with imagined future events being characterized by a greater positivity bias
than remembered past events (Berntsen and Bohn 2010; Rasmussen and Bertsen 2013). Perhaps
this evidence of phenomenological discrepancies only suggests a difference of degree between
episodic memory and FMTT, but other evidence seems to suggest a difference in kind. Imaging
studies have revealed that imagining is more cognitively demanding than remembering and in
fact draws on brain regions that are not solicited by remembering (D’Argembeau and Van der
Linden 2004; Schacter and Addis 2007a; Addis et al. 2007; Szpunar et al. 2007; McDonough
230
and Gallo 2010; Martin et al. 2011), and there is evidence that impairments of certain brain
regions affect FMTT but not episodic memory (Berryhill et al. 2010). Indeed, some research-
ers have even argued that two subsystems can be distinguished within the core network, with
only episodic memory requiring reactivation of regions involved in the original processing of
remembered information (Addis et al. 2009), and others have argued that imagining future
events, in contrast to remembering past events, relies on conceptual knowledge to provide a
scaffolding for the integration of episodic details (Irish et al. 2012; Duval et al. 2012); conse-
quently, FMTT may be more schema-driven than episodic memory (Szpunar 2010; Irish et al.
2012; Duval et al. 2012; Rasmussen and Berntsen 2013; Klein and Steindam 2016).
Further support for discontinuism is provided by functional considerations. Remembering
and imagining may have different functions with respect to the self, emotion, and behav-
iour regulation (Shao et al. 2010; Rasmussen and Berntsen 2013). For instance, an increased
positivity bias for imagined future relative to remembered past events has been documented
and it has been proposed this would show that episodic memory is more devoted to prob-
lem solving and behaviour regulation, while FMTT is more dedicated to securing positive
self-image, regulating emotions and prompting us to approach novelty. And the constructive
episodic simulation hypothesis itself suggests that episodic memory has the function of storing
information which is then recombined in FMTT, thus assigning a distinctive functional role
to memory. Against this view, continuists have argued that, while episodic memory indeed has
this function, it performs the function with respect to the process of FMTT and the process
of episodic remembering alike, both of which involve the recombination of stored information
(Michaelian 2016a). But it may be possible to distinguish between a process of recombination,
at work in imagining the future, and a process of recapitulation, at work in remembering the
past (Addis et al. 2009; Storm and Jobe 2012), in which case there might still turn out to be a
functional difference between FMTT and episodic memory.
5. Philosophical arguments for discontinuism

Given that there is empirical evidence for both continuities and discontinuities between epi-
sodic memory and FMTT, such evidence is unlikely to settle the debate between continuists
and discontinuists. It is not immediately clear, for example, whether continuism is incompatible
with the existence of distinct subsystems for episodic memory and FMTT within the core net-
work, nor is it clear whether recombination and recapitulation amount to qualitatively different
functions. Conceptual considerations thus take centre stage, allowing philosophical arguments
to play a potentially decisive role in the debate. This section and the next review the main argu-
ments that have been advanced so far, focusing in turn on epistemological, metaphysical, and
phenomenological considerations.
Beginning with epistemology, one argument for discontinuism turns on the possibility or
impossibility of error through misidentification in episodic memory and FMTT (Perrin 2016).
On the one hand, when one episodically remembers, it is arguably possible for one to mistake
an individual (object or person) figuring in the remembered episode for a similar but distinct
individual, since the identity of the remembered individual is determined by one’s past causal
interactions with it. In other words, one can misidentify a remembered individual – episodic
memory is not immune to error through misidentification.4 On the other hand, when one
engages in FMTT, it is arguably impossible for one to mistake an individual figuring in the
imagined episode for another individual, since the identity of the imagined individual is in
effect stipulated by the imagining subject. In other words, one cannot misidentify an imagined
individual – FMTT is immune to error through misidentification.
231
Turning to metaphysics, another argument for discontinuism concerns the nature of the
mental contents involved in remembering the past and imagining the future. Intuitively, epi-
sodic memory provides the subject with a form of direct contact with past events. According
to relationalists about the objects of memory (e.g. Campbell 2002; Debus 2008, 2013, 2014;
cf. Recanati 1993, 2007; Perrin 2016), when a subject episodically remembers an event, the
remembered event literally constitutes part of the content of his memory. Consequently, the
event must play a role in the individuation of the memory content. In this sense, episodic
memory is ‘world-involving’. When a subject imagines the future, however, the imagined
event cannot play a role in the individuation of content, since, in order for an event to consti-
tute part of the content of a mental state, it must be actual, not merely possible. Thus FMTT is
not world-involving.
A related but distinct metaphysical argument concerns the particularity of the objects of
MTT. Episodic memory seems to involve thinking about particular events with which one is
acquainted through experience. FMTT, in contrast, seems to involve thinking about general
types of events. For example, some have suggested that we imagine future events by thinking of
general types of events and mentally projecting them forwards in time (Martin 2001; Campbell
2002; Debus 2008, 2014). While this argument is distinct from the previous argument, the
generality-particularity discrepancy may stem from the possibility-actuality discrepancy, assum-
ing that only actual events can be particular (Debus 2014).
A final metaphysical argument, or rather a pair of arguments, concerns the role of causation
in episodic memory and FMTT. One argument maintains that episodic memory requires an
appropriate causal link with the remembered experience, in contrast to FMTT, which obvi-
ously cannot require an analogous link (Perrin 2016). Consider a case of ‘quasi-memory’ in
which the identity of the subject whose experiences are ‘remembered’ does not coincide with
the identity of the subject who ‘remembers’ the experiences. In such a case, the ‘remembering’
subject is arguably not remembering at all, but merely imagining. When a subject imagines the
future, in contrast, he would seem to be engaged in the same process regardless of whether
he imagines his own experience or that of another subject. Another argument maintains that
episodic memory and FMTT involve different kinds of awareness of potential causal links
(Debus 2016). When one imagines a future event, one is arguably aware of the fact that it is
causally open, in the sense that someone—perhaps oneself—might intervene to bring about or
prevent the occurrence of the event. When one imagines a past event, however, one has no
such awareness.
Further arguments for discontinuism point to apparent qualitative phenomenological differ-
ences between episodic memory and FMTT. One such argument claims that episodic memory
alone involves experiential awareness (Debus 2014; cf. Martin 2001; Campbell 2002). This
claim is related to the point concerning the actuality of the objects of episodic memory noted
above. In order for someone to be experientially aware of an event, the event must presum-
ably be actual; in other words, one cannot be experientially aware of merely possible events.
Thus episodic memory, which concerns actual events, may involve experiential awareness, but
FMTT, which concerns possible events, cannot.
A distinct phenomenological argument claims that autonoetic consciousness arises in a dif-
ferent manner in episodic memory and FMTT (Klein 2016; Perrin 2016). In episodic memory,
awareness that the remembered event ‘belongs’ to oneself seems to arise in a prereflective man-
ner. In FMTT, in contrast, it seems to be the result of a deliberate, inferential process. Indeed,
while some have taken autonoesis to be a necessary feature of FMTT, others have argued that
a simulated future experience remains essentially unaltered when the identity of the imagined
232
subject shifts from the identity of the imagining subject to that of another subject. If this is right,
autonoetic consciousness, which concerns the self, turns out not to be a necessary feature of
FMTT, strengthening the contrast between FMTT and episodic memory (Perrin 2016).
6. Philosophical arguments for continuism

These arguments for discontinuism fit well with traditional philosophical analyses of remember-
ing, one of the main goals of which has been to provide criteria for demarcating remembering
from imagining. The causal theory, in particular, presupposes that there is a difference in kind
between remembering the past and imagining it and maintains that this difference is to be
understood in terms of the presence or absence of an appropriate causal link with the repre-
sented event. What makes the difference between remembering a past event and imagining it,
according to the theory, is that the former involves an appropriate causal link with the repre-
sented event, while the latter does not. According to standard versions of the causal theory, what
makes a causal link appropriate, in the relevant sense, is that it is sustained by the preservation of
traces of the subject’s original experience of the event.
The presupposition that there is a difference in kind between remembering the past and
imagining it is threatened by the MTT framework, and some philosophers have argued that it
should be abandoned entirely. MTT research suggests that the boundary between remembering
the past and imagining it may be blurry (Shanton and Goldman 2010) and perhaps even that
there is no boundary at all (Michaelian 2016b). In line with the constructive episodic simulation
hypothesis, simulationist accounts of episodic memory thus reject the goal of providing criteria
for demarcating remembering from imagining, instead classifying remembering as a kind of
imagining. According to these accounts, there is no difference in kind between remembering
the past and imagining it – to remember simply is to imagine the past.
Assuming that there is no difference in kind between imagining the past and imagining the
future, simulationism entails continuism. It would be question-begging to appeal directly to
simulationism in an attempt to defeat discontinuist arguments, and such an appeal would over-
look the fact that MTT research has also provided evidence for discontinuism. Nevertheless,
simulationism has a role to play in the debate. The empirical evidence for discontinuism is
compatible with a broadly simulationist approach. The claim that distinct subsystems of the
core network are involved in episodic memory and FMTT, for example, is compatible with the
claim that episodic memory and FMTT are both simulational in character. And if remember-
ing is simulational in character, it need not involve a causal link – simulation inevitably draws
on traces originating in past experience, but simulation of a given past event need not draw on
traces originating in experience of that particular event. Arguments for discontinuism assume
that such a causal link is necessary, and the simulationist approach, by reminding us that that
assumption is problematic, points to potential responses to those arguments.
As we saw above, for example, discontinuists might claim that there is an epistemological
difference between episodic memory and FMTT, arguing that the latter is immune to error
through misidentification, whereas the former is not. The idea is that, since FMTT is a form
of imagination, the identities of the individuals figuring in the representations produced by it
are simply stipulated by the subject, whereas, since episodic memory requires a causal link with
past experience, the identities of the individuals figuring in the representations produced by it
may be inherited from the subject’s past causal interactions with them, opening up the possibil-
ity of error in the latter case but not the former. If we accept that both episodic memory and
FMTT are simulational in character, however, this apparent discontinuity may be undermined
233
(Michaelian 2016a). On the one hand, since both episodic memory representations and FMTT
representations are generated by simulations based on past experience, identities may be inher-
ited from causal interactions in both cases. On the other hand, since both episodic memory
representations and FMTT representations are produced by simulations, identities may not be
inherited in a straightforward manner, and stipulation by the subject may play a role in both cases.
In short, both forms of MTT may sometimes be immune to error through misidentification and
sometimes fail to be immune to error through misidentification.
Turning to metaphysical arguments, we saw that discontinuists might argue that past events
are constituents of episodic memories, whereas future events are not constituents of episodic
future thoughts. There are real worries about how a past event might come to serve as part of
a current mental state, but even if we set these aside, the argument encounters a difficulty analo-
gous to one known to afflict direct realism in the philosophy of perception. Direct realists argue
that perceived objects are constituents of perceptual states. This forces them to posit a difference
in kind between genuine perceptual states and hallucinations, a position known as disjunctiv-
ism. Similarly, discontinuists are forced to posit a difference in kind between genuine memories
and ‘memory hallucinations’ – i.e. false memories – and thus to adopt a form of disjunctivism
about memory. Disjunctivism is implausible in both instances, because it posits metaphysical
differences where there are no corresponding empirical differences: in cases of genuine per-
ception and perceptual hallucination, everything might be the same in neural, cognitive, and
phenomenological terms, and the same thing goes for cases of genuine memory and memory
hallucination. Discontinuists might insist that the presence of a causal link with a past event, in
the case of genuine memory, grounds a metaphysical difference between genuine memory and
memory hallucination, but, if remembering is a simulational process, there need be no such link.
Moreover, the initial intuitive appeal of direct realism or relationalism is largely undermined by
the simulational character of remembering: if there need be no causal link between a memory of
a past event and the subject’s original experience of it, the claim that the past event is a constituent
of the subject’s memory of it becomes difficult to motivate.
Discontinuists have also argued that episodic memory involves thinking about particular
events, whereas FMTT involves thinking about generic events. Bearing in mind the simulational
character of both episodic memory and FMTT, continuists might respond by arguing that, if
episodic memory can involve thinking about past events, then so can FMTT and that, if FMTT
can involve thinking about generic events, then so can episodic memory (Michaelian 2016a).
On the one hand, we might sometimes imagine future events not by imagining generic events
and projecting them forwards in time, as discontinuists suggest, but rather by remembering past
events and projecting them forwards in time. In such cases, if the memory refers to a particular
event, then so, presumably, does the imagination. On the other hand, we might sometimes
remember past events by imagining generic events and projecting them backwards in time. In
such cases, if the imagination refers to a generic event, then so, presumably, does the memory.
The arguments discussed so far depend indirectly on assumptions about causation, but, as
we saw above, discontinuists have also argued more directly that episodic memory and FMTT
differ with respect to causation. Focusing on memory, they have argued straightforwardly that
a causal link to the represented event is necessary in the case of remembering, in contrast to
FMTT, where it is neither necessary nor possible. Here, continuists can reply that, given what
we know about the simulational nature of remembering, there can be no guarantee that any
given case of remembering involves a causal link to the represented event. Discontinuists are
free to insist that only where there is a causal link does genuine remembering occur, but, again,
this is to posit a metaphysical difference without a corresponding empirical difference – MTT
234
research suggests that remembering involves the same process, regardless of whether a causal
link is preserved. The continuist alternative is to treat cases of quasi-memory, for example, not
as cases in which the subject merely seems to remember but rather as cases in which the subject
genuinely remembers but does so inaccurately. Focusing on FMTT, discontinuists have argued
that FMTT involves awareness of the causal openness of the future; episodic memory does
not involve a parallel form of awareness, since the past is causally closed. Given the similarity
between the simulational processes involved in episodic memory and FMTT, however, con-
tinuists can argue that we should expect to find that subjects’ judgements about whether they
are remembering the past or imagining the future are sometimes mistaken (Michaelian 2016b).
Whether subjects do indeed make such mistakes is an empirical question, but, assuming that
they do, there will be cases of FMTT which fail to involve awareness of causal openness, as well
as cases of episodic memory which involve (mistaken) awareness of causal openness.
An appreciation of the simulational character of remembering likewise grounds continuist
responses to arguments for phenomenological discontinuities. Discontinuists have argued that
episodic memory, because it refers to actual events, can involve experiential awareness, whereas
FMTT, because it does not refer to actual events, cannot. Absent a causal link to the event
in question, sustained by the preservation of traces of the subject’s experience of the event,
the actuality of the event cannot underwrite experiential awareness. Research on constructive
memory and MTT demonstrates that, in routine cases, many components of the memory rep-
resentation are imported from sources other than experience of the remembered event. In such
cases, the subject is presumably not experientially aware of all aspects of the event. And the same
research suggests that there are cases in which all components of the memory representation
are imported from sources other than experience of the remembered event. In such cases, the
subject is presumably not experientially aware of any aspect of the event. Discontinuists might
argue that genuine remembering occurs only where at least some trace of the original experi-
ence is preserved, but, given that similar simulational processes are at work in the full range of
cases, it would seem to be preferable to abandon talk of experiential awareness altogether.
In an argument for another sort of phenomenological asymmetry, discontinuists have argued
that episodic memory and FMTT are unalike in that only the former necessarily involves auto-
noetic consciousness. The thought here is that, since episodic memory necessarily concerns
experienced events, it may involve autonoetic consciousness as an essential feature; since FMTT
might concern either oneself or another, autonoetic consciousness in FMTT must be the out-
come of an inferential process. The simulational character of remembering may undermine
this alleged disanalogy. If both past-oriented MTT and future-oriented MTT are simulational
processes, both might concern either oneself or another. There may be a difference between
simulating one’s own past experience (i.e. remembering) and simulating the past experience
of another, but, if so, there is presumably likewise a difference between simulating one’s own
future experience and simulating the future experience of another. Autonoetic consciousness
might thus be absent in both MTT into the past of another and MTT into the future of another
and present in both MTT into one’s own past and MTT into one’s own future. If it is the out-
come of an inferential process in the case of future-oriented MTT, it is presumably the outcome
of an inferential process in the case of past-oriented MTT as well.
7. Conclusion: continuism, discontinuism, and causation

Ultimately, the debate between continuists and discontinuists seems to boil down to a disa-
greement over the necessity of causation for episodic memory. If episodic memory, despite
235
its simulational character, necessarily involves a causal link to the subject’s experience of the
remembered event, discontinuism would appear to be the more plausible of the two views. If
episodic memory does not necessarily involve a causal link to the subject’s experience of the
remembered event, continuism may carry the day.
MTT research, we have seen, does suggest that remembering need not involve the pres-
ervation of traces of the subject’s experience of the remembered event. But this might not
be enough to undermine the claim that remembering necessarily involves a causal link to the
subject’s experience of the remembered event. Standard versions of the causal theory assume
that appropriate causal links are sustained by the preservation of traces of experience, but there
are alternatives to this version of the theory. Attributionalist accounts of memory, in particular,
favour a procedural characterization of the causal links involved in remembering (Jacoby et al.
1989; Mitchell and Johnson 2000; Addis et al. 2009), claiming that the detection of procedural
properties (such as fluency) of subpersonal processes triggers attribution to past experience.
Such properties, in turn, reflect past experience, so, if attributionalism is right, episodic memory
would result from the monitoring of a procedural effect of past experience. Attributionalism
has considerable advantages. It allows us to preserve the distinction between remembering and
merely imagining the past which is abandoned by simulationist approaches, since fluency can
trigger attribution to past experience even when it is due to other causes. And it can explain the
observation that greater neural activity is observed in FMTT than in episodic memory, since the
simulation involved in episodic memory will in general be more fluent, due to the effect of past
experience, than that involved in FMTT. But its main import, in the present context, is that it
promises to identify a causal difference between episodic memory and FMTT, which may in
turn sustain the various epistemological, metaphysical, and phenomenological differences for
which discontinuists argue.
Whether attributionalism can ultimately fulfil this promise remains to be seen. Attributionalism
may secure a causal difference between episodic memory and FMTT without securing a differ-
ence sufficient to sustain the qualitative differences that discontinuists posit. And it may yet turn
out that attributionalism does not in fact secure a causal difference between episodic memory
and FMTT. In essence, attributionalism says that representations resulting from more fluent
simulations tend to be attributed to past experience. Fluency, in turn, is typically an effect of past
experience. Thus, in cases where the subject takes himself to be remembering, there will usu-
ally be a causal link with past experience. There is, however, no guarantee of such a link, since
fluency may result from other causes. When it does, discontinuists must claim that the subject
merely seems to remember, on penalty of admitting that a causal link is not in fact necessary for
remembering. Continuists, however, will respond that this claim is unmotivated: if the same
simulational process unfolds in two cases, and if both cases are characterized by the same level
of fluency, both should be categorized as cases of remembering. In the final analysis, then, the
continuist–discontinuist debate may bottom out in a clash of intuitions over the necessity of
causation for remembering.
Notes
1 Psychologists have tended to favour continuism, whereas philosophers have tended to favour discon-
tinuism. But the theoretical divide need not align with the disciplinary border, and some psychologists
remain sceptical of the evidence for continuism (e.g. Friedman 2007; Klein and Steindam 2016), while
some philosophers have developed continuist analyses that classify episodic memory and FMTT alike as
forms of imagination (Hopkins 2014; Michaelian 2016b).
2 See Suddendorf and Corballis (2007), and Szpunar (2010) for further detail.
236
3 Compare, e.g. the self-projection approach (Buckner and Carroll 2007) and the scene construction
approach (Hassabis and Maguire 2009).
4 There is a debate among philosophers as to whether episodic memories are immune to error through
misidentification or not. Evans (1982) and Hamilton (2007) answer in the affirmative, while Shoemaker
(1970) and Coliva (2006) maintain this is only a contingent feature of remembering. Our take here is
closer to the latter, which we cannot argue for here for reason of space.
Related topics
• Taxonomy and unity of memory
Further reading
For an introduction to mental time travel, see Chapter 6 of Michaelian, K. 2016b. Mental Time Travel:
Episodic Memory and Our Knowledge of the Personal Past. Cambridge, MA: MIT Press.
References
Addis, D. R., R. P. Roberts, and D. L. Schacter (2011) Age-related neural changes in autobiographical
remembering and imagining. Neuropsychologia 49(13): 3656–69.
Addis, D. R., A. T. Wong, and D. L. Schacter (2007) Remembering the past and imagining the future:
Common and distinct neural substrates during event construction and elaboration. Neuropsychologia
45(7): 1363–77.
Addis, D. R., R. Musicaro, L. Pan, and D. L. Schacter (2010) Episodic simulation of past and future events
in older adults: Evidence from an experimental recombination task. Psychology and Aging 25(2): 369–76.
Addis, D. R., L. Pan, M.-A. Vu, N. Laiser, and D. L. Schacter (2009) Constructive episodic simulation of
the future and the past: Distinct subsystems of a core brain network mediate imagining and remember-
ing. Neuropsychologia 47(11): 2222–38.
Atance, C. M. (2008) Future thinking in young children. Current Directions in Psychological Science 17(4):
295–8.
Bartlett, F. C. (1932) Remembering: A Study in Experimental and Social Psychology. Cambridge: Cambridge
University Press.
Bernecker, S. (2010) Memory: A Philosophical Study. New York: Oxford University Press.
Berntsen, D. and A. Bohn (2010) Remembering and forecasting: The relation. Memory & Cognition 38(3):
265–78.
Berryhill, M. E., L. Picasso, R. Arnold, D. Drowos, and I. R. Olson (2010) Similarities and differ-
ences between parietal and frontal patients in autobiographical and constructed experience tasks.
Neuropsychologia 48(5): 1385–93.
Brewer, W. F. (1996) What is recollective memory? Remembering Our Past: Studies in Autobiographical
Memory. Ed. D. C. Rubin. Cambridge: Cambridge University Press.
Buckner, R. and D. Carroll (2007) Self-projection and the brain. Trends in Cognitive Sciences 11(2): 49–57.
Campbell, J. (2002) Reference and Consciousness. New York: Oxford University Press.
Coliva, A. (2006) Error through misidentification: Some varieties. Journal of Philosophy 103(8): 403–25.
D’Argembeau, A. and M. Van der Linden (2004) Phenomenal characteristics associated with projecting
oneself back into the past and forward into the future: Influence of valence and temporal distance.
Consciousness and Cognition 13(4): 844–58.
D’Argembeau, A. and M. Van der Linden (2006) Individual differences in the phenomenology of mental
time travel: The effect of vivid visual imagery and emotion regulation strategies. Consciousness and
Cognition 15(2): 342–50.
237
D’Argembeau, A., O. Renaud, and M. Van der Linden (2011) Frequency, characteristics and functions of
future-oriented thoughts in daily life. Applied Cognitive Psychology 25(1): 96–103.
De Brigard, F. and K. S. Giovanello (2012) Influence of outcome valence in the subjective experience of
episodic past, future, and counterfactual thinking. Consciousness and Cognition 21(3): 1085–96.
Debus, D. (2008) Experiencing the past: A relational account of recollective memory. Dialectica 62(4):
405–32.
Debus, D. (2013) Thinking about the past and experiencing the past. Mind & Language 28(1): 20–54.
Debus, D. (2014) ‘Mental time travel’: Remembering the past, imagining the future, and the particularity
of events. Review of Philosophy and Psychology 5(3): 333–50.
Debus, D. (2016) Temporal perspectives in imagination: On the nature and value of imagining the future.
Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel. Eds. K. Michaelian,
S. B. Klein, and K. K. Szpunar. New York: Oxford University Press.
Duval, C., B. Desgranges, V. de La Sayette, S. Belliard, F. Eustache, and P. Piolino (2012) What happens to
personal identity when semantic knowledge degrades? A study of the self and autobiographical memory
in semantic dementia. Neuropsychologia 50(2): 254–65.
Evans, G. (1982) The Varieties of Reference. New York: Oxford University Press.
Friedman, W. J. (2007) The meaning of time in episodic memory and mental time travel. Behavioral and
Brain Sciences 30(June): 323.
Hamilton, A. (2007) Memory and self-consciousness: Immunity to error through misidentification.
Synthese 171: 409–17.
Hassabis, D. and E. A. Maguire (2009) The construction system of the brain. Philosophical Transactions of the
Royal Society B: Biological Sciences 364(1521): 1263–71.
Hassabis, D., D. Kumaran, S. D. Vann, and E. A. Maguire (2007) Patients with hippocampal amnesia
cannot imagine new experiences. Proceedings of the National Academy of Sciences 104(5): 1726–31.
Hopkins, R. (2014) Episodic memory as representing the past to oneself. Review of Philosophy and Psychology
5(3): 313–31.
Irish, M., D. R. Addis, J. R. Hodges, and O. Piguet (2012) Considering the role of semantic memory in
episodic future thinking: Evidence from semantic dementia. Brain 135(7): 2178–91.
Jacoby, L. L., C. M. Kelley, and J. Dywan (1989) Memory attributions. Varieties of Memory and Consciousness:
Essays in Honour of Endel Tulving. Eds. H. L. Roediger and F. I. M. Craik. Hillsdale, NJ: Lawrence
Erlbaum.
Klein, S. B. (2016) Autonoetic consciousness: Reconsidering the role of episodic memory in future-
oriented self-projection. Quarterly Journal of Experimental Psychology 69(2): 381–401.
Klein, S. B. and C. Steindam (2016) The role of subjective temporality in future-oriented mental time
travel. Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel. Eds. K. Michaelian,
Klein, S. B., J. Loftus, and J. F. Kihlstrom (2002) Memory and temporal experience: The effects of epi-
sodic memory loss on an amnesic patient’s ability to remember the past and imagine the future. Social
Cognition 20(5): 353–79.
McDonough, I. and D. Gallo (2010) Separating past and future autobiographical events in memory:
Evidence for a reality monitoring asymmetry. Memory & Cognition 38(1): 3–12.
Martin, C. B. and M. Deutscher (1966) Remembering. Philosophical Review 75(2): 161–96.
Martin, M. G. F. (2001) Out of the past: Episodic recall as retained acquaintance. Time and Memory: Issues
in Philosophy and Psychology. Eds. C. Hoerl and T. McCormack. New York: Oxford University Press.
Martin, V. C., D. L. Schacter, M. C. Corballis, and D. R. Addis (2011) A role for the hippocampus in
encoding simulations of future events. Proceedings of the National Academy of Sciences 108(33): 13858–63.
Michaelian, K. (2016a) Against discontinuism: Mental time travel and our knowledge of past and future
events. Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel. Eds. K. Michaelian,
Michaelian, K. (2016b) Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past.
Michaelian, K., S. B. Klein, and K. K. Szpunar (2016) The past, the present, and the future of future-
oriented mental time travel. Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time
Travel. Eds. K. Michaelian, S. B. Klein, and K. K. Szpunar. New York: Oxford University Press.
Mitchell, K. J. and M. K. Johnson (2000) The Oxford Handbook of Memory. Eds. E. Tulving and F. I. M.
Craik. New York: Oxford University Press.
238
Perner, J., D. Kloo, and M. Rohwer (2010) Retro- and prospection for mental time travel: Emergence
of episodic remembering and mental rotation in 5- to 8-year-old children. Consciousness and Cognition
19(3): 802–15.
Perrin, D. (2016) Asymmetries in subjective time. Seeing the Future: Theoretical Perspectives on Future-Oriented
Mental Time Travel. Eds. K. Michaelian, S. B. Klein, and K. K. Szpunar. New York: Oxford University
Press.
Rasmussen, A. and D. Berntsen (2013) The reality of the past versus the ideality of the future: Emotional
valence and functional differences between past and future mental time travel. Memory & Cognition
41(2): 187–200.
Rathbone, C. J., M. A. Conway, and C. J. A. Moulin (2011) Remembering and imagining: The role of
the self. Consciousness and Cognition 20(4): 1175–82.
Recanati, F. (1993) Direct Reference: From Language to Thought. Oxford: Blackwell.
Recanati, F. (2007) Perspectival Thought: A Plea for (Moderate) Relativism. New York: Clarendon Press.
Rosenbaum, R. S., S. Köhler, D. L. Schacter, M. Moscovitch, R. Westmacott, S. E. Black, F. Gao, and
E. Tulving (2005) The case of K.C.: Contributions of a memory-impaired person to memory theory.
Neuropsychologia 43(7): 989–1021.
Schacter, D. L. and D. R. Addis (2007a) The cognitive neuroscience of constructive memory: Remembering
the past and imagining the future. Philosophical Transactions of the Royal Society B: Biological Sciences
362(1481): 773–86.
Schacter, D. L. and D. R. Addis (2007b) Constructive memory: The ghosts of past and future. Nature
445(7123): 27.
Schacter, D. L., D. R. Addis, and R. L. Buckner (2007) Remembering the past to imagine the future: The
prospective brain. Nature Reviews Neuroscience 8(9): 657–61.
Schacter, D. L., D. R. Addis, D. Hassabis, V. C. Martin, R. N. Spreng, and K. K. Szpunar (2012) The
future of memory: Remembering, imagining, and the brain. Neuron 76(4): 677–94.
Shanton, K. and A. Goldman (2010) Simulation theory. Wiley Interdisciplinary Reviews: Cognitive Science
1(4): 527–38.
Shao, Y., X. Yao, S. J. Ceci, and Q. Wang (2010) Does the self drive mental time travel? Memory 18(8):
855–62.
Shoemaker, S. (1970) Persons and their pasts. American Philosophical Quarterly 7(4): 269–85.
Storm, B. C. and T. A. Jobe (2012) Remembering the past and imagining the future: Examining the
consequences of mental time travel on memory. Memory 20(3): 224–35.
Suddendorf, T. (2010) Episodic memory versus episodic foresight: Similarities and differences. Wiley
Interdisciplinary Reviews: Cognitive Science 1(1): 99–107.
Suddendorf, T. and J. Busby (2005) Making decisions with the future in mind: Developmental and
comparative identification of mental time travel. Learning and Motivation 36(2): 110–25.
Suddendorf, T. and M. C. Corballis (2007) The evolution of foresight: What is mental time travel, and is
it unique to humans? Behavioral and Brain Sciences 30(03): 299–313.
Szpunar, K. K. (2010) Episodic future thought: An emerging concept. Perspectives on Psychological Science
5(2): 142–62.
Szpunar, K. K., J. M. Watson, and K. B. McDermott (2007) Neural substrates of envisioning the future.
Proceedings of the National Academy of Sciences 104(2): 642–7.
Tulving, E. (1972) Episodic and semantic memory. Organization of Memory. Eds. E. Tulving and
W. Donaldson, pp. 381–402. New York: Academic Press.
Tulving, E. (1985) Elements of Episodic Memory. New York: Oxford University Press.
Tulving, E. (2002) Episodic memory: From mind to brain. Annual Review of Psychology 53(1): 1–25.
Viard, A., B. Desgranges, F. Eustache, and P. Piolino (2012) Factors affecting medial temporal lobe
engagement for past and future episodic events: An ALE meta-analysis of neuroimaging studies. Brain
and Cognition 80(1): 111–25.
Wagoner, B. (2016) The Constructive Mind: Frederic Bartlett’s Psychology in Reconstruction. Cambridge:
Williams, N. Ellis, C. Tyers, H. Healy, G. Rose, and A. Macleod (1996) The specificity of autobiographical
memory and imageability of the future. Memory & Cognition 24(1): 116–25.
239
PART VI
The social dimension of memory

19
EXTENDED MEMORY
Robert W. Clowes
1. A new technological ecology of memory?

In the early 1990s, Merlin Donald provided an agenda-setting approach to external memory
when he distinguished between biological memory that resides within the brain, and external
memory that
may reside in a number of different external stores, including visual and electronic
storage systems, as well as culturally transmitted memories that reside in other individ-
uals. The key feature is that it is external to the biological memory of a given person.
(Donald 1991: 308)
Donald argued that it is external memory, not our biological capacities, that is responsible for
all of our complex cultural achievements, and makes us human.
Throughout the 1990s, researchers, prompted by Edwin Hutchins’ analyses of Distributed
Cognition (Hutchins 1995) and Andy Clark’s situated/embodied (Clark 1997) and then
extended mind views (Clark and Chalmers 1998), pointed toward a need for the human
mind to be understood not as an individual biological entity, but instead as inseparable from
and deeply reliant upon a background of cultural activities and material culture. On distrib-
uted, extended, and embedded theories of cognition, our cognitive life is essentially bound
up with culture, other people and especially material culture beyond the biological body.
This perspective raises questions about why and how we develop and recruit any particular
cognitive tools, as well as what the limits and nature of the effects of material culture may be
on the workings of our minds.
These questions are posed with new urgency by the way the twenty-first century confronts
us with a host of new digital media technologies which are rapidly restructuring our cognitive
habits and abilities. This new regime of extended memory technology, dubbed “E-memory”
(Bell and Gemmell 2009; Clowes 2013), is archetypally mediated through an array of ever-
present mobile and wearable devices such as smartphones and tablets. These devices connect
us to cloud technology: the now ubiquitous Internet where massive data warehouses accessed by
wireless Internet technology provide a vast array of highly personalized informational services.
243
Robert W. Clowes
These services are often anchored to individual profiles which store and provide personalized
information while supporting and tracking our every action.
Which memory functions might this new technology effect? Sellen and Whittaker (2010)
note that E-memory can support and perhaps replicate a variety of human memory func-
tions including the five Rs of recollecting, reminiscing, retrieving, reflecting and remembering
intention. This last, remembering intention, or prospective memory is already widely imple-
mented in a host of gadgetry that prompt and regulate our actions, such as the incorporated
calendar, planning and alarm functions of applications such as Google Calendar or Microsoft
Outlook (see Smart et al. 2017). Such applications often now reside in the cloud and interact
with us through our mobile and wearable devices. They form an ever-present accompaniment
that many of us constantly interact with and rely upon to structure our lives. We can view this
technology as an externalization of existing cognitive functions. But does it also represent a
replacement of existing cognitive functions? Might it degrade our biological and/or autono-
mous abilities? Or is it a new part of ourselves? Might the new regime of technology actually
enhance some of our cognitive abilities?
To tackle these questions, in Section 2, we discuss some of the properties of the new
E-memory systems concentrating on totality, autonomy, social entanglement, and incorporabil-
ity. We discuss ways in which they contrast and complement biological memory. In Section 3,
we introduce Wegner’s framework for analysing how memory is socially distributed amongst
human beings: transactive memory. We discuss how some have used this framework to analyse
human/technological systems, and one controversial claim that the Internet is a “supernormal
stimulus” disrupting normal human memory distribution. Section 4 examines the Extended
Mind (EM) hypothesis which suggests that external resources may count as parts of our minds.
EM offers the theoretical possibility that even if we come to replace many of our biologi-
cal memory capacities with artificial ones, this may not entail that we are thereby cognitively
diminished. It is clear that external memory systems embody different properties from bio-
logical systems. One major difference is that biological memory is highly reconstructive. We
review the discussion over whether such “fine-grained” biological differences are the most use-
ful for developing theories of mind. Section 5 returns to how we should understand ourselves
as agents against the background of cloud technology using this deepened theoretical apparatus.
It attempts to address whether and how our heavy reliance on a new regime of E-memory may
change our biological capacities, our cognitive abilities and nature as agents.
In so doing, we take on important philosophical questions that intersect with the increas-
ing importance of external memory in (especially) individual human lives, namely: Is extended
memory really memory? What is the nature of the interaction between Extended Memory and
Biological Memory? Might our usage of this new regime of memory diminish or even poten-
tially enhance our existing biological memory and our wider cognitive abilities?
2. E-memory, complementarity and practical incorporation

Donald (1991) suggests the deep underlying tendency of extended memory technologies has
been to make it possible to remember that which our unenhanced biological systems would
otherwise forget. The history and pre-history of regimes of external memory technology indi-
cate that they do not tend simply to duplicate our biological memory profiles. On the contrary,
as the archaeological record shows us, we seem to build extended cognitive systems that func-
tion to circumvent some of the fragilities of our native biological systems and innovate beyond
their limitations (Malafouris 2013).
244
Extended memory
John Sutton notes that tools are recruited, not because they duplicate biological cognitive
functions, but because they offer something different. Sutton’s complementarity principle
suggests that
In extended cognitive systems, external states and processes need not mimic or repli-
cate the formats, dynamics or functions of inner states and processes. Rather, different
components of the overall (enduring or temporary) system can play quite different
roles and have different properties while coupling in collective and complementary
contributions to flexible thinking and acting.
(Sutton 2010: 194)
External resources will tend to be incorporated when they offer functions that our unenhanced
brains cannot.
E-memory devices afford cognitive properties which are arguably very different from previ-
ous technological regimes (Clowes 2013). A central tendency is toward totality: the possibility,
made available by advances in the capacity and speed of digital media, to record high-fidelity fac-
similes of every event as it happens (Bell and Gemmell 2009). The SenseCam, the pre-eminent
and most widely used research tool to investigate this tendency to date, is a digital camera with
a fisheye lens that can fit into the palm of a hand, or more typically, worn on a chain around a
user’s neck. It can be set to take a picture every couple of seconds or whenever its face recogni-
tion software detects a person approaching its wearer. The SenseCam can be used to capture a
photographic digital record of a day’s event, saving it up to a database or the Internet (Hodges
et al. 2006; Sellen and Whittaker 2010). Pioneers like Gordon Bell have used this and other
technologies to “remember everything” by attempting to create a complete digital record of
their lives. Although such technologies have so far only been used in such extreme ways by
scientists and enthusiasts, the drive toward recording ever more digital traces of our everyday
lives is becoming a mass pursuit. Millions of us carry portable technologies like digital phones
and tablets which make ever more detailed and complete digital records of our lives. As search
technology becomes better at retrieving all this data, the tendency to use it to augment or replace
our biological capabilities becomes ever-greater.
E-memory can also be autonomous in a way no previous extended memory has been. A book
left in a library could reasonably be expected to be the same when you come back to it; perhaps
with underlining added by other readers. Digital memory traces in our current moment are in
a constant process of transformation, mined by apps looking for novel statistical regularities, and
open to myriad forms of annotations and modification. A set of photographs stored on Google
might be recolourized, turned into a video clip or automatically tagged to include the names
or other attributes of those photographed. Returning to a digital photograph in ten years’
time (assuming it is still there) will likely not return the memory trace you stored. Content has
become fluid and malleable in ways that partially mirror, albeit inexactly, the reconstructive
character of biological memory.
E-memory systems are also socially entangled. What we might recall with systems like Facebook
is not merely what we upload, but what others tag, comment upon, or connect to. A digital
memory trace that my Facebook friends interact with will tend to be more available both to
myself and others in the future. Traces that are ignored will tend to continue to be ignored.
The algorithms and interactions of social media parallel some reconstructive aspects of individual
biological memory, but the precise nature of E-memory reconstructions are determined by soft-
ware engineers and social interactions. These are very different from the evolutionary processes
245
Robert W. Clowes
that shaped our biological memory systems. This raises questions over whether such multiply
reconstructable and labile memory traces should ever be counted on a par with memories cap-
tured by our biological systems.
E-memory devices are at our fingertips and ripe for deployment and incorporation in a host
of cognitive operations. They have become the constant background context of our cogni-
tive and emotional lives. Often it is easier to rely on these resources than our own biological
capacities, and so, how we use our biological memories is also undergoing rapid change. This
ubiquitous readiness for deployment into so many diverse cognitive operations may be consid-
ered a cognitive property in itself: incorporability (Clowes 2013). For some, this tendency to lean
upon the resources of the Internet as a one-size-fits-all solution to all of our cognitive problems
is itself a worrying problem. It is claimed that our tendency to over-rely on these resources is
an unwelcome extension of basic human capacity to form socially distributed memory systems
which is now undermining individual human memory.
3. Transactive memory and its extensions

The notion of transactive memory (TM) “draws deeply on the analogy between the mental
operations of the individual and the processes of the group” (Wegner 1987: 85). Daniel Wegner
(1948–2013) originally used the idea to illustrate the workings of what he called the group mind:
the transactive memory system in a group involves the operation of the memory systems
of the individuals and processes of communication that occur within the group.
Transactive memory therefore is not traceable to any of the individuals alone, nor can
it be found somewhere “between” individuals. Rather, it is a property of a group.
(Wegner 1987: 85)
The TM approach holds that this social distribution is a fundamental property of human memory.
TM systems rely upon this cognitive division of labour. In order to function, someone either
formally or informally becomes considered the expert on any particular memory domain and is
then relied upon in order to store and recall the information at a future time. In his 1987 arti-
cle, Wegner looks at several sorts of TM systems, from intimate couples, through relationships
between a professor and a student, to organizations such as a firm where a TM system needs to
be explicitly developed.
A paradigm case is collaborative recall in older couples (Harris et al. 2011), where memory
is retrieved through a discursive interaction between the couple. Within such relationships,
one partner might become the specialist in remembering birthdays and important occasions,
the other in which bills have been paid. One advantage of such collaborative recall is that it can
actually help aid failing memory in older couples (Johansson et al. 2005). Apart from the actual
division of labour of storing memory, there must be a system of meta-memory where mem-
bers of the group can find the person who has responsibility for remembering any particular
domain of information. In dyads, this is likely not very difficult, but such meta-memory might
be quite complex in itself for moderately complex organizations. First, it requires individuals
to be experts on a given domain of memory, and second, it requires other individuals to be
able to find the right expert.
TM and collaborative recall can be contrasted with more mainstream experimental psy-
chology, which has focused on core internal biological and especially neural processes of the
individual, which still, often uncritically see social (and technological) influences on memory
as “inherently disrupting or distorting” (Harris et al. 2010). Hitherto, much experimental
246
Extended memory
psychology tended to create ad hoc TM systems for laboratory-based experiments. (Individuals

do indeed appear to perform worse on a host of memory tasks such remembering lists of
words when asked to collaborate with strangers in laboratory conditions.) Much empirical
memory research continues to be primarily individualistic in orientation (Michaelian and
Sutton 2013). Work on collaborative recall and other examples of TM in more naturalistic
settings, however, show that intimate groups can demonstrate powerful memory-enhancing
effects (Harris et al. 2011).
Some of Wegner’s colleagues have been applying the idea of TM to our use of Internet
and computer technology. From early on, Wegner (1987) had defined memory in the func-
tional terms of information processing, in a rough-grained manner. Memory just is, on this
“received view” the encoding, storage, and retrieval of information. On this conceptualiza-
tion, Internet technology itself can be seen as a transactive memory partner (Ward 2013). Our
everyday experience of technologies like Google Search tends to cause us to factor them into
our cognitive activities in ways that are analogous to the way we treat partners in transactive
memory systems. The problem is that the Internet, when considered as a memory system, has
its own properties – as we have seen – which are very different from the memory systems of
individual persons.
One worry is that we may quickly come to just rely on the Internet – as a sort of universal
expert – to encode memories for us, and thereby not bother ourselves. Indeed, evidence from
experiments reported by Sparrow and colleagues (2011) implies that we are already doing just
that. When subjects believe that they will later be able to access a computer file, they tend to
forget that file’s contents, but remember how to access it. This work does indeed fit nicely
into the TM paradigm where members of a transactive memory system will tend to remember
only how to access information when a handy expert can be relied on to remember the actual
detail for them. This lends weight to the hypothesis that we might treat the Internet itself as a
transactive memory partner. Familiar Internet search engines might quickly become considered
universal experts on all matters allowing us to forget almost everything. As Sparrow and col-
leagues put it: “Because search engines are continually available to us, we may often be in a
state of not feeling we need to encode the information internally. When we need it we will
look it up” (2011: 777).
Ward describes the Internet as a supernormal stimulus, figuratively analogous to junk food,
that operates by “highjacking pre-existing cognitive tendencies and creating novel outcomes”.
Moreover, the Internet “seems to outperform all other external storage devices, potentially
leading people to offload responsibility for the vast majority of information to this single digital
resource” (Ward 2013: 341). Ward claims that an effect of these systems is that we develop a
heightened and over-inflated Cognitive Self Esteem (CSE). That is, we tend to believe that
information that we can rapidly locate through search systems is really our own knowledge,
something we personally know. Moreover, Ward’s experiments find that we tend to view
knowledge that is available to us on the Internet as our own, even when we are offline. For
Ward and Wegner, this is a straightforward cognitive error (Wegner and Ward 2013).
4. Is E-memory really memory?

Might some highly skilled Internet search experts be correct that, in some circumstances, the
Internet is best treated as a part of their cognitive and memory systems? It is worth considering
the possibility that treating knowledge and indeed memories that we can rapidly, effectively,
and reliably access from Google may not be a mistake at all, but rather a metacognitive adjust-
ment to new epistemic conditions.
247
Robert W. Clowes
According to the hypothesis of the extended mind (EM), processes that take place outside of
an agent’s body can still count as part of her mind providing that: (1) the agent can constantly
access the resource; (2) the information supplied by the resource is directly available; (3) infor-
mation retrieved is automatically endorsed; and (4) such information has at some time in the past
been endorsed by the agent. These conditions have since come to be known as trust and glue.1
On the EM view, it is the nature of the agent’s use of, and functional coupling to, an artefact
that determines whether the agent’s mind is extended. Clark and Chalmers’s original (1998)
paper proposed a thought experiment involving Otto and Inga both heading to New York’s
Museum of Modern Art (the MOMA). Inga decides to go to the museum and, thanks to her
normally functioning memory, arrives without difficulty. Otto, who has Alzheimer’s disease,
has difficulty remembering where he is going and how to get there. However, he has a note-
book which stands in for his failing biological memory and which he consults whenever need
arises. Otto uses his notebook to arrive at the MOMA in ways that strongly parallel Inga’s use of
her biological memory. Clark and Chalmers ask us to consider that, in virtue of Otto’s constant
reliance upon, trust and ready use of his notebook, it should count as part of his mind.
There have been many objections (and a few proposed extensions) to the trust-and-glue cri-
teria over the years, including some with special regard to memory (e.g. Rupert 2004). A central
objection hinges on the (re)constructive nature of biological memory. Michaelian (2012a) has
pointed out that the Otto thought experiment is highly unrealistic and uses a container model
of memory, where memories are understood as distinct and compartmentalized items. Empirical
research suggests that human memory traces are nothing like such discrete stable items (Schacter
and Addis 2007). Events that Inga recollects from the past do not access some pristine memory
trace. Rather, her brain reconstructs an event reflecting her current interests and situation.
Recollection is thus better pictured as an interaction between the memory traces carried by
Inga’s brain and her current cognitive goings on, intentions, environmental situation, and emo-
tional states. Human recollection is better envisaged, less like an archivist accessing a pristine
set of images from a video library, and much more like the piecing together of an hypothesis,
based upon the best available evidence. The memory traces recorded in Otto’s notebook seem
not to meet this reconstructive model. They are apparently the same, independent of context.
(Although, insofar as Otto is responsive to environmental contingencies the reasons and context
for accessing the notebook will indeed be contextually driven.) If memory is a container, it is a
leaky container where we lose boxes, mix things up, and construct new contents as the current
situation dictates.
Moreover, Michaelian argues that the trust-and-glue conditions may present an unrealistic
and distorting picture of what memory is. Michaelian concedes that the constancy criterion is not
problematic, as we do indeed have access to our internal biological memory everywhere. But
the other criteria are problematic indeed. Biological memories are not always directly available
because “little information is rendered permanently unavailable once stored, (and) much stored
information is inaccessible to the agent at any given point in time” (2012a: 1156). Biological
memory is much more situational than the EM picture credits.
Endorsement is problematic on several grounds, because, Michaelian contends, storage is not
a consequence of endorsement at all, but of depth of processing which in turn is determined
by several factors including “an assessment of the relevance or significance of incoming infor-
mation” (2012a: 1157). Equally, whether retrieved memory traces are endorsed depends on
metacognitive processes with endorsement being only one possible outcome. Endorsement
of biological memory traces is not automatic. We only selectively endorse recollections based
on a variety of metacognitive strategies. (We shall return to this point shortly.) Given that even
biological memory does not apparently meet the trust-and-glue conditions, the EM hypothesis,
248
Extended memory
Michaelian argues, is a poor guide for deciding whether extended memory is real memory. He
contends therefore that the standard trust-and-glue conditions give us no reason to think of
extended memory as real memory.
Clark and other advocates of EM contend that such objections turn on an overly restrictive
notion of what memory is. EM in fact comports with the received view of memory, which has
it that “memory is any state or process that results from the sequential stages of encoding, stor-
age, and retrieval” (Klein 2015: 1). (We have already seen how Wegner endorsed this view.)
Critics, they believe, focus on overly fine-grained functional properties of memory, whereas the
coarse-grained view is of more value for accommodating the complexity of the real world and
developing research (Clark 2008).
One argument against requiring the fine-grained functional profile of memory is that it
appears to rule out any memory system that works moderately differently from human biologi-
cal memory. Animals, aliens (should we ever meet any), and subjects with cognitive deficits
or non-standard cognitive equipment all seem to be potentially ruled out from having real
memory. This may turn out to be particularly problematic for theoretically grasping the effects
of increasingly exotic forms of apparently extended memory we are currently inventing. On the
fine-grained view, agents post-cognitive enhancement may appear not to have memory systems
at all. Such “external memory systems” – for reasons we have seen – tend to have a functional
profile that complements rather than duplicates biological memory. If, as Donald argues, such
memory extension is the specifically human form of memory, then we are in danger of (ironi-
cally) ruling our specifically human memory from our investigations.
Even if one accepts the importance of a more fine-grained approach to equivalence classes of
human memory, there are a number of ways of responding to this challenge. One possibility is
to concede that extended memory is not real memory, while still claiming that it is nevertheless a
genuine cognitive system. Extended memory systems might only meet the rough-grained func-
tional profile of internal systems, but they should nevertheless count as bona fide cognitive systems
because we use them to do cognitive work. Clark (2010) has argued that the active and constructive
nature of biological memory means that it is difficult to differentiate between memory and reason-
ing systems.2 But if biological memory is so tightly tied into other organic systems that it cannot
easily be disentangled, it might also be so entangled with a host of extended systems. As cognitive
systems become increasingly heterogeneous, we will need to press the coarse-grained view into
service to accommodate human memory as it exists in naturalistic settings. It may become more
difficult to maintain the theoretical purity of bio-memory. This continues to be a rather unresolved
argument, with proponents of the extended mind view arguing the coarse-grained allows for more
open-ended research, whereas its opponents (Adams and Aizawa 2009; Rupert 2009; Weiskopf
2008) argue that it thereby doesn’t do justice to the unities of biological memory.
Another type of response is to weaken the trust-and-glue conditions or otherwise redraft
them to respect bio-memory as science finds it. One way of developing this response is to refor-
mulate the EM view to take account of recent findings on meta-memory and epistemic feelings
(Arango-Muñoz 2014). Work on epistemic feelings suggests that human agents do not simply
endorse any memory trace that comes to mind, but carefully balance a variety of metacogni-
tive abilities to decide whether they should trust their own memories. Metacognitive feelings
include the tip of the tongue sensation, the feeling of error (Mazzoni et al. 2010), and perhaps
most importantly for our discussion, the feeling of truth (Reber and Unkelbach 2010).
As Duncan Pritchard has pointed out, one reason that the original Otto thought experi-
ment seemed convincing was precisely because of what was assumed but not explicitly stated:
that is, Otto’s ongoing abilities to check on his interaction with his notebook. Otto is “gradu-
ally becoming aware that his memory is fading” (Pritchard 2010: 144) and uses his notebook
249
Robert W. Clowes
precisely to try to deal with his failing biological faculties. In addition, Otto is a careful note-
book user and one who is careful to maintain epistemic hygiene in terms of what is entered
into the notebook. If Otto’s beliefs are to be understood as extending to his notebook, this
is in part because of Otto’s ongoing good practices and epistemic virtues. Otto’s abilities to
police and control his own mental states and memories, can be understood as metacognitive,
and Otto’s manipulation of his notebook can be viewed as a sort of metacognitive activity.
Otto’s epistemic feelings and (non-standard) meta-memory more generally can play an ongo-
ing role in deciding just what should be recalled and what stored.
This leads to a dilemma for the EM theorist. As Clark notes: “Ordinary biological memory,
for the most part, functions in a kind of automatic, subterranean way. It is not an object for us,
we do not encounter it perceptually. Rather, it helps constitute the cognitive beings we are”
(Clark 2015b: 3762). At least on the original conception of EM, for a technology to be a good
candidate for the mind it should be used in an unreflective and automatic way, i.e. be transpar-
ent in use. As Clark reflects, the Pritchard interpretation is
adding something Clark and Chalmers rather deliberately left out. They are making
the notebook strategy an object of Otto’s mental focus. The reason we left that out is,
of course, because it works subtly against the extended mind claim itself. For this is not
the role played by ordinary biological memory.
(Clark 2015b: 3762)
And yet in order to meet the sorts of epistemic conditions that Pritchard discusses, some meta-
cognitive agent activity appears to be needed. The worry here is that if we consciously encoun-
ter external resources in order to assess them, they appear less like the generally unconsciously
encountered part of our biological cognitive apparatus.
If the metacognitive picture of memory is correct, then it may just be the case that memory
technologies, and cognitive technologies more generally, do not need to meet the transparency
conditions originally assumed by Clark and Chalmers. Epistemic feelings might provide means
for checking memory sources in a subterranean manner, without them becoming an explicitly
perceptual part of the mental focus rather than a subject of mental focus themselves. But this
implies that we need at least reasonably good metacognitive abilities to continually assess the
contributions of external devices as well as the contributions of our cognitive systems. The con-
sideration of epistemic feelings allows us to note that these do not need be utterly unconscious.
The tip of the tongue sensation is not unconscious.
Arango-Muñoz (2013) argues that our use of epistemic feelings and other metacognitive
abilities might contribute toward our abilities to make virtuous usage of extended technologies.
One interesting aspect of metacognitive abilities is that at least some of them appear to work on
external sources of information as much as internal ones. Thus, epistemic feelings may provide
a way into integrating external resources. What is needed for our epistemic feelings to do the
correct epistemic checking is that the new technologies of E-memory in some sense wear their
epistemic virtues on their sleeves, or otherwise provide cues that our epistemic feelings can track.
Our epistemic capacities will need to keep pace with our technologies.
Clark’s Principal of Ecological Assembly (PEA) suggests that the “canny cognizer” may
simply put together the best set of resources for solving any particular cognitive problem, inde-
pendently of whether they be internal biological systems or external technological ones. There
is a need to show how the PEA is solved practically and how the mind meets the so-called
endorsement and selection problems (Arango-Muñoz 2013; Michaelian 2012b). The endorsement
250
Extended memory
problem refers to how the mind chooses – consciously or unconsciously – whether to endorse
the information coming from the selected resources, whether internal or external. This selection
problem addresses which cognitive resources are used. Given any particular cognitive task, an
agent may solve it in one of a number of ways. Multiplication may be achieved by recollecting
one’s time tables, or by using paper or pencil. A memory trace may be accessed by asking one-
self a question or using Google search. Each way of solving a problem requires a different mix
of internal and external processes, which may or may not be considered cognitive, considering
the particular theoretical resources on offer. Epistemic feelings may play a central role in these
processes, but there is much work to be done to show exactly how this is accomplished.
5. Cognitive enhancement and interpreting the mind

A final line of response to objections to the extended mind is to notice that Otto’s notebook
is not the be all and end all of extended memory. Contemporary and future E-memory tech-
nologies may already have a significantly more biological function profile. Contemporary
technologies, because e.g. of their somewhat reconstructive character, may just be much better
at meeting the original conditions of extended memory than the fictional case of Otto’s note-
book. The Cloud Technology resources we discussed at the beginning of this chapter may be
in several relevant ways more closely resemble the fine-grained functional profile of biological
memory (Clowes 2015).
The human brain is already practically factoring in the presence of E-memory systems
(Sparrow et al. 2011) and is at the centre of current controversies over the cognitive effects of
the Internet. The effects are broadly in line with Clark’s PEA. We have seen how many of us
now attribute knowledge that is stored on the Internet as our own. On Ward’s view, Internet
applications tend to hijack humans’ natural tendencies to distribute memory among groups. The
idea that we thereby possess knowledge that we can merely quickly access is a straightforward
mistake. Moreover, there is danger that as we come to increasingly rely upon such resources,
our onboard biological cognitive capacities may become diminished.
If the EM hypothesis is correct, these self-attributions might just be recognizing the new
epistemic conditions as realities (Clowes 2015). Perhaps Ward’s subjects have just adapted to the
new ecological conditions of Cloud Tech where the cognitive resources of the Internet, highly
tailored to their own needs, are constantly available through their iPhones and tablets. The
nature of what it is practicable to consider their own knowledge has just changed. Temporarily
being disconnected from the Internet does not change what they usually know.
Epistemic feelings, as we have seen, might aid us in incorporating resources that are not
biological parts of ourselves. While we may not be able to guarantee that our epistemic feelings
can indeed give us good cues to the epistemic status of a technology, it is important to consider
how our metacognitive abilities and the epistemic readability of technology might intersect. It
may be possible to build technologies that are designed to signal their epistemic status in ways
that make them more readable to us. Whether they can be trusted, whether the information
they purport to present has been evaluated by a relevant epistemic community, whether it has
been recently edited, are questions in part of design and use.
Achieving extended epistemic virtue requires a good practical grasp upon, and accurate epis-
temic estimation of, the new tech. A problem of our present moment is that we have not developed
good social and intellectual resources, and indeed norms and laws, to make the epistemic status of
Internet resources reliably readable. The current tendency is toward technologies, like Facebook’s
Edgerank algorithm, which while transparent in use, are cognitively opaque (Clowes 2013: 116).
251
Robert W. Clowes
Technologies can structure memory, of course, even if they are not actually parts of our minds.
The view that this external structuring is primarily the best way of understanding the role of
material culture is often developed in terms of scaffolding (Sterelny 2010). This idea was implicit
in Donald’s thoughts on memory, and much earlier work on the cognitive role of tools concen-
trated precisely on the way that tools and the development of material culture helped develop
our peculiarly human cognitive abilities (Gregory 1981; Vygotsky 1978). It may be better to
conceptualize what we “remember” with Google, not as part of anyone’s mind, but as a sort of
cognitive commons upon which we can all draw (Clowes 2015; Dror and Harnad 2008). Systems
like Wikipedia and Google are readily recruited to such models and in many cases this theoreti-
cal model may be more economical than parts of those systems considered as proper parts of us.
Cognitive integration may be best located on a continuum (Heersmink 2012; Sutton et al.
2010), between deeply integrated systems which can be viewed as proper parts of an agent, and
those that should be considered as a kind of cognitive scaffolding or part of a cognitive com-
mons. Although there is a spectrum here, it is worth remembering that locating things on one
side or other of the divide has real consequences. Where a technology can be considered to
form a part of an agent’s identity, its removal or the prohibition of its use might seem like an
attack on that agent (Heersmink 2015). When such interpretations seem reasonable, we are on
the EM end of the spectrum.
If we view E-memory as neither real memory, nor cognitive, then the selection and endorse-
ment problems seem to arise in just the same way, but with new added ethical force. Any
practical choice between the biological and extended systems, now also requires choosing to
select between ‘genuine’ and ‘pseudo’ cognitive resources. Good epistemic hygiene may help
guide us in not over-relying on machines to carry out our memory functions where it might
diminish us in the process.
Against this, hybrid-agents may make more sense according to our folk psychological norms
when interpreted in an extended light. The canons of folk psychology may better predict and
explain an agent’s activities when her extended belief set is incorporated in the interpretation
base.3 Otto is more easily predictable and explicable when we postulate the contents of his
notebook are really beliefs. Explaining his museum-finding abilities become more complex if
we say that Otto knows very little but consults his notebook. It is easier and more economical,
and more in tune with the canons of folk psychology, just to say that Otto knows how to arrive
at the MOMA.
Where E-memory is deeply integrated as part of an agent’s sense of self and even identity,
our attempts to interpret, predict, and interact with such apparently hybrid agents might just
require us to practically consider an agent’s extended resources as part of the basis of our folk-
psychological ascriptions. Interpreting E-memory as real memory, then, seems to help in much
predictive and explanatory work when making sense of such agents. On these grounds we
may need a revised version of the EM – taking account of epistemic feelings – to produce the
best overall account of the use and integration of E-memory. E-memory may not appear to be
memory from certain fine-grained vantage points, but it can still be really cognitive and really
constitutive of the agent’s identity. But if this is so, we must take seriously the idea that the
human mind is currently undergoing profound, perhaps unprecedented, changes as we incorpo-
rate a diverse set of digital media into our cognitive arsenal (Clowes 2012). This may itself imply
limitations upon, and the need for revisions to, our current folk psychology. We need to spend
more intellectual energy to understand the shape and the ethical and epistemic implications of
these new kinds of minds.
252
Extended memory
Notes
1 It is notable that at least some transactive memory systems readily meet the trust-and-glue conditions.
2 Also see Clark’s latest book (2015a) on how memory and other cognitive functions such as imagination,
may, at a neural level be closely bound together.
3 Although see problems around the lack of integration of beliefs suggested in Weiskopf (2008).
Related topics
• Habit memory
• Memory traces
Further reading
Clowes, R. W. (2015). Thinking in the cloud: The cognitive incorporation of cloud-based technology.
Philosophy and Technology 28(2): 261–96.
Michaelian, K. and Sutton, J. (2013). Distributed cognition and memory research: History and current
directions. Review of Philosophy and Psychology 4(1): 1–24.
Malafouris, L. (2013). How Things Shape the Mind: A Theory of Material Engagement. Cambridge, MA: MIT
Press.
References
Adams, F. and Aizawa, K. (2009). Why the mind is still in the head. In P. Robbins and M. Aydede (eds.),
The Cambridge Handbook of Situated Cognition (pp. 78–95). Cambridge: Cambridge University Press.
Arango-Muñoz, S. (2013). Scaffolded memory and metacognitive feelings. Review of Philosophy and
Psychology 4(1): 135–52.
Arango-Muñoz, S. (2014). The nature of epistemic feelings. Philosophical Psychology 27(2): 193–211. doi:1
10.1080/09515089.09512012.09732002.
Bell, C. and Gemmell, J. (2009). Total Recall: How the E-Memory Revolution Will Change Everything. New
York: Dutton.
Clark, A. (1997). Being There: Putting Brain, Body, and World Together Again. Cambridge, MA: MIT Press.
Clark, A. (2008). Supersizing the Mind. Oxford: Oxford University Press.
Clark, A. (2010). Memento’s revenge: Objections and replies to the extended mind. In R. Menary (ed.),
Extended Mind. Cambridge, MA: MIT Press.
Clark, A. (2015a). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. New York: Oxford
University Press.
Clark, A. (2015b). What ‘Extended Me’ knows. Synthese April: 1–19. doi:10.1007/s11229-015-0719-z.
Clark, A. and Chalmers, D. (1998). The extended mind. Analysis 58: 10–23.
Clowes, R. W. (2012). Hybrid memory, cognitive technology and self. In Y. Erdin and M. Bishop (eds.),
Proceedings of AISB/IACAP World Congress 2012.
Clowes, R. W. (2013). The cognitive integration of E-memory. Review of Philosophy and Psychology 4: 107–33.
Clowes, R. W. (2015). Thinking in the cloud: The cognitive incorporation of cloud-based technology.
Philosophy and Technology 28(2): 261–6.
Donald, M. (1991). Origins of the Modern Mind. Cambridge, MA: Harvard University Press.
Dror, I. E. and Harnad, S. (2008). Offloading cognition onto cognitive technology. In I. E. Dror and
S. Harnad (eds.), Cognition Distributed: How Cognitive Technology Extends Our Minds (pp. 1–23).
Amsterdam: John Benjamins Publishing.
Gregory, R. L. (1981). Mind in Science: A History of Explanations in Psychology. Cambridge: Cambridge
University Press.
Harris, C. B., Keil, P. G., Sutton, J. and Barnier, A. J. (2010). Collaborative remembering: When can
remembering with others be beneficial? www.cogsci.mq.edu.au/news/conferences/2009/ASCS2009/
pdfs/Harris.pdf.
253
Robert W. Clowes
Harris, C. B., Keil, P. G., Sutton, J., Barnier, A. J. and McIlwain, D. J. (2011). We remember, we forget:
Collaborative remembering in older couples. Discourse Processes 48(4): 267–303.
Heersmink, R. (2012). Mind and artifact: A multidimensional matrix for exploring cognition–artifact rela-
tions. Proceedings of AISB/IACAP World Congress 2012.
Heersmink, R. (2015). Extended mind and cognitive enhancement: Moral aspects of cognitive artifacts.
Phenomenology and the Cognitive Sciences, 1–16.
Hodges, S., Williams, L., Berry, E., Izadi, S., Srinivasan, J., Butler, A. . . . Wood, K. (2006). SenseCam:
A retrospective memory aid. UbiComp 2006: Ubiquitous Computing 4206: 177–93.
Hutchins, E. (1995). Cognition in the Wild. Cambridge, MA: MIT Press.
Johansson, N., Andersson, J. and Rönnberg, J. (2005). Compensating strategies in collaborative remembering
in very old couples. Scandinavian Journal of Psychology 46(4): 349–59.
Klein, S. B. (2015). What memory is. WIREs Cogn. Sci. 6: 1–38.
Malafouris, L. (2013). How Things Shape the Mind. Cambridge, MA: MIT Press.
Mazzoni, G., Scoboria, A. and Harvey, L. (2010). Nonbelieved memories. Psychological Science 21(9): 1334–40.
Michaelian, K. (2012a). Is external memory memory? Biological memory and extended mind. Consciousness
and Cognition 21(3): 1154–65.
Michaelian, K. (2012b). Metacognition and endorsement. Mind and Language 27(3): 284–307.
Michaelian, K. and Sutton, J. (2013). Distributed cognition and memory research: History and current
directions. Review of Philosophy and Psychology 4(1): 1–24.
Pritchard, D. (2010). Cognitive ability and the extended cognition thesis. Synthese 175(1): 133–51.
Reber, R. and Unkelbach, C. (2010). The epistemic status of processing fluency as source for judgments
of truth. Review of Philosophy and Psychology 1(4): 563–81.
Rupert, R. D. (2004). Challenges to the hypothesis of extended cognition. Journal of Philosophy 101:
389–428.
Rupert, R. D. (2009). Cognitive Systems and the Extended Mind. New York: Oxford University Press.
Schacter, D. L. and Addis, D. R. (2007). The cognitive neuroscience of constructive memory:
Remembering the past and imagining the future. Philosophical Transactions of the Royal Society of London
B: Biological Sciences 362(1481): 773–86.
Sellen, A. J. and Whittaker, S. (2010). Beyond total capture: A constructive critique of lifelogging.
Communications of the ACM 53(5): 70–77.
Smart, P. R., Heersmink, R. and Clowes, R. W. (2017). The cognitive ecology of the Internet. In
S. J. Cowley and F. Vallée-Tourangeau (eds.), Cognition Beyond the Brain, 2nd edition (pp. 251–82).
New York: Springer.
Sparrow, B., Liu, J. and Wegner, D. M. (2011). Google effects on memory: Cognitive consequences of
having information at our fingertips. Science 333(6043): 776–8.
Sterelny, K. (2010). Minds: Extended or scaffolded? Phenomenology and the Cognitive Sciences 9(4): 465–81.
Sutton, J. (2010). Exograms and interdisciplinarity: History, the extended mind, and the civilizing process.
In R. Menary (ed.), The Extended Mind (pp. 189–225). London: Bradford Book, MIT Press.
Sutton, J., Harris, C. B., Keil, P. G. and Barnier, A. J. (2010). The psychology of memory, extended
cognition, and socially distributed remembering. Phenomenology and the Cognitive Sciences 9(4): 521–60.
Vygotsky, L. S. (1978). Mind in Society: The Development of Higher Psychological Processes. Cambridge, MA:
Harvard University Press.
Ward, A. F. (2013). Supernormal: How the Internet is changing our memories and our minds. Psychological
Inquiry 24(4): 341–8.
Wegner, D. M. (1987). Transactive memory: A contemporary analysis of the group mind. In B. Mullen
and G. R. Goethals (eds.), Theories of Group Behavior (pp. 185–208). New York: Springer-Verlag.
Wegner, D. M. and Ward, A. F. (2013). How Google is changing your brain. Scientific American 309(6):
58–61.
Weiskopf, D. A. (2008). Patrolling the mind’s boundaries. Erkenntnis 68(2): 265–76.
254
20
COLLECTIVE MEMORY
Over the past decades, the topic of collective memory has become an increasingly central
theoretical concern in a wide spectrum of academic disciplines, from the human and social
sciences to literature, aesthetics, and the cognitive and natural sciences. It is perhaps due to the
breadth of its fields of application that it has often seemed to be a nebulous concept that is in
need of philosophical clarification.
The term “collective memory” and the concept to which it corresponds are of recent vin-
tage and have gained currency since the pioneering work of authors such as Walter Benjamin
(Benjamin 1969a: 83–110; 1969b: 155–200; 1969c: 217–52) and above all Maurice Halbwachs
(Halbwachs 1958, 1994, 1997). The period following World War II, and especially since the
1970s and 1980s, has witnessed intense preoccupation with this phenomenon, which many
recent authors have characterized as a “memory boom” (Olick 1999: 333–48; Huyssen 2000:
21–38; Klein 2000: 127–50; Kansteiner 2002: 179–97; Nora 2002: 1–8; Winter 2006). This
recent emergence of the concept of collective memory as a central theoretical concern on a
global scale is paradoxical since, as a social function, collective remembrance has been embodied
in age-old practices that are as ancient as human communal existence and the symbolic realms
that forge its cohesion. The intense contemporary preoccupation with collective memory sug-
gests that this phenomenon which was not previously of central theoretical interest has been
drawn into the arena of theoretical discourse in relation to ideas, attitudes, and assumptions that
are specific to our times. In this chapter, I begin my elucidation of the concept of collective
memory by characterizing its historical genealogy and the social function that set the backdrop
for its new theoretical visibility. This will lead in the following sections to a critical discussion of
recent attempts to redefine it in social, psychological, and historical analysis,1 a theoretical inves-
tigation of its scope and temporal articulations in the public sphere and, finally, an interpretation
of its public configuration in the era of the mass media.
1. The emergence of collective memory as a topic of

modern discourse and theory
The rise of theoretical preoccupation with collective memory coincided with the decline of
more traditional ways of accounting for collective cohesion in the socio-political sphere. Its
new theoretical visibility corresponded to a series of long-term developments that emerged
255
over the course of the modern period. These involved a weakening of the traditional convic-
tion inherited from ancient philosophy and Christian theology that immutable metaphysical
ideas might definitively account for human identity and socio-political existence. Whereas
in the relatively stable framework of pre-modern societies, the continuity of collective exist-
ence was postulated in terms of fixed metaphysical principles, modern currents of thought,
from Lockean empiricism to the French Enlightenment and Kantian critical idealism, in
spite of other points of divergence, each sharply undermined traditional presuppositions
concerning the intelligibility of such principles as the basis for understanding self and world.
This questioning of ideas of human essence and social cohesion in terms of pre-given meta-
physical schemata admitted a more open view of the self-determination of human groups
in the course of their historical development. Eighteenth-century thought at the same time
witnessed the elaboration of a series of new perspectives concerning the singular and vari-
able “spirit” underlying human collectivities conceived as individual unities. Montesquieu’s
“spirit of the laws,” Voltaire’s “spirit of an epoch,” or Herder’s “spirit of a people” are
notable examples of novel conceptions of collective unities that define themselves over time
that were elaborated in the decades prior to the French Revolution. The subsequent expe-
rience of radical change and dislocation brought about by the French Revolution and the
demise of the traditional European socio-political order inspired widespread reflection on
the phenomenon of historical discontinuity and set the phenomenon of collective cohesion
in a problematic light.
It is in this context that “memory,” conceived not only as a faculty of human understanding,
but as the organ of continuity and cohesion in the collective sphere, began to play a paradig-
matic role in social theory. In the aftermath of the French Revolutionary period Hegel, in his
Phenomenology of the Spirit (1807), accorded an epoch-making status to memory (Erinnerung) as
a source of collective identity, for it is memory that provides the dynamic principle of cohe-
sion of the Spirit amid the diversity of its expressions in human historical development (Hegel
1970: 16–24, 590–91). In view of the fundamental differences that distinguish human groups
and historical epochs, Hegel conferred on the self-understanding of the Spirit as it is maintained
through recollection of past experience the capacity to lend cohesion and continuity to the
diverse moments of its development over the course of history.
Without dwelling on the question of Hegel’s influence, his interpretation of the role of
memory in the collective sphere marked an epoch-making shift for subsequent generations,
even after the absolute claim of his philosophy of history had lost its persuasive power. In a very
different and more directly political perspective, Ernst Renan, an avowed admirer of Hegel’s
philosophy of history (Renan 1905: 172–3), identified memory shared by national groups—as
opposed to geography, language, or race—as the principle source of their cohesion over time
(Renan 1992: 54). In a later and different context, the historicism of Benedetto Croce was
inspired by the idea that memory among national groups is the spiritual organ of historical con-
tinuity (Croce 1929), just as his German contemporary, Wilhelm Dilthey, identified a people’s
memory (Erinnerung) of its past as the source of its cohesion (Zusammenhang) over time, lending
continuity to the diverse moments of its development (Dilthey 1973).
Beside this conception of memory as a spiritual source of historical continuity, the decades
prior to World War I witnessed the emergence of a series of highly influential conceptions of
“organic memory” that was taken to infuse the physiological characteristics inherited by large
groups. Ewald Hering’s speculations about a physiologically determined organic memory
found an echo in the theories of Wilhelm Roux and Ernst Haeckel and were popularized in
England by Samuel Butler (Hering 1921; Butler 1910). All of them espoused the conviction
that a physiologically inscribed and unconsciously transmitted memory provides the principle
256
Collective memory
of cohesion of human collectivities over time. In his mature writings, Friedrich Nietzsche
adopted an analogous conviction. In works like Beyond Good and Evil or The Genealogy of
Morals, Nietzsche assumed that organic memory is transmitted among generations and pro-
vides continuity to long-term historical development.2 The epoch-making shift that Hegel
inaugurated in situating group memory at the source of historical continuity and develop-
ment, albeit on a wholly different organic basis, came to a novel form of expression at the
end of the century.
2. Maurice Halbwachs’s concept of collective memory

and its recent critics
In the context of the late nineteenth and early twentieth centuries, rapid industrialization,
transformations in the conditions of travel and communication, and mass migration of rural
populations to urban centers were factors that contributed to an unprecedented sense of change
and dislocation which reached its paroxysmal climax during the cataclysm of World War I.
The radical historicity and contingency of the socio-political order fueled a widespread sense
of “crisis,” in which earlier assumptions emanating from the philosophy of history concern-
ing the unity of past and present in a meaningful process of development came into question.
At this precise juncture, theoretical attention began to turn to the phenomenon of “collective
memory” which, in the writings of Walter Benjamin and Maurice Halbwachs, placed the theme
of social cohesion in a novel perspective. In the wake of the contemporary experience of dislo-
cation, they broke with earlier assumptions concerning the role of memory in forging historical
cohesion and continuity.
Halbwachs’s manner of situating discontinuity at the heart of his reflection comes to light in
his conception of recollection shared among overlapping living generations, which he sharply
distinguished from awareness of the historical past. In delineating this concept of collective
memory, Halbwachs recognized that memory at all of its levels is anchored in personal aware-
ness that is most immediately interwoven with the living groups to which it belongs in specific
social frameworks (cadres sociaux). Beyond the shared memories of groups whose members are
personally familiar with each other, Halbwachs readily acknowledged that mass associations such
as vast nations provide few occasions for direct face-to-face contact. Their shared remembrance
is almost always indirectly acquired from the works and deeds of others; it is in Halbwachs’s
words a “borrowed memory” (Halbwachs 1997: 128–9). For this reason, Halbwachs’s concept
of collective memory attributed a particular importance to smaller intermediary groups, both
as a source of shared experience and of the transformations which directly orient the lives of
their members. Each individual is in this sense “submerged” (plongé), simultaneously or succes-
sively, in many different groups (Halbwachs 1997: 129). Each individual, through participation
in different groups comprising different spheres of experience and of memory, establishes lines
of continuity between past and present which, by virtue of the unity of temporal context, lends
a measure of stability to individual identities anchored in the framework of each of the groups
in which she participates. The fundamental discontinuity in temporal horizon intervenes, as
Halbwachs wrote, where living generations disappear, and with them all direct contact with
ways of being, gestures, idioms, and attitudes belonging to the living context of still-existing
generations. For the most part, these discontinuities emerge gradually and the shifts in the tem-
poral horizon they occasion remain largely unperceived. Yet they mark a radical distance of the
living present from the remote past (Halbwachs 1958: 22; 1997: 130). Once it is obliterated in
the memory of living generations, only the work of historical deciphering can indirectly recon-
struct, in piecemeal fashion, the lost theater of the past.
257
In recent decades, the concept of collective memory that Halbwachs elaborated has elicited
important criticism. In view of elucidating the concept of collective memory, I will briefly con-
sider two contemporary criticisms that have been raised against his theory.
Halbwachs has frequently been taken to task for his tendency to consider individual aware-
ness as a function of the different social groups to which it belongs and thus to limit it to its
socially determined expressions. Paul Ricoeur, above all, has emphasized what he has taken to
be the proximity of Halbwachs’s theory to a form of collective determinism that leaves little
space for individual spontaneity and autonomy (Ricoeur 2004: 120–24). Certainly this tendency
can be inferred from a number of Halbwachs’s statements but, as Ricoeur himself acknowl-
edges, Halbwachs would seem to attenuate this position at other points in his analysis. Jeffrey
Olick has stressed the tension in Halbwachs’s thought between a collective memory that exists
only in the minds of remembering individuals and the collectively shared languages and sym-
bolic representations in which individual memories are anchored (Olick 1999: 335–6). Since, as
Olick has noted, a measure of personal autonomy is not incompatible with the idea of collective
frameworks of recollection that lend stability to personal identity, a determinist sociology is by
no means a necessary consequence of Halbwachs’s theory.
A second objection to Halbwachs’s concept of collective memory was first voiced by Peter
Burke and has been amplified by Aleida and Jan Assmann. Burke’s critique placed in question
the distinction that Halbwachs drew between collective memory and historical understand-
ing. Like remembered events, the historical record also belongs to group memory. As with
the collectively remembered past, historical narratives reconstruct the past from a given group
perspective. In order to highlight this reconstructive aspect of all kinds of approach to the past,
Peter Burke proposed to fuse together memory and historical understanding under the single
rubric of “social memory” (Burke 1989: 98–9).
In explicit reference to Burke’s theory, Jan Assmann adopted a similar point of view, and he
proposed to include all forms of comprehension of the past, both group memory and long-term
historical awareness, under the heading of “collective memory.” He at the same time further
subdivided this category by introducing a distinction between what he and Aleida Assmann
term “communicative memory,” corresponding roughly to the collective memory of living
generations in the sense of Halbwachs, and “cultural memory” encompassing the entire herit-
age of a literary, legendary, and historical past (Assmann and Assmann 1988: 28–9; Assmann,
2011: 34–41). Jan Assmann further broadens his definition of “cultural memory” where, after
the manner of Freudian theory, it encompasses unconscious and repressed aspects of past group
experience that former concepts of cultural tradition tended to exclude (Assmann 2000: 38–44).
In integrating the domain of psychological latency and in according a central place to leg-
end and myth in their definition of cultural memory, the revisions proposed by Jan and Aleida
Assmann are helpful in reminding us that myths, legends, and repressed collective fantasies
often have greater vivacity and longevity than the remembrance of factual events, and also
that mythical beliefs lend orientation to historical actions. Once this point is acknowledged,
however, the question remains whether the conflation of all temporal dimensions of the past
under the rubric of “collective memory”—encompassing “communicative” and “cultural”
memory in the Assmanns’s sense—is not too general to be useful as an analytical tool. Cultural
memory is, of course, a form of group remembrance, but it is so in a secondary and, indeed,
metaphorical sense, which must be distinguished from collective memory in its original forms.
Its generality, which overrides Halbwachs’s finer categories of analysis of the essential discon-
tinuity of collective experience over time, raises the question concerning the relevance for a
theory of collective memory of the distinction Halbwachs initially drew between the memory
retained by overlapping living generations and the historical past beyond its pale. I will return
258
Collective memory
to this question in Section 5 of this chapter in light of the possibility that, in spite of disconti-
nuities in group remembrance, latent aspects of the historical past may continue to function in
the present beyond the scope of collective awareness.
3. Collective memory as a category of contemporary social analysis

According to my interpretation, it is above all necessary to envision, in relation to Halbwachs’s
theory, a reformulation of the level of analysis of the concept of collective memory. Where
Halbwachs, as we have seen, elaborated his concept of collective memory primarily in relation
to the interwoven social frameworks of smaller groups, such as families and other associations,
I propose to shift the focus of investigation to the broad domain of mass social life and of the
corresponding public sphere. This shift is based on the consideration that the communication of
what is collectively experienced and remembered, far from specific to smaller social frameworks,
is conveyed through language and other symbolic forms that presuppose publicly intelligible
patterns of social interaction. The cohesion of small groups, as private as they may be, depends
upon symbols which are not specific to them but derive from a broader mass social framework.
For this reason, an adequate concept of collective memory must encompass at its fundamental
level mass social existence in a public sphere situated in an increasingly globalized world.
This shift in focus nevertheless raises what might first appear to be an insurmountable dif-
ficulty. As Halbwachs emphasized, memory in its original sense always transpires in the personal
sphere of individual rememberers that is readily shared with members of families and small
groups. On the level of publicly significant events, however, direct personal experience and
remembrance are for the most part excluded, for they are usually possible only for a tiny minor-
ity of eyewitnesses. Remembrance of events endowed with public significance is almost always
based on indirect reports or accounts diffused among the vast strata of contemporary mass socie-
ties. In view of the gap between personal remembrance or that retained by small groups and the
indirect quality of diffuse representations of vast collectivities, we might wonder whether the
concept itself of “collective remembrance” might be appropriate. It might indeed be claimed
that collective memory, since it rarely corresponds to any direct and original form of remem-
bered experience, is essentially a figment of the social imagination.
The resolution of this problem depends upon our understanding of the role of imagination
in the social sphere. Certainly, imagination is a complex faculty which corresponds to a series of
operations, reaching from the production of fantasies to the deliberative activity involved in a
whole range of cognitive capacities, from mathematics to historical reconstruction of past factual
events. Moreover, where we commonly take imagination and memory to be separate opera-
tions, they are always interrelated in their everyday employment and primarily work as a unity
in conjunction with a totality of anthropological capacities.
In the collective sphere, imagination plays a role that is of particular importance for a theory of
collective memory, for it enables the embodiment of meaning in concrete, communicable images
and symbols. For this reason, in accord with the privilege I accord to the public realm of con-
temporaneous social existence, a closer examination must be undertaken of symbols and symbolic
interaction through which collectively significant experience and recollection are communicated.
4. The symbolic embodiment of collective memory

and its temporal articulations
Symbols are, indeed, many-faceted phenomena that all communication of collective memory
presupposes. They are conventionally understood in two different ways, both of which are per-
tinent to our analysis (Barash 2012: 183–95; 2016). At one level, symbols may be defined in a
259
narrow sense where they represent what is absent or what cannot be grasped in the realm of
sense-experience: for example, the lamb representing the divinity of Christ, or the flag standing
for a nation. At another and more fundamental level, symbols also have a broader function, for
they confer spatio-temporal pattern and logical order on immediate experience itself.3 In this
broad sense, symbols lend intelligibility to experience as it is communicated through language,
gesture, or style and is embodied in memory. As such, they lend spontaneous intelligibility
to the public world in which more particular forms of communication among small groups
and individuals are deployed. In an urban environment, for example, I immediately familiarize
myself with spatial differences between private yards and public parks or semi-public shopping
malls, even before I explicitly reflect on them, just as the background music I hear in an airport
or supermarket, a restaurant or church gives me direct clues concerning the surrounding social
milieu. Collective memory is rooted in a many-layered web of interwoven shared symbolic
structures that orient spatio-temporal awareness and conceptual logic. The continuity of this web
attests the ongoing link between past and present within the horizon of experience recalled by
overlapping living generations.
As they are marshaled at all levels of the collective life-world that draw on a publicly com-
municable sense, symbolic configurations of group experience and remembrance are by no
means static elements, for they enter into complex interaction over time. In time, collective
remembrance brings to group awareness a tie between past and present that marks lines of group
continuity. If it is also in time that mutation and discontinuity with the past occur and are later
brought to mind, group perception of discontinuity nonetheless presupposes an underlying
continuity or sameness in the remembering collectivity in relation to which discontinuity may
be placed in relief. In close association with the imaginative work of symbolic embodiment of
experience, the specific province of collective memory lies in the articulation of the modes of
synthesis through which the web of remembered experiences shared by groups is interwoven in
a given present and recalled as common experience.
At all of its levels, the symbolically configured forms of social cohesion woven by collective
memory conform to a variety of temporal patterns. In the space of this brief analysis, I empha-
size the role of three such temporal articulations, which are each essential preconditions of
social life. I equate the first of these articulations with “reiterative” practices that are patterned
in collective memory in line with continually recurring cycles of social activity. These are the
rhythms of everyday life commanded by specific habits, which vary among different sectors of
the population in accord with particular kinds of activity. The division between hours of work,
of recreation and sleep, the regular intervals at which, in given social contexts, market days and
other similar social institutions are enacted, exemplify the reiterative rhythms that punctuate
social life. Commemorative practices exemplify the second kind of temporal articulation of col-
lective memory. Like reiterative temporal rhythms, commemorative practices recur at regular
intervals, yet their periodicity corresponds not to habitual everyday activities, but to unique past
events that are generally recalled on a yearly basis (or of groups of years marked off in decades
and centuries) to designate symbolically meaningful occurrences and to reaffirm collectively
held ideals through which groups renew their cohesion. At times when reiterative or com-
memorative acts are not performed, they are not generally themes of explicit reflection and fade
into the horizon of latency until, at regular intervals, their symbolic charge is summoned anew
in view of solidifying group cohesion and orienting group identities in a current context.
The third temporal articulation concerns the ongoing subsistence of group dispositions that
customary practices mold into long-term propensities that span generations. Where reiterative
and commemorative rhythms that punctuate collective life are, in their multifarious expressions,
straightforward and readily recognizable, long-term group dispositions are more difficult to
260
Collective memory
identify, since they span all reaches of the public sphere and, for the most part, spread over the
implicit layers of group life. Here we encounter the deeper strata of collective existence, which,
by virtue of the symbols in which they are couched, trace the links between past and present
elaborated in successive moments of contemporaneity, lending continuity to group experience
and collective self-conceptions as they are mobilized in view of what is to come. Rooted in
an underlying network of passive dispositions that encompass the different sectors of social life,
these long-term customary patterns constitute the deeper and usually implicit horizons of tem-
poral continuity residing in collective memory. Beneath the surface strata of group existence,
these diffuse levels of group existence, from the latent reaches in which they perdure, account
for the anonymous threads of social cohesion that are woven amid changes, shifts, and upheav-
als. In this perspective, collective memory does not only concern commemorative ceremonies,
vestiges of the past displayed in museums, or even the ongoing public recognition of a past that
has been experienced and transmitted by contemporaries but, first and foremost, vast reservoirs
of meaning that are latent sources of interpretation underlying public existence and configuring
group identities. If they are often unperceived, they are not, however, invisible occult qualities,
nor are they reducible to general laws governing unconscious psychic mechanisms. Specific to
the particular mode of existence of a given group, these passive symbolic configurations come
to expression in group behavior and social styles that often correspond to corporeal dispositions
and habits. Over the course of the twentieth century, sociological research has emphasized the
role of the body and of corporeal dispositions as the locus of socially mediated symbolic expres-
sions. In more recent years, the theme of the body as a vehicle of such long-term articulations of
collective memory through gestures, postures, and styles has provided a seminal topic of anthro-
pological analysis (Mauss 1983: 368; Bourdieu 1977: 82–3; Connerton 1989; Fassin 2007).
The continuities interwoven by this deep level of passively inscribed dispositions, as they
draw on the reservoirs of sedimented group experience that have been symbolically elaborated
over the long term, are not to be confused with secondary elaborations, such as codified tradi-
tions or cultural legacies which, indeed, presuppose this fundamental experiential dimension.
In view of their mostly latent character and their diffuseness, these long-term collective con-
tinuities are not so much the direct objects of historical representation as patterns of regularity
accounting for the social cohesion of experience that historical representation presupposes. Far
from monolithic structures, the weave of symbolic significations flowing from these passive
depths is in a constant state of fragmentation as it is transmitted and interpreted among a plurality
of groups that make up a given social context (Mendels 2004).
5. The hiatus between collective memory and

the historical past
The interpretation of collective memory elaborated thus far permits us to set on a new foun-
dation insight into the discontinuity between group remembrance and the historical past that
originally inspired Halbwachs’s work. If collective remembrance of the past is always fragmented
in accord with a plurality of different group perspectives, public communication depends upon
a shared web of spontaneously graspable symbols that defines the contours of group contempo-
raneity shared by overlapping, living generations, and distinguishes them from the historical past
beyond living memory. This discontinuity is not simply a consequence of the demise of single
individuals and groups, for the disappearance of living generations signals the evanescence of
the concrete context in which their symbolic interaction transpired. Following the disappear-
ance of this theater of group interaction, the legibility of the symbolic structures embedded in it
begins to weaken. Even where the broad intelligibility of general linguistic and other symbolic
261
categories is retained over centuries, the more specific nuances groups invest in them, constitut-
ing the living context and intrinsic sense of their coexistence, are subject to remarkable, if often
barely palpable variability, as collective memory recedes into the historical past. In a situation
of radical discontinuity, the passage of each successive generation marks a drift in the symbolic
framework of communication and interaction. Such an abrupt change in context, calling forth
mostly imperceptible displacements of its passive recesses, casts in its wake a deepening shroud
over the essential significance of the symbolic patterns that constitute the past’s singular texture.
Discontinuity in the shared context of remembrance marks the finitude of living groups.
Beyond the finite existence of mortal individuals, group awareness is subject to a specific kind
of finitude set in relief by the limited reach of group remembrance beyond the horizon of recall
of living generations. The finite temporal scope of this context comes to light in the collective
inability, by virtue of any specific capacity of memory, to penetrate the remote depths of the his-
torical past. This can only be achieved to a very limited extent through historical reconstruction
in which memory in its proper sense plays only an indirect role.
Beyond historical continuities of the spirit presumed to be capable of comprehending
and retaining the essential significance of past experience, and beyond continuities taken to
be unconsciously elaborated by organically transmitted dispositions, this concept of collective
memory highlights an incommensurable diversity of temporal contexts that underlie the histo-
ricity of human existence. As the nuances of contextual divergence fade from living memory,
subtle metamorphoses in the symbolic underpinnings of group experience flowing from its
passive reaches are no longer recalled to group awareness. The more fully the contours of con-
textual divergence between past and present are forgotten, the more readily nuances are effaced
that distinguish collective memory from the historical past.
If the fundamental distinction that, in company with Halbwachs, I have drawn between
collective memory and history has an essentially theoretical aim, it also has a practical scope.
Its practical intention is inspired by insight into the elusiveness and, indeed, the opacity of
those strata of group existence that reach into the passive layers of symbolic configuration.
This consideration places in a dubious light facile presuppositions concerning long-term group
continuity, allegedly forged by memory, that have become a contemporary rallying point for
extremist political ideologies. From the works of Maurice Barrès, the late nineteenth-century
French novelist and nationalist prophet, up to the pronouncements of more recent proponents
of nationalist extremism, the discourse of collective memory has been marshaled to attest an
alleged homogeneity of national groups whose shared ethnic virtues are taken to constitute the
durable matrix of national unity (Barash 2016). Against this attitude, the theory of collective
memory, in highlighting the discontinuity between generations and the finite reach of group
remembrance, reveals its eminently critical vocation.
6. The public configuration of collective memory in

the era of the mass media
In our contemporary world, the emergence of mass societies in a context of global interaction
marks an essential break with the past that distinguishes the current horizon of group inter-
action from previous forms of social existence. Corresponding to this development, the public
sphere has undergone essential metamorphoses over the past century and a half that have been
channeled by the technical evolution of the mass media. Their increasing predominance as
organs of public information has tended to accentuate the disparity between the life-world
of original personal and small-group experience and remembrance, and the vast public realm
that the mass media configure. The development of the mass media since the introduction of
262
Collective memory
mass-circulation illustrated newspapers and magazines that exercised a preponderant influence in

the late nineteenth and early twentieth centuries began to transform the public sphere through
their ways of configuring and transmitting information and thus of conferring public significance
on reported actions and events. This transformation was further accentuated through mass-
produced images favoring communication that, independently of the articulated language they
accompany, spontaneously convey a visible message that can be immediately grasped beyond the
confines of linguistic or cultural borders. From the development of the still photo and the mov-
ing picture newsreel to that of television, animated digital imagery and the World Wide Web,
the mass media, whatever their distance from the life-world, have been able to simulate, in an
ever more technically precise manner, direct experience of personal and small-group encounters
and to make this simulated experience resonate in public memory.
The radical transformation that mass media communications have brought to the public
world may be placed clearly in relief if we consider that the potency of the mass media is
not the result of a simple reproduction or replication of what transpires in the everyday life-
world of face-to-face encounters. In an ever more technologically precise manner television, in
CNN fashion, presenting “hypermediated” combinations of written text and visual image and,
subsequently, hypermediated World Wide Web and social media Internet displays, transfigure
reported information in function of a spatio-temporal pattern and logical order specific to the
mass media format. This format configures information in terms of a symbolic order that is spe-
cific to mass communications.
From the introduction of mass circulation newspaper journalism and then radio, television
and the Internet, news reporting has provided the paradigmatic format for mass communica-
tions (Luhmann 2004: 120, 141). It is in accord with this format that phenomena are endowed
with public visibility and are consigned to public remembrance. Let us briefly characterize the
salient characteristics of this format in order to identify its implications for public modes of
experience and remembrance. First, images and accompanying information are condensed and
tailored to the space of a small screen and subjected to the split-second timing required by the
rapidly moving sequence of reports. Second, information is adapted to the generally undiffer-
entiated interests of the largest possible number of spectators. This characterizes what has often
been qualified as its anonymity (Luhmann 2004: 11). If, to a certain extent, the new interactive
forms of social media are designed to introduce a measure of personal interrelation, encounters
are nonetheless mediated by an interface through which face-to-face presence is simulated and
formatted for potential transmission to larger groups or to a mass audience. Third, informa-
tion is selected in function of its recency in the field of current events, of its status as “breaking
news,” that is subjected to a constant turnover in which older items are replaced by more recent
ones. The currency of information and the rapidity of its turnover place it in a decontextualized
perspective in which all but the most recent past is excluded from consideration. Fourth, in the
manner of mass circulation newspapers and television news, different items of information are
typically subjected to haphazard arrangements, setting commercial messages, sports, weather,
and breaking news alongside or following one another in a seemingly random order. This
mass communication format prevails on hypermediated CNN-like screens and Internet displays
where, analogous to news reporting, information in different sectors is transmitted. In sectors
of entertainment, advertisement, and sports, the attention of a mass public may thus be drawn
to the “latest” sports results, to ever-renewed entertainment programming, and to “new and
improved” commercial products.
The uncanny capacity of the mass media to simulate in ever more precise manner what we
collectively take to be the real world ordinarily leads us to overlook the reconfigured spatio-
temporal pattern and logic of representation according to which information is disseminated.
263
Where events are publicly remembered, their mass media format is rarely brought to mind.
Nonetheless, even where we bring this format to awareness, it is not always possible to detect
the operation of external factors influencing the selection and organization of reported events.
In the political sphere, on one hand, the condensed and concentrated format of representation
facilitates simplification of the real plurality of the public world, where the viewpoints of a pre-
dominant group and its elite monopolize the space of public visibility and of public recollection.
On the other hand, in countries where the content of television broadcasts is not dictated by
the political regime, the survival of news programs, like entertainment and commercial features,
depends on ratings by which their popularity is constantly measured. Programming is conse-
quently commanded by the competition to attract the largest possible audience (Bourdieu 1998:
28–9; Postman and Powers 2008: 75–89). Certainly the mass media, and particularly the Internet
and social media, may be credited with opening otherwise closed information systems to a plural-
ity of external viewpoints that may contradict the officially accepted slant. It is well known that
the Internet and social networks have contributed in recent years to the overthrow of oppressive
dictatorships. Yet this should not lead us to overlook the influence of extraneous political and
commercial factors in even the most open contemporary societies, which may not only limit
but also distort what is brought by the media to mass awareness and retained in public memory.
In view of the altered symbolic sense infused by the media format, public recognition gener-
ated by mass media communications must be distinguished from collective memory where it is
nourished by the contextual depth of a life-world shared by overlapping contemporaneous gen-
erations. Publicly retained remembrance engendered by a simulated video format is essentially
a memory of association, arising out of what are generally programmed assemblages of images
presented through the video interface. Here, beyond the choice of events to endow with public
significance, the mass media format imposes the decontextualized, anonymously configured,
and continually shifting schemata of media events that take precedence over the contextual logic
of experience and remembrance serving as the principal mode of orientation in a shared com-
mon life-world. The mass communications format, in orienting attention to the up-to-date
and in simulating a presence in the life-world, or in providing a satiety of hypermediated pres-
ence that may be substituted for reality (Bolter and Grusin 2000: 53), tends to homogenize and
standardize nuances of perspective, since the short-term format leaves little room for more than
a cursory contextual analysis of fragmented points of view. In this essential way, the logic of
video communication contrasts with the contextual reasoning capable of identifying sources of
fragmented collective remembrance retained by disparate groups, whose respective viewpoints
draw on the deep strata of sedimented symbolic layers in a given contemporaneous framework.
It similarly contrasts with the contextual logic of historical discourse which, when guided by
critical historiographical principles, emphasizes the heterogeneity of interpretations and attempts
to construct coherent accounts of actions and events in the past in the face of different orders
of contextual explanation to which the immediacy of eyewitness accounts is subordinated. The
contextual logic of both collective memory and historical discourse carries with it a singular
assumption: circumstantial evaluations drawn from the immediacy of a given present do not
provide sufficient grounds for understanding the present, for such understanding requires elu-
cidation of current events in relation to a more removed vantage point of the past from which
the present has developed. Current attitudes, according to this logic, find their source in a past
that, even where neglected, forgotten, or otherwise obscured, retains an essential pertinence for
interpreting the concrete reality of the immediately given world.
This identification of the discontinuity that sets our present context in a unique horizon and
distinguishes it from the historical past leads us, in conclusion, to insist on the wholly paradoxi-
cal role that collective memory has assumed in the public sphere in our present era. On one
264
Collective memory
hand, public remembrance serves as a primary vehicle for political identification; on the other
hand, the events on which remembrance focuses, as they are brought to public visibility by the
mass media, become ever more elusive as concrete contents of representation. It is perhaps this
paradox which accounts for the ever-growing proliferation of monuments and archives seek-
ing to collect and to preserve traces of public memory. They provide tangible symbolic images
to reinforce the precarious ties between personal experience and the public sphere of political
action which has become opaque.
This quest has its own inherent dangers. Where memory is assigned a task it cannot hope to
fulfill, that of bridging the abyss between personal and small group identity and a mass public,
this may lead in extreme forms to a denial of the reality of events which recollection cannot
hope to fathom. Where the many-layered complexity of social existence and its public organi-
zation is forgotten, fragmented recollections may all too readily be manipulated to promote the
illusion that they are direct “experiences,” capable of symbolically configuring the coherence
of events as a whole. The yawning gap between the living everyday context of experience
and remembrance and their reconfiguration in the format of public information systems is all
too readily invaded by collective fantasies which radically distort its factual texture. The abyss
between memory and political reality may then be filled by fictional representation of public
identity in the guise of political myths which have become an all too familiar facet of our con-
temporary political world.
Notes
1 In spite of the important contributions of the cognitive sciences to the analysis of memory, it would
reach beyond the scope of this brief chapter to deal with the cognitive approach to collective memory. As
useful as cognitive methods may be for the examination of personal and small-group memory, their role
in elucidating the symbolic embodiment of publicly communicable experience and the symbolic con-
figuration of the public sphere by the mass media, which are central to the theory of collective memory
presented in this chapter, remains to be explored in future investigations.
2 As Nietzsche wrote, for example, in a characteristic passage of Beyond Good and Evil: “It is in no way pos-
sible that a person should not incorporate the traits and the past life of his or her parents, however much
appearances may indicate otherwise. This is the problem of race (Rasse)” (Nietzsche 1968: 228). Unless
otherwise indicated, all translations are my own.
3 My approach to what I describe here as the broader function of the symbol is in part inspired by the
thought of Ernst Cassirer. What I have borrowed from him concerns less the theory of symbolic forms
that he presented in the three volumes of his Philosophy of Symbolic Forms than what he conceived to be
the “primordial forms of synthesis” (Urformen der Synthesis) – space, time, and number – for which sym-
bols provide the ordering principle (Cassirer 1994: 17). In a different philosophical framework, Nelson
Goodman made at least implicit use of the symbol in what I take to be both its narrower and broader
senses. For Goodman, symbols may stand for things or elements other than themselves. Yet the symbol
is not limited to this role for, at another level, it exercises what he aptly termed a more general, “world-
making” function. A “world,” as he wrote, “may be made up of atoms or qualities, of ordinary objects of
certain kinds or of other kinds, of riotous Soutine-like or geometric Braque-like patterns” (Goodman
1981: 130). In his book Ways of World-Making, Goodman elucidated his concept of the symbol’s world-
making function with explicit reference to the work of Ernst Cassirer (Goodman 1978: 1).
Related topics
• Memory and social identity
• Memory and time
• G.W.F. Hegel
• Maurice Halbwachs
• Paul Ricoeur
265
Further reading
Barash, J.A. (2016) Collective Memory and the Historical Past. Chicago, IL and London: University of Chicago
Press.
Erll, A. and Rigney, A. (eds.) (2012) Mediation, Remediation, and the Dynamics of Cultural Memory. Berlin/
Boston, MA: Walter de Gruyter.
Huyssen, A. (1995) Twilight Memories: Marking Time in a Culture of Amnesia. New York and London:
Routledge.
Kansteiner, W. (2006) In Pursuit of German Memory. History, Television, and Politics after Auschwitz. Athens,
OH: University of Ohio Press.
Mendels, D. (ed.) (2007) On Memory: An Interdisciplinary Approach. Frankfurt am Main: Peter Lang.
Olick, J.K., Seroussi, V.V. and Levy, D. (eds.) (2011) The Collective Memory Reader. Oxford: Oxford
University Press.
References
Assmann, A. and Assmann, J. (1988) “Schrift, Tradition und Kultur,” in W. Raible (ed.), Zwischen Festtag
und Alltag: Zehn Beiträge zum Thema ‘Mündlichkeit und Schriftlichkeit’. Tübingen: Narr.
Assmann, J. (2000) Religion und kulturelles Gedächtnis. Zehn Studien. Munich: Beck.
Assmann, J. (2011) Cultural Memory and Early Civilization. Writing, Remembrance, and Political Imagination.
Barash, J.A. (2012) “Articulations of Memory: Reflections on Imagination and the Scope of Collective
Memory in the Public Sphere,” Partial Answers 10, no. 2.
Barash, J.A. (2016) Collective Memory and the Historical Past. Chicago, IL and London: University of Chicago
Press.
Benjamin, W. (1969a) “The Storyteller,” in Illuminations. Trans. H. Zohn. New York: Schocken.
Benjamin, W. (1969b) “On Some Motifs in Baudelaire,” in Illuminations. Trans. H. Zohn. New York:
Schocken.
Benjamin, W. (1969c) “The Work of Art in the Age of Mechanical Reproduction,” in Illuminations. Trans.
H. Zohn. New York: Schocken.
Bolter, J.D. and Grusin, R. (2000) Remediation: Understanding New Media. Cambridge, MA: MIT Press.
Bourdieu, P. (1977) Outline of a Theory of Practice. Trans. R. Nice. Stanford, CA: Stanford University Press.
Bourdieu, P. (1998) Sur la télévision suivi de l’empire du journalism. Paris: Raisons d’Agir.
Burke, P. (1989) “History as Social Memory,” in T. Butler (ed.), Memory. Oxford: Blackwell.
Butler, S. (1910) Unconscious Memory. London: Fifield.
Cassirer, E. (1994) Philosophie der symbolischen Formen, vol. 3. Darmstadt: Wissenschaftliche Buchgesellschaft.
Connerton, P. (1989) How Societies Remember. Cambridge: Cambridge University Press.
Croce, B. (1929) Aesthetic as a Science of Expression and General Linguistic (1902). Trans. D. Ainslie. London:
Macmillan.
Dilthey, W. (1973) Der Aufbau der geschichtlichen Welt in den Geisteswissenschaften. In Gesammelte Schriften,
vol. 7. Stuttgart: Teubner.
Fassin, D. (2007) When Bodies Remember: Experiences and Politics of AIDS in South Africa. Trans. A. Jacobs
and G. Varro. Berkeley and Los Angeles, CA: University of California Press.
Goodman, N. (1978) Ways of World-Making. Indianapolis, IN: Hackett.
Goodman, N. (1981) “Routes of Reference,” in Critical Inquiry 1, no. 1.
Halbwachs, M. (1958) The Psychology of Social Class. Trans. C. Delavenay. Glencoe, IL: Free Press.
Halbwachs, M. (1994) Les cadres sociaux de la mémoire. Paris: Albin Michel.
Halbwachs, M. (1997) La mémoire collective. Paris: Albin Michel.
Hegel, G.W.F. (1970) Phänomenologie des Geistes. Vol. 3 of Werke. Frankfurt am Main: Suhrkamp.
Hering, E. (1921) Über das Gedächtnis als eine allgemeine Funktion der organischen Materie (1870). Leipzig:
Akademische Verlagsgesellschaft.
Huyssen, A. (2000) “Present Pasts: Media, Politics, Amnesia,” Public Culture 12, no. 1.
Kansteiner, W. (2002) “Finding Meaning in Memory: A Methodological Critique of Collective Memory
Studies,” History and Theory 41.
Klein, E.L. (2000) “On the Emergence of Memory in Historical Discourse,” Representations 69.
Luhmann, N. (2004) Die Realität der Massenmedien. Wiesbaden: Vs Verlag.
266
Collective memory
Mauss, M. (1983) “Les techniques du corps,” (1934) in Sociologie et Anthropologie. Paris: Presses Universitaires
de France.
Mendels, D. (2004) Memory in Jewish, Pagan and Christian Societies of the Graeco-Roman World. London and
New York: Continuum and T&T Clark International.
Nietzsche, F. (1968) Jenseits von Gut und Böse, Zur Genealogie der Moral, 1886–1887, in G. Colli and
M. Montinari (eds.), Vol. 6 of Kritische Gesamtausgabe. Berlin: Walter de Gruyter.
Nora, P. (2002) “L’avènement mondial de la mémoire,” Transit 22.
Olick, J. (1999) “Collective Memory: The Two Cultures,” Sociological Theory 17, no. 3.
Postman, N. and Powers, S. (2008) How to Watch TV News. London: Penguin.
Renan, E. (1905) L’avenir de la science: Pensées de 1848. Paris: Calmann-Lévy.
Renan, E. (1992) Qu’est-ce qu’une nation? et autres essais politiques. Paris: Agora.
Ricoeur, P. (2004) Memory, History, Forgetting. Trans. K. Blamey and D. Pellauer. Chicago, IL: University
of Chicago Press.
Winter, J. (2006) Remembering War: The Great War between Memory and History in the Twentieth Century.
New Haven, CT: Yale University Press.
267
21
MEMORY AND SOCIAL
IDENTITY
1. Memory and social identity

We are the stories we tell about ourselves. Of the many ways to characterize “the self,” we
argue that narrative practice is a necessary component. Our identity is critically defined by the
narratives we create about ourselves, and this narrative self affords a sense of emotional coher-
ence and continuity in life, that is, a sense that one is the same person with the same inner life,
same goals, values, and commitments across time (Barnes 1998; Conway et al. 2004; Fivush
2010; Habermas and Reese 2015; McAdams 2001). In this chapter, we examine psychological
theory and empirical evidence demonstrating that a coherent narrative identity is created in
everyday social interactions, beginning early in child development, and that individual narra-
tive identity is very much an evolving socially constructed process. We use the term “self” and
“identity” interchangeably to refer to the sense of personal continuity and emotional coher-
ence conferred by the narrative life story. We make no claims that this is the only definition or
aspect of self; only that this is a critical aspect of self/identity that is created through storying
our lived experience.
We begin by explicating narrative identity in more detail before turning to examine how
narrative identity begins to develop in early parent–child reminiscing. We argue that the early
emergence and ongoing maintenance of a narrative identity is a socioemotional construction,
forged in interactions in which we co-narrate the experiences of our lives with others who
validate, contest, negate, and negotiate what happened and what it means. In early parent–child
reminiscing, children learn the forms and functions of narrating the personal past, and begin to
develop the initial skills that provide the foundation for the construction of identity. We dem-
onstrate that this early reminiscing occurs within more embedded socio-emotional relationships
between parent and child, such that the parent–child attachment relationship both contributes
to and emerges from forms of parent–child reminiscing. We argue that dialectical relations
between reminiscing and socio-emotional attachment undergirds the process by which children
internalize the narrative tools and cultural values to understand and evaluate new experiences
across development. Importantly, both the attachment relationship and forms of parent–child
reminiscing are individually variable, and these variations facilitate individual differences in the
development of a narrative identity across development. In this way, reminiscing about specific
268
Memory and social identity
life experiences leads to larger life themes about self in the world, that filter back into how
individuals narrate the events of their lives and create selves that are competent, worthy and
loved, leading to an emotionally coherent narrative identity.
2. Narratives and narrative identity

Reminiscing about our past experiences is ubiquitous in social interaction. Narratives of past
experiences, whether of the previous day or from decades ago, occur approximately every
5 minutes in spontaneous conversations (Bohanek et al. 2009; Miller 1994), and approxi-
mately 90 percent of even moderately emotional experiences are shared in conversation within
48 hours of occurrence (Rime 2007). Importantly, sharing memories is not about “accuracy”
(Marsh 2007), or “getting things right.” We share our experiences with others, especially our
emotional experiences, in order to create a shared understanding of what this event means,
and in these tellings, the narrative is shaped and reshaped by both the teller and the listener
to create a negotiated shared understanding. Thus co-constructed narratives help shape both
the teller’s and the listener’s sense of self. Our approach is sympathetic to recent philosophical
approaches that argue that conceptions of “memory” as storage and retrieval of information
need to be revised (Brockmeier 2015; Goldie 2012). Instead, memory, or perhaps more aptly,
remembering, is a fluid dynamic process. Each and every time a past experience is recalled
and narrated with others, the current context shapes how that remembering occurs and what
will subsequently be remembered. Evidence from neuroscience (Dudai 2004), cognitive psy-
chology (Rubin 2005), collective memory (Coman et al. 2009) and literary studies (Herman
2012) converge on remembering as a process that occurs in the present in ways that re-create
anew re-presentations of past experiences (see Brockmeier 2015, for an extensive and excel-
lent review). Moreover, this kind of remembering creates meaning through creating narratives
(Bruner 1990). We do not only know ourselves through narratives; as Hutto (2007) argues,
an integral way to gain insight into other persons’ identities and our own is through narrative
practice. Knowing someone entails more than adequately understanding current beliefs and
feelings in order to explain behaviors. Knowing someone further entails that individuals are
influenced by past behaviors and history, current circumstances and commitments, goals, and
future projects, i.e., knowing their “story” (Fivush and Nelson 2006; Lagattuta and Wellman
2001). We make sense of ourselves and others through sharing narratives.
Perhaps it is not surprising, then, that infants are born into a densely narrated world; par-
ents and grandparents begin telling family stories to newborns well before they could possibly
understand let alone participate in these stories (Fiese et al. 1995). The prevalence of narratives
in everyday social interactions also strongly suggests that they play a role in creating personal
meaning. Indeed, many theorists have argued that narrative is the fundamental way in which
humans turn the passage of time into lived experience, from a chronology of actions in the
world to a meaningful sequence of one’s life (Bruner 1987; Ricoeur 1991; Sarbin 1986).
When sharing personal stories, people do not just narrate the “landscape of actions” but also
the “landscape of consciousness” (Bruner 1987). A narrative moves beyond recollecting a
series of actions to retelling a storied subjective experience that includes agents with thoughts,
feelings, goals, obstacles, and outcomes. We construct personal stories filled with such com-
ponents in order to organize and understand an event as a personal experience (Fivush 2010;
Hutto 2007). Through narration, we turn our memories of experiences into a subjective
understanding and evaluation of the world and ourselves. In narrating the experiences of our
269
lives, we create a narrative identity, a sense of self as a lived history of emotionally coherent
experiences (Conway et al. 2004; Schectman 2003).
McAdams (1996, 2008) has laid out the most widely accepted model for narrative identity.
The model is a framework for personality, in which narrative identity captures individuals’ efforts
to make meaning in life and understand their selves as individuals. The model specifies five
levels to personality: genetics, traits, character adaptations, identity and life story, and culture.
Genetics refer to the biological underpinnings to traits. Character traits are defined as stable,
dispositional behavioral patterns. Character adaptations include motives, goals, and values. The
primary emphasis of the model is the identity and life-story level because it focuses on the subjec-
tive meaning in one’s life through narrative processing. In McAdams’s framework, traits lay out a
framework, adaptations fill in details, and stories give rise to meaning. In particular, the narrative
identity model seeks to investigate narratives for personal themes and intentions as reflecting an
emotionally coherent self.
Importantly, McAdams proposes a developmental theory, but he does not focus on how nar-
rative identity is constructed, negotiated and created from autobiographical memories shared in
everyday reminiscing beginning early in life that sets the stage for adult narrative identity. It is to
these issues that we now turn. We focus on how narrative identity, including personal themes
and intentions, are constructed beginning in early childhood. We argue that these components
of self-identity are socially mediated, such that they give rise to different kinds of experiences
and create individual differences in narrative identity.
3. Early parent–child reminiscing and narrative development

From the moment of birth, infants are embedded in complex cultural worlds, and are brought
into becoming competent participating members of cultural practices. As Vygotsky (1978) laid
out, children are “scaffolded” into culture. Parents, and other competent cultural members,
engage children in activities deemed to be appropriate for members of that culture, and through
this engagement they model and elicit appropriate cultural behaviors in an increasingly complex
stepped system. For example, literacy, a clearly critical cultural skill in industrialized cultures,
begins with the simple visual array of letters and numbers on infants’ clothing, toys, and books.
Quickly, infants are drawn into participating more in nascent literary activities, initially learn-
ing to name letters and numbers, and then to sing alphabet songs and recite rhymes. Again, as
these skills are elicited and mastered in socially structured contexts children are “asked” to do
“more”—to begin to read and write, and so on, each scaffolded step bringing children into a
fully literate cultural world.
Stemming from Vygotsky’s (1978) sociocultural theory, Nelson and Fivush (2004) pro-
posed the developmental sociocultural model of autobiographical memory. They argue that
telling coherent stories of self is a culturally critical skill and that narratives are the cultural
tools provided to help structure experience into normative coherent forms. Whether it is
narrating self for a job interview, meeting a possible romantic partner, or simply interacting
with family and friends, adults must be able to present a coherent and compelling narrative
of who they are and how they came to be this way. This skill may be especially important in
Western industrialized cultures where individual autonomy and uniqueness is prized (Wang
2013), and individuals move from place to place, creating the need for continuity through
a self-narrative rather than through family or community relationship systems (Fivush 2010;
McAdams 1990; Nelson 2003). This approach emphasizes the cultural specificity of the value
and forms of narrative identity. Cultures create opportunities throughout development for
270
children to become competent members of their culture, and the ways in which identities
are valued and constructed in narrative are culturally variable. In this chapter, we focus on
broadly Western European cultures that generally value autonomous independent selves, but
see Wang (2013) for an excellent explication of culture, autobiographical memory, and self.
The forms of narrative reminiscing

Because creating a narrative about the self is a critical skill in Western cultures, children are
drawn into activities that will facilitate their learning of this skill early in development, with
adults, usually parents, taking on the role of “scaffolding” the child’s performance. With
development, the child internalizes these cultural tools. Substantial research has now demon-
strated that parents, especially mothers, engage frequently in reminiscing with their children
beginning as early as 18 months of age and continue this practice throughout development
(see Fivush et al. 2006, and Fivush 2012, for reviews). The majority of research has examined
mothers, for both theoretical and pragmatic reasons that are beyond the scope of this chapter
(but see Fivush and Zaman 2015, for a full discussion of this issue). Importantly, mothers
show consistent individual differences in maternally guided reminiscing along a dimension
of elaboration. More elaborative mothers reminisce more frequently with their children, co-
construct more richly detailed and elaborately textured narratives, as well as more coherent
and more emotionally expressive narratives with their children than do less elaborative moth-
ers. Maternal elaborative style is consistent over time and across siblings, but, importantly,
mothers who are highly elaborative during reminiscing are not necessarily more elabora-
tive during free play, book reading, or conversation during caregiving activities (Haden and
Fivush 1996; Hoff-Ginsburg 1991). Thus, elaborative reminiscing reflects a specific context,
and possibly a specific set of goals. Highly elaborative mothers self-report using reminiscing
as a means of creating close emotional bonds with their child and of helping their child learn
about their identity (Kulkofsky et al. 2009). In doing so, the functions of reminiscing appear
to be threefold by providing: (1) sociocultural norms of narrative practice that include sharing
and co-constructing narratives with others; (2) structure, or form, for personal experiences
that provide culturally mediated forms of narrative identity; and (3) a means of creating and
maintaining social bonds that help create and maintain a sense of emotional coherence across
time. Equally important, individual differences in maternal reminiscing style suggest that child
outcome is variable; level of maternal elaboration leads to differences in how individual chil-
dren begin to construct their own life narratives.
In fact, both longitudinal and intervention research has demonstrated that more elaborative
maternal reminiscing earlier in the preschool years facilitates the development of more coherent
and elaborative narratives as children grow older (see Fivush et al. 2006 for a review). Research
extending these ideas into adolescence has shown that mothers continue to display differences in
elaborative reminiscing style (Fivush et al. 2009; Habermas et al. 2010) and that these maternal
differences in elaboration continue to be related to children’s autobiographical memories, such
that adolescent children of more highly elaborative mothers have earlier age of first memory
(Jack et al. 2009), narrate more coherent and detailed narratives (Reese et al. 2010a), and have
a more complex understanding of how the various events of their life coalesce into a coherent
life story (Reese et al. 2010b). Thus it is clear that children’s developing autobiographical nar-
ratives are constructed within social interactions, and that the form of these social interactions
has a significant impact on how children come to tell the stories of their life. But how might the
functions of these early scaffolded narratives develop?
271
The functions of narrative reminiscing

Highly elaborative mothers help scaffold structurally more complex personal narratives. Highly
elaborative mothers also scaffold more emotionally expressive and regulated personal narratives.
That is, highly elaborative mothers help their children to understand their emotional experi-
ences in more complex and regulated ways, and, in this way, help children construct a more
emotionally coherent sense of self. For example, in this excerpt, a mother is discussing a difficult
event with her 4-year-old daughter; the daughter’s friend has moved away:
Mother: Well, one thing that made you really sad is when your best friend Sheila
moved away, right?
Child: (nods)
Mother: Yeah, and did we watch all her things go on the moving truck?
Child: (nods)
Mother: Uh-huh, and do you remember why she had to move away?
Child: . . . Because Sheila’s Dad had to work.
Mother: Sheila’s Daddy was going to start working at a new job . . . And do you still
miss Sheila when you think about it? Yes?
Child: Yes.
Mother: It makes you sad. Doesn’t it?
Child: (nods)
Mother: But is she still your friend even far away?
Child: (nods)
Mother: Yes! What can you do even though she’s far away?
Child: Give her a happy letter with a (drawing) on it.
Mother: Give her a happy letter, right, and we have a drawing, don’t we?
In this small everyday narrative interaction, we see multiple ways in which the mother scaffolds
emotional regulation and a positive sense of self. First, the mother acknowledges and validates
the child’s feelings, making it clear that this reaction is appropriate given the situation. Note also
that the narrative is about a relationship and the importance of relationships more generally is
clearly highlighted (use of the term “best friend,” and the statement that “she is still your friend”
are ways to highlight the importance of relationships for defining who one is). And, of course,
the mother helps the child develop a strategy that will both manage her emotions and simultane-
ously help maintain the relationship. Thus within this brief narrative, we see validation of self,
of relationships, and strategies to regulate difficult emotional experiences.
Importantly, mothers and children are not always in such strong agreement about the experi-
ence; they often negotiate and validate the child’s emerging sense of identity, as shown in this
excerpt between a mother and her 5-year-old daughter discussing going to the wake of the
child’s pre-school teacher, who had recently died of cancer:
Mother: Well, I think what we’re going to talk about next is going to Susan’s [the
child’s pre-school teacher] wake.
Child: Uh, well I remember when they were crying and Sally [the teacher’s daughter],
Sally started crying.
Mother: . . . mmm hmmm. And it was very sad to see her so sick.
Child: Yes.
Mother: And then what happened?
272
Child: Well she, they, they, did everything but it didn’t work and she died.
Mother: Yeah, and it was sad, wasn’t it?
Child: Yeah. (very softly)
Mother: And you begged Mommy and Daddy to take you to the wake.
Mother: And what was the wake like?
Child: Well it had sadly music and it was really sad to talk about so I didn’t want to
talk about it.
Mother: . . . Didn’t we go talk to her daughter?
Mother: And we gave her big hugs.
Child: . . . Yeah . . . but I don’t want to talk about this because you’re almost going
to make me cry.
Mother: Okay, I won’t, we won’t talk about it anymore.
This is clearly a difficult experience for this young girl, and she struggles with her emotions.
The mother helps provide an emotional framework that both validates and places the daugh-
ter’s response in a larger framework; the mother immediately validates that it was sad to see the
teacher sick, and sad that she died, but the mother further helps provide a coherent framework
for the child about these events: that the teacher was ill, that the doctors did everything that they
could, and that the daughter really wanted to go to the wake even though, as it turns out, it was
an emotionally difficult experience. The mother pushes the child to explore her feelings and
share them, and then turns to the fact that these feelings are shared with others, and their role in
this situation is to comfort others. The daughter finally gets across that she really is overwhelmed
and cannot discuss this and the mother accepts this. Still, by the end of the conversation, we have
a sense of the shape of the event, and a form of resolution in that it was sad for everyone and they
all shared hugs. Thus, although the mother acknowledges and accepts that this is too difficult for
her daughter to talk about, she at the same time models that it is important to talk about these
things. Further, she helps define the child’s identity as one of a caring and sympathetic person,
who wants to comfort those in distress. We note that in both of these cases, mothers are also
implicitly validating fairly gender-typed identities for their daughters. Gender differences and
gendered narrative voices are beyond the scope of this chapter, but are an important part of
narrative identity (see Fivush and Zaman 2015, for a theoretical discussion and review of gen-
dered narrative identity). Here, the important point is that beginning very early in development,
mothers are helping their children understand their difficult emotional experiences in ways that
provide regulation. Even when bad things happen, the child can depend on others for help and
support, and for strategies to actively engage in the world in more proactive positive ways.
Thus event narratives that explain why the child might experience certain emotions and
what the child might do to cope with aversive emotions helps the child understand their
inner life, and create a life narrative that connects the landscape of actions with the landscape
of consciousness. This intertwining of events in the world and events “in the head” facilitates
the construction of a reflective and interpretive life narrative (Fivush 2010, 2012). Through
co-constructing more emotionally expressive and regulated narratives, children of more emo-
tionally elaborative mothers develop a more nuanced understanding of emotion (Laible and
Thompson 2000) and tell more emotionally expressive personal narratives, suggesting that they
have a more complex subjective identity (Fivush et al. 2003; Fivush and Nelson 2006).
This analysis raises the question of how and why some mothers may be better able to scaf-
fold emotionally complex and regulated narratives with their children, narratives that help their
273
children develop a sense of a coherent and regulated self. We argue that these scaffolded remi-
niscing conversations occur within enduring socio-emotional relationships between parent and
child that help create larger narrative themes. More specifically, more elaborative emotionally
coherent mothers also have more secure mother–child attachment bonds with their children, and
these secure bonds facilitate narratives of self as competent, safe, and secure. These bonds, in turn,
affect the way individuals interpret their stories—their past histories, goals, and commitments
that compose a personal identity. We outline theory and assumptions about the attachment rela-
tionship and then discuss how attachment and reminiscing are dialectically related throughout
development in ways that facilitate individual variability in narrative identity.
4. Attachment theory
Attachment refers to the strong emotional bond felt between an infant and caregiver, typically
the mother (Bowlby 1988). Interdependent innate attachment-related systems are theorized to
work with each other to achieve physical and psychological security for the infant. The two
primary systems are the exploratory system and the attachment behavior, or support-seeking
system. The exploratory system leads to actively seeking and experiencing novel stimuli in the
environment in order to promote curiosity and growth. But when threat arises during explora-
tion, the support-seeking system leads to seeking proximity to the attachment figure in order
to promote survival, safety, and emotional comfort and psychological regulation. To the extent
that infants consistently receive psychological comfort from their attachment figures, they gain a
sense of security, trust, and safety in the world, and especially the secure knowledge that comfort
will be received in times of distress. Threats to physical accessibility, responsiveness, and open
communication may lead to distrust and insecurity. These early attachment-related childhood
experiences lay the groundwork for an enduring sense of safety and security in the world, and,
consequently, we argue, set the stage for overarching themes of self in the life story.
Internal working model

Bowlby (1979) proposed that, based on maternal sensitive and responsive caregiving behaviors,
infants create generalized event representations, or schematic knowledge, on how and when
to obtain psychological comfort. In early childhood, this schematic knowledge is theorized to
form the “internal working model” (IWM) that guides understanding of past experiences, the
current environment, and anticipates the future. Infants that receive consistent and responsive
care, especially when distressed, develop a general sense of safety and care, which is the hallmark
of secure attachment. Infants may develop an anxious-ambivalent attachment if their attachment
figures are inconsistent with both the recognition and response to their distress. Infants may
develop an avoidant attachment style if their attachment figures consistently lack the receptivity
and responsiveness to their bids for comfort when distressed. From these experiences, securely
attached children construct representations of the world as safe; they are loved and people can
be trusted to comfort and help them in times of difficulty. In contrast, insecurely attached chil-
dren represent the world as unsafe; they are not loved and may not be lovable; people cannot
be trusted to come to their aid when needed. These attachment schema, or internal working
models, provide a lens through which individuals understand the current environment, antici-
pate threats, and make future plans.
Bowlby (1979) argued attachment style is utilized from “cradle to grave,” beginning
in infancy and maintained into adulthood. Indeed, the IWM continues to provide a lens
into lived experienced in adulthood. Early attachment experiences set the stage for ongoing
274
attachment relationships and behaviors with significant others across the lifespan. For example,
more securely attached adults are more trusting of significant others (Arriaga et al. 2013) and
have less marital conflicts (Campbell et al. 2005). There are also cognitive consequences of
attachment style. More insecurely attached adults tend to be more hypervigilant to threats,
suggesting that they approach the world as unsafe and monitor their world for possible dif-
ficulties (Shaver and Mikulincer 2002). There also appear to be differences in motivational
profiles between securely and insecurely attached individuals (Elliot and Reis 2003). In par-
ticular, more securely attached people view experiences from a more achievement-oriented
motivational profile than insecurely attached people. For example, securely attached college
students were more likely to think optimistically about their class performance compared
to insecurely attached students, who appeared more preoccupied with a fear of failure and
constructed their goals accordingly. These patterns suggest that individuals process ongoing
experience in line with their well-developed attachment-related schema, and these schema
facilitate quicker and deeper processing of schema-consistent information. Thus, these sche-
mas, or internal working models, are employed in stylistic ways based on personal history
to respond to threats and regulate emotions throughout adulthood. An important question
becomes how these attachment schemas, or internal working models, are maintained across
development. We argue that one critical part of this process is through narrative reminiscing.
5. Reminiscing and attachment

As parents and children begin to reminisce about their shared past in early childhood, the co-
constructed narratives are related to the parent–child attachment bond in at least two ways. First,
more securely attached parent–child dyads are able to communicate in more open and expres-
sive ways than less securely attached dyads, especially about distressing events (Bost et al. 2006;
Etzion-Carrasso and Oppenheim 2000; Oppenheim et al. 1997). Second, in the process of
co-narrating more elaborate and emotionally regulated narratives about these distressing events,
parents are helping their children build a more secure internal working model by providing the
narrative tools for these representations (Bretherton 1995; Fivush and Reese 2002). More spe-
cifically, within more securely attached dyads, parents elaborate on experiences of distress that
provide resolutions within caring relationships. The narrative excerpts above illustrate this, as, in
both examples, the mother ends the narrative with reassuring the child that comfort is available.
Such narrative practice not only helps a child down-regulate their distress in the moment, it also
helps the child internalize a coherent regulated narrative style that facilitates a sense of security in
the world and confidence they can handle stressors in the future due to past successful emotional
resolutions. Thus early attachment relationships contribute to emerging reminiscing style and
reminiscing style contributes to the evolving attachment relationship across childhood, forming
dialectical relations that help maintain stability of the internal working model.
In this way, early attachment and reminiscing has a long-lasting effect on narrative identity.
Substantial research has demonstrated that more securely attached infants develop into more
socially adept, empathic, and emotionally regulated children and adolescents than less securely
attached children (see Sroufe 2005, for review). More important for our arguments, early
attachment continues to be related to individuals’ developing personal narratives and identity.
Attachment and narrative identity

Attachment style influences the way in which individuals recall their past experience, and thus
impacts on narrative identity. Securely attached individuals are able to employ more flexible and
275
open cognitive and emotional processing styles that allow them to engage with ongoing expe-
rience in more complex ways. Because of this open processing of experience, and a coherent
IWM based on safety and security, securely attached individuals are subsequently able to recall
specific past experiences in a coherent and realistic manner. More securely attached children tell
more coherent, more emotionally expressive, and regulated narratives of self through childhood
and early adolescence than do less securely attached children (Gini et al. 2007; Oppenheim
et al. 2007; Zaman and Fivush 2011). As adults, securely attached individuals provide coherent
and emotionally regulated narratives about their childhood experiences (Main 1995), and about
their current relationships (Treboux et al. 2004). In contrast, more insecurely attached individu-
als recall memories in a more confused, overly detailed and even contradictory manner.
Through narrating more coherent and emotionally regulated personal narratives, securely
attached individuals are able to construct a personal identity around themes of being safe and
being loved, and a sense of understanding that others will be there for support through difficult
times. More specifically, we argue that the way in which attachment-related themes of support-
seeking and exploration are expressed in personal narratives is both related to attachment style,
and has effects on broader constructs of identity and well-being. In support of this argument,
Graci and Fivush (2016) asked young adults to narrate highly stressful experiences and assessed
both their attachment status and their identity through personal growth. Narratives were explic-
itly coded for the attachment themes of exploration and support-seeking. Participants who
expressed more exploration in their narratives of highly distressing life events reported more
positive personal growth from the experience, compared to others who narrated the experience
from a single, flat viewpoint. For example, in this excerpt, a young adult chose to narrate about
her father’s physical illness as her most traumatic experience:
Most traumatic was when my father was in the hospital. He was diagnosed with a heart
condition. This was difficult because it was during finals time. My grades took a hit
because I was always at the hospital. I found out when I was leaving a psych test. This
hit me hard because my father went through a lot of pain. I felt as if I got closer to him
and my entire family. I realized how important my parents are. I developed in several
ways after this event. I became a lot more mature. I would take care of my little sister.
For example, I’d pick her up from school all the time.
This narrative has similar themes to those expressed in the excerpts of mother–child reminiscing
discussed earlier. This young woman spends most of her narrative time evaluating how and why
this event was important to her. Most intriguing, in addition to narrating how important family
relationships are, she goes a step further to reflect on how this event was a springboard for per-
sonal growth, and mentions one specific way in which she has implemented this new reflection
on life meaning through her relationship with her sister. As in the mother-scaffolded excerpts,
we see the narrative end with a way on acting on one’s values in order to maintain important
relationships. Thus this is highly suggestive that the ways in which mothers scaffold these early
conversations become templates for emotional understanding of self as children grow older.
Importantly, Graci and Fivush also found that increased narrative exploration and personal
growth were highest for securely attached individuals. Avoidantly attached individuals appeared
to be less likely to grow from their distressing experience, perhaps because they do not ben-
efit from processing their narratives in an exploratory way. These findings suggest that avoid-
antly attached individuals may be inhibited from adaptively growing from distressing situations,
whereas securely attached individuals’ exploration systems allow for more complex emotional
reflection. Thus attachment status influences how individuals narrate past experiences, and the
276
way in which they narrate, in turn, affects their identity growth. We can extrapolate cautiously,
but these findings suggest a cascading effect of attachment-related themes expressed in personal
narratives on creating and stabilizing a sense of self as competent to deal with stressful situations
in ways that promote positive identity growth.
6. Narrative identity and life themes as socially constructed

We have traced how narrative identity begins to be constructed in early development in
everyday social interactions. Moreover, these interactions occur within socio-emotional
relationships that color the way in which lived experience is constructed in narrative form.
Securely attached parents and children engage in elaborated and emotionally expressive and
regulated co-constructed narratives of everyday and challenging experiences, and these chil-
dren develop more coherent, emotionally expressive, and regulated narratives as they grow
older. As these securely attached and narratively elaborative individuals develop into adoles-
cents and young adults, these early foundations set the stage for the formation of a narrative
identity and life story. As adults, more securely attached individuals continue to narrate their
experiences more coherently and in more emotionally regulated ways, and, importantly, nar-
ratives that coherently express attachment-related themes of exploration and support seeking
are related to greater personal growth. It is in this sense that attachment and reminiscing
mutually contribute to an emotionally coherent narrative identity through the lens of these
overarching themes of security and comfort. Moreover, this lens is self-perpetuating. The
more an individual narrates and understands the world as safe, and the self as secure and
loved, the more experiences will be framed in this way in ongoing narratives, and the more
that experiences are expressed in this narrative style, the more it reinforces this sense of self.
Provocatively, there is emerging evidence that these individual developmental trajectories are
echoed in the generational transmission of narrative form and function; the stories that cross
from one generation to the next recreate emotionally coherent narratives of individual experi-
ence across the generations, thus perpetuating the creation of emotionally coherent identities
over generational time (Fivush and Merrill 2016).
Thus our very sense of self is constructed in social interactions that facilitate both the forms
and functions of narrative identity. Narratives of self are formed within broader socio-emotional
contexts that provide frameworks for self as situated in relationships, and these relationships, in
turn, facilitate certain forms of narrating self. Importantly, we have shown how differences in
mother–child reminiscing emerge from and contribute to the ongoing emotional relationship
and lead to individual differences in an emotionally coherent sense of self. These differences in
developmental trajectory strongly imply that these outcomes are the result of social interaction
and subsequently facilitate different forms of social interaction across the lifespan. Thus narrative
identity is both constructed and understood within broader social interactions. As we said at
the beginning of this chapter, we are the stories we tell about ourselves. For securely attached
individuals, these stories reflect and reinforce development histories of emotional relationships
that create and maintain a sense of self as loved, safe, secure, and competent.
Related topics
• Memory and emotion
• Memory and self-consciousness
277
• Memory and narrativity

• Maurice Halbwachs
• Paul Ricoeur
Further reading
Fivush, R. (2010). “The development of autobiographical memory.” Annual Review of Psychology, 62: 2.1–2.24.
doi:10.1146/annurev.psych.121208.131702.
Fivush, R. (2012). “Subjective perspective and personal timeline in the development of autobiographi-
cal memory.” In D. Bernsten and D. Rubin (eds.), Theories of Autobiographical Memory (pp. 226–45).
McAdams, D. P. (2001). The psychology of life stories. Review of General Psychology, 5: 100–122. doi.
org/10.1037/1089-2680.5.2.100.
Shaver, P. R. and Mikulincer, M. (2002). “Attachment-related psychodynamics.” Attachment and Human
Development, 4: 133–61.
References
Arriaga, X. B., Kumashiro, M., Finkel, E. J., VanderDrift, L. E., and Luchies, L. B. (2013). “Filling
the void bolstering attachment security in committed relationships.” Social Psychological and Personality
Science, 5: 398–405.
Barnes, H. E. (1998). The Story I Tell Myself: A Venture in Existentialist Autobiography. Chicago, IL:
Bohanek, J. G., Fivush, R., Zaman, W., Thomas-Lepore, C., Merchant, S., and Duke, P. (2009).
“Narrative interaction in family dinnertime conversations: Relations to child well-being.” Merrill
Palmer Quarterly, 55(4): 488–515.
Bost, K. K., Shin, N., Mcbride, B. A., Brown, G. L., Vaughn, B. E., Coppola, G., Veríssimo, M., Monteiro,
L., and Korth, B. (2006). “Maternal secure base scripts, children’s attachment security, and mother–child
narrative styles.” Attachment and Human Development, 8(3): 241–60. doi:10.1080/14616730600856131.
Bowlby, J. (1988). A Secure Base: Parent–Child Attachment and Healthy Human Development. New York:
Basic Books.
Bretherton, I. (1995). “A communication perspective on attachment relationships and internal working
models.” Monographs of the Society for Research in Child Development: 310–29.
Brockmeier, J. (2015). Beyond the Archive: Memory, Narrative, and the Autobiographical Process. New York:
Bruner, J. (1987). “Life as narrative.” Social Research, 54(1): 11–32.
Bruner, J. (1990). Acts of Meaning. Cambridge, MA: Harvard University Press.
Campbell, L., Simpson, J. A., Boldry, J., and Kashy, D. A. (2005). “Perceptions of conflict and support
in romantic relationships: The role of attachment anxiety.” Journal of Personality and Social Psychology,
88(3): 510–31.
Coman, A., Brown, A. D., Koppel, J., and Hirst, W. (2009). “Collective memory from a psychological
perspective.” International Journal of Politics, Culture, and Society IJPS, 22(2): 125–41.
Conway, M. A., Singer, J. A., and Tagini, A. (2004). “The self in autobiographical memory: Correspondence
and coherence.” Social Cognition, 22: 491–529. doi:10.1080/01650250344000460.
Dudai, Y. (2004). “The neurobiology of consolidations, or, how stable is the engram?” Annu. Rev. Psychol.,
55: 51–86.
Elliot, A. J. and Reis, H. T. (2003). “Attachment and exploration in adulthood.” Journal of Personality and
Social Psychology, 85(2): 317–31.
Etzion-Carasso, A. and Oppenheim, D. (2000). “Open mother–pre-schooler communication: Relations
with early secure attachment.” Attachment and Human Development, 2: 347–70. doi:10.1080/146167
30010007914.
Fiese, B. H., Hooker, K. A., Kotary, L., Schwagler, J., and Rimmer, M. (1995). “Family stories in the early
stages of parenthood.” Journal of Marriage and the Family, 57: 763–70. doi:10.2307/353930.
278
Fivush, R. (2010). “The development of autobiographical memory.” Annual Review of Psychology, 62:
2.1–2.24. doi:10.1146/annurev.psych.121208.131702.
Fivush, R. (2012). “Subjective perspective and personal timeline in the development of autobiographi-
cal memory.” In D. Bernsten and D. Rubin (eds.), Theories of Autobiographical Memory (pp. 226–45).
Fivush, R., Berlin, L., Sales, J. D., Mennuti-Washburn, J., and Cassidy, J. (2003). “Functions of parent–
child reminiscing about emotionally negative events.” Memory, 11: 179–92.
Fivush, R., Haden, C. A., and Reese, E. (2006). “Elaborating on elaborations: The role of maternal
reminiscing style in cognitive and socioemotional development.” Child Development, 77: 1568–88.
doi:10.1017/CBO9780511527913.014.
Fivush, R., Marin, K., McWilliams, K., and Bohanek, J. G. (2009). “Family reminiscing style: Parent
gender and emotional focus in relation to child well-being.” Journal of Cognition and Development, 10(3):
210–35.
Fivush, R. and Merrill, N. (2016). “An ecological systems approach to family narratives.” Memory Studies,
9(3): 305–14.
Fivush, R. and Nelson, K. (2006). “Parent–child reminiscing locates the self in the past.” British Journal of
Developmental Psychology, 24: 235–51.
Fivush, R. and Reese, E. (2002). “Reminiscing and relating: The development of parent–child talk about
the past.” In J. Webster and B. Haight (eds.), Critical Advances in Reminiscence Work. New York: Springer.
Fivush, R. and Zaman, W. (2015). “Gendered narrative voices: Sociocultural and feminist approaches
to identity.” In K. McLean and M. Syed (eds.), The Oxford Handbook of Identity Development. New
York: Oxford University Press.
Gini, M., Oppenheim, D., and Sagi-Schwartz, A. (2007). “Negotiation styles in mother–child narra-
tive co-construction in middle childhood: Associations with early attachment.” International Journal of
Behavioral Development, 31: 149–60.
Goldie, P. (2012). The Mess Inside: Narrative, Emotion, and the Mind. Oxford: Oxford University Press.
Graci, M. E. and Fivush, R. (2016). “Narrative meaning making, attachment, and psychological growth
and stress.” Journal of Social and Personal Relationships. doi:10.1177/0265407516644066.
Habermas, T., Negele, A., and Mayer, F. B. (2010). “‘Honey, you’re jumping about’ – Mothers’ scaffolding
of their children’s and adolescents’ life narration.” Cognitive Development, 25(4): 339–51.
Habermas, T. and Reese, E. (2015). “Getting a life takes time: The development of the life story in
adolescence, its precursors and consequences.” Human Development, 58(3): 172–201.
Haden, C. A. and Fivush, R. (1996). “Contextual variation in maternal conversational styles.” Merrill-
Palmer Quarterly, 42: 200–227.
Herman, D. (2012). “Exploring the nexus of narrative and mind.” In D. Herman, J. Phelan,
P. J. Rabinowitz, B. Richardson, and R. Warhol (eds.), Narrative Theory: Core Concepts and Critical
Debates (pp. 10–14). Columbus, OH: Ohio State University Press.
Hoff-Ginsburg, E. (1991). “Mother–child conversations in different social classes and communicative
settings.” Child Development, 62: 782–96.
Hutto, D. D. (2007). “The narrative practice hypothesis: Origins and applications of folk psychology.”
Royal Institute of Philosophy Supplement, 60: 43–68.
Jack, F., MacDonald, S., Reese, E., and Hayne, H. (2009). “Maternal reminiscing style during early child-
hood predicts the age of adolescents’ earliest memories.” Child Development, 80: 496–505.
Kulkofsky, S., Wang, Q., and Koh, J. B. K. (2009). “Functions of memory sharing and mother–child remi-
niscing behaviors: Individual and cultural variations.” Journal of Cognition and Development, 10: 92–114.
doi:10.1080/15248370903041231.
Lagattuta, K. H. and Wellman, H. M. (2001). “Thinking about the past: Early knowledge about links
between prior experience, thinking, and emotion.” Child Development, 72(1): 82–102.
Laible, D. J. and Thompson, R. A. (2000). “Mother–child discourse, attachment security, shared positive
affect, and early conscience development.” Child Development, 71(5): 1424–40. doi:10.1111/1467-
8624.00237.
McAdams, D. P. (1990). “Unity and purpose in human lives: The emergence of identity as a life story.” In
A. I. Rabin, R. A. Zucker, R. A. Emmons, and S. Frank (eds.), Studying Persons and Lives (pp. 148–200).
New York: Springer.
McAdams, D. P. (1996). “Personality, modernity, and the storied self: A contemporary framework for
studying persons.” Psychological Inquiry, 7(4): 295–321.
279
McAdams, D. P. (2001). “The psychology of life stories.” Review of General Psychology, 5: 100–122. doi.
org/10.1037/1089-2680.5.2.100.
McAdams, D. P. (2008). “Personal narratives and the life story.” In J. O. Robins and L. A. Pervin (eds.),
Handbook of Personality: Theory and Research (3rd edn., pp. 241–61). New York: Guildford Press.
Main, M. (1995). “Recent studies in attachment: Overview, with selected implications for clinical work.”
In S. Goldberg, R. Nuir, and J. Kerr (eds.), Attachment Theory: Social, Developmental and Clinical
Perspectives (pp. 407–74). Hillsdale, NJ: Analytic Press.
Marsh, E. J. (2007). “Retelling is not the same as recalling implications for memory.” Current Directions in
Psychological Science, 16(1): 16–20.
Miller, P. J. (1994). “Narrative practices: Their role in socialization and self-construction.” In U. Neisser
and R. Fivush (eds.), The Remembering Self: Construction and Accuracy in the Life Narrative. New York:
Cambridge University Press, pp. 158–79. doi:10.1017/CBO9780511752858.010.
Nelson, K. (2003). “Narrative and self, myth and memory.” In R. Fivush and C. Haden (eds.), Connecting
Culture and Memory: The Social Construction of an Autobiographical Self. Hillsdale, NJ: Erlbaum.
Nelson, K. and Fivush, R. (2004). “The emergence of autobiographical memory: A social cultural devel-
opmental model.” Psychological Review, 111: 486–511.
Oppenheim, D., Koren-Karie, N., and Sagi-Schwartz, A. (2007). “Emotion dialogues between mothers
and children at 4.5 and 7.5 years: Relations with children’s attachment at 1 year.” Child Development,
78: 38–52.
Oppenheim, D., Nir, A., Warren, S., and Emde, R. N. (1997). “Emotion regulation in mother–child
narrative co-construction: Associations with children’s narratives and adaptation.” Developmental
Psychology, 33: 284–94.
Reese, E., Jack, F., and White, N. (2010a). “Origins of adolescents’ autobiographical memories.” Cognitive
Development, 25(4): 352–67.
Reese, E., Yan, C., Jack, F., and Hayne, H. (2010b). “Emerging identities: Narrative and self from early
childhood to early adolescence.” In K. C. McLean and M. Pasupathi (eds.), Narrative Development in
Adolescence: Creating the Storied Self (pp. 23–44). New York: Springer-Verlag.
Ricoeur, P. (1991). “Narrative identity.” Philosophy Today, 35(1): 73–81.
Rime, B. (2007). “The social sharing of emotion as an interface between individual and collective processes
in the construction of emotional climate.” Journal of Social Issues, 63: 307–22. doi:10.1111/j.1540-
4560.2007.00510.x.
Rubin, D. C. (2005). “A basic-systems approach to autobiographical memory.” Current Directions in
Psychological Science, 14(2): 79–83.
Sarbin, T. R. (1986). Narrative Psychology: The Storied Nature of Human Conduct. Westport. CT: Praeger
Publishers/Greenwood Publishing Group.
Schectman, M. (2003). “Empathic access: The missing ingredient in personal identity.” In R. Martin and
J. Barresi (eds.), Personal Identity (pp. 238–59). Oxford: Oxford University Press.
Shaver, P. R. and Mikulincer, M. (2002). “Attachment-related psychodynamics.” Attachment and Human
Development, 4: 133–61.
Sroufe, L. A. (2005). “Attachment and development: A prospective, longitudinal study from birth to
adulthood.” Attachment and Human Development, 7(4): 349–67.
Treboux, D., Crowell, J., and Waters, E. (2004). “When ‘new’ meets ‘old’: Configurations of adult
attachment representations and their implications for marital functioning.” Developmental Psychology,
40: 295–314.
Vygotsky, L. S. (1978). Mind in Society: The Development of Higher Psychological Processes. Cambridge, MA:
Harvard University Press.
Wang, Q. (2013). “The cultured self and remembering.” In P. J. Bauer and R. Fivush (eds), The Wiley
Handbook on the Development of Children’s Memory (pp. 605–25). New York: Wiley.
Zaman, W. and Fivush, R. (2011). “Intergenerational narratives and adolescents’ emotional well-being.”
Journal of Adolescence, 21: 703–16.
280
PART VII
The epistemology of memory

22
INTERNALISM AND
EXTERNALISM
Brent J. C. Madison
Introduction
We have countless beliefs, and many of them are justified. Most of our beliefs, at any given
time, are not occurrent, but are stored in memory. In addition, many of our beliefs remain
justified while stored in memory. How is this all possible? An epistemology of memory will
in part explain the nature of memorial justification, and how it is possible. Epistemic internal-
ists and externalists disagree deeply about the fundamental nature of epistemic properties like
epistemic justification. That is, is justification wholly determined by what goes on inside and
from the first-person perspective, or does how beliefs are caused, formed, and what relations
subjects bear to their environment prove relevant to whether or not justification obtains? This
chapter will first survey general issues in the epistemic internalism / externalism debate: what is
the distinction, what motivates it, and what arguments can be given on both sides. The second
part of the chapter will examine the internalism / externalism debate as regards to the specific
case of the epistemology of memory belief.
1. The internalism / externalism distinction: intuitive motivations

The internalism / externalism distinction in epistemology concerns the question of what kind of
factors can contribute to the positive epistemic status of a belief. While one can endorse versions
of internalism or externalism for a variety of epistemic statuses, such as epistemic rationality,
warrant, entitlement, etc., the focus of this chapter will specifically be on epistemic justification.
Justification is taken to be a supervenient property: whether a belief is justified is not a brute
fact, but depends upon some further conditions obtaining. Internalists maintain that all the
factors upon which justification depends are internal, either in the sense of being reflectively
accessible to the subject, or by being the agent’s mental states. Externalists deny this. The ver-
sions of internalism or externalism that epistemologists have tended to endorse have been shaped
by reactions to at least two kinds of central thought experiments: those that aim to show that
paradigmatically external factors, such as causal relations or the reliability of the method that
gave rise to the belief, are not sufficient for justification, and those that aim to show that such fac-
tors are not necessary for justification. Reactions to these cases have also influenced the positive
accounts of justification offered. The two kinds of thought experiment are as follows.
283
Brent J. C. Madison
Is reliability sufficient for justification?

Actual and possible cases of blindsight (e.g., Smithies 2014), serendipitous brain-tumors
(e.g., Plantinga 1993), chicken-sexers (e.g., Foley 1987: 168–701) and clairvoyance (e.g.,
Bonjour 1980, 1985) share this basic structure: unbeknownst to the subjects, they have highly
reliable belief-forming processes that produce and sustain true beliefs regarding some subject
matter. The subjects are not aware that they have this ability, and nor are they aware of how
they could have reliably true beliefs on that topic. In addition, the subjects are not aware of
any reasons to think that the belief is true. For example, take this classic description of such a
case by Laurence Bonjour:
Norman, under certain conditions that usually obtain, is a completely reliable clair-
voyant with respect to certain kinds of subject matter. He possesses no evidence or
reasons of any kind for or against the general possibility of such a cognitive power or
for or against the thesis that he possesses it. One day Norman comes to believe that the
President is in New York City, though he has no evidence either for or against this
belief. In fact the belief is true and results from his clairvoyant power under circum-
stances in which it is completely reliable.
(Bonjour 1985: 41)
One can then ask about such cases: are the beliefs of such subjects, for example, justified?
If one were to answer “yes,” then one is likely expressing externalist sympathies. One might
reason as follows: after all, the aim of belief is truth, and the beliefs in question are not acciden-
tally true, but are reliably produced. Therefore these beliefs have a lot going for them, from an
epistemic point of view, which implies that they are justified.
If one were to answer “no,” that such beliefs are not justified, then one will hold that justi-
fication must be more than a reliably produced belief. But what is missing in such cases? What
would need to be added to such cases to make them cases of justified belief? So-called epistemic
internalists tend to stress that what is missing is the involvement of the subject’s point of view:
the first-person perspective. According to a traditional sort of epistemic internalism, what needs
to be added is some kind of awareness on the subject’s part of something like reasons, grounds, or
evidence to think that the contents of their beliefs are true. More on what form that awareness
might take, and what one must be aware of, will be discussed below.
Is reliability necessary for justification?

The other kind of thought experiment that has tended to move people in the internalism /
externalism debate is versions of what has become known as the “New Evil Demon” prob-
lem. (For the original presentation of the New Evil Demon problem, see Cohen 1984; Lehrer
and Cohen 1983. For an overview, see Littlejohn 2009.) A common version of the New Evil
Demon problem is based on drawing new morals from an old thought experiment. Like the evil
genius that Descartes introduces in the context of discussing skepticism, the New Evil Demon
problem proceeds from a similar set-up. We are asked to consider a world where a powerful
evil demon (or in some variations, super scientists controlling brains in vats) is radically deceiv-
ing the inhabitants of that world in a systematic way such that their perceptual experiences are
largely hallucinatory, and their beliefs about the external world based on these experiences are
by and large false. Despite this, from the subjects’ own point of view, things seem exactly as they
would if their beliefs were by and large veridical. Unlike Descartes’, which concerns skepticism
284
Internalism and externalism
and the possibility of knowledge, the new question is an evaluative question about the presence
and nature of justification. That is, do we judge that the victims of the evil demon are justified
in believing as they do? If they are justified, do they share sameness of justification with their
non-deceived counterparts?
Traditional internalists have registered the judgment of sameness of justification, from
which one can conclude that external factors like the reliability of the process that gave
rise to a belief, or the truth of those beliefs, or whether or not one’s perceptual experiences
are veridical, are not necessary for justification, since the demon can ensure that the sub-
jects’ beliefs are utterly unreliable by, among other things, systematically ensuring that their
beliefs are always false. Internalists often draw a positive moral from such cases: whatever
else justification must be like, it is constrained by principles which hold that internally alike
counterparts must also be counterparts in justification: the same beliefs are justified for each
of them, and to the same extent.
Those with externalist sympathies have tended to judge that victims of the New Evil Demon
lack justification for their beliefs, or if they are justified, then to a lesser degree than their
normal-world counterparts. Some externalists concede that while the beliefs of the radically
deceived in the demon world might have other possible virtues, such as, e.g., being blamelessly
held, they contend that such beliefs lack justification (e.g., Littlejohn 2012; Pritchard 2012).
Depending on how one responds to the above two cases, one ends up with increasingly
internalist or externalist views about the nature of epistemic justification.
How should we understand the epistemically “internal”?

In response to the above test cases on whether factors like reliability are sufficient for justifica-
tion, some internalists have held that what is missing is any essential involvement of the subject’s
point of view. On one way of construing this requirement, what needs to be added is some kind
of (actual or potential) awareness on the subject’s part of something like reasons, grounds, or evi-
dence to think that the contents of their beliefs are true. Awareness is usually taken to be a form
of consciousness. But if an internalist does not demand actual awareness, they might still hold
that a subject must have the disposition to be consciously aware of the grounds of their belief, if
they reflected upon them. Intuitively, as an internalist might put it, what Norman needs to be
justified in believing that the President is in New York, is to be aware of some kind of reason
or evidence that counts in favor of believing that the President is in New York. For example,
Norman could see a news report to that effect, or some other kind of testimony. In the absence
of being aware of any such considerations, then from Norman’s point of view, his belief is com-
pletely groundless. There is dispute among internalists of what a subject must be consciously
aware of in order to hold a justified belief. Some options present themselves:2
a) Grounds: Whether one is justified in believing that p supervenes on facts which one is in
a position to be consciously aware of. In order for a fact to contribute to justification, its
content must be accessible to the agent. This position has been called “Simple Internalism”
(Pryor 2001) (and in another form, “Internalist Externalism” (Alston 1989a).
b) Adequacy of grounds: Access Internalism, by contrast, maintains that one always has “special
access” to one’s justificatory status (where “access” and “conscious awareness” are often
used interchangeably). So unlike simple internalism which only requires access to the
contents of one’s grounds, access internalism in addition insists that all of one’s justified
beliefs are such that not only are one’s grounds accessible, but also that the grounds are
285
Brent J. C. Madison
adequate (e.g., Bonjour 1985; Chisholm 1989). Simple Internalists do not affirm this lat-
ter claim. For example, an Access Internalist in this sense will insist that for Norman to
be justified in believing that the President is in NYC, it is necessary that he is aware of
some reason to think that the President is in NYC, and in addition, that it is necessary
that he is also able to determine through reflection alone (which is usually taken to be
restricted to introspection and a priori reasoning) that this reason justifies his belief about
the President’s whereabouts.
The other main way of understanding the epistemically internal has become known as
Mentalism. Mentalists do not stress the epistemological significance of conscious awareness.
Instead, they construe the epistemically internal as internal to the subject’s mind in the sense
of holding that epistemic justification depends wholly on the subject’s mental states. Ralph
Wedgwood (2002, forthcoming) offers a version of Mentalism about epistemic rationality: he
argues that what is rational to believe supervenes on one’s non-factive mental states.3 Richard
Feldman and Earl Conee, the chief advocates of Mentalism, define Mentalism as the thesis
that “a person’s beliefs are justified only by things that are internal to the person’s mental
life” (Feldman and Conee 2001: 233). Specifically, their official formulation of Mentalism is
as follows: “S: The justificatory status of a person’s doxastic attitudes strongly supervenes on
the person’s occurrent and dispositional mental states, events, and conditions” (Feldman and
Conee 2001: 234).
This statement of Mentalism, unlike Wedgwood’s, allows for factive mental states / events /
conditions to serve as justifiers. From this formulation, they express the main implication of S
as follows: “M: If any two possible individuals are exactly alike mentally, then they are alike
justificationally, e.g., the same beliefs are justified for them to the same extent” (Feldman and
Conee 2001: 234). They take it that this latter claim simply spells out a consequence of their
supervenience thesis S.
There are different ways of arguing for Mentalism. Wedgwood appeals to versions of the
New Evil Demon cases introduced above in arguing for Mentalism: if normal world subjects
and their radically deceived counterparts are equally justified, then what best explains this is that
they are mentally alike (Wedgwood 2002, forthcoming). Wedgwood suggests that it makes
no difference to which beliefs are rationally held whether or not the subject’s experiences are
veridical or whether the beliefs based on those experiences are true or false.
Feldman and Conee offer their view as the best explanation of intuitive judgments about
cases they provide (e.g., Feldman and Conee 2001). In defending Mentalism, pairs of cases
are introduced where Feldman and Conee invite the intuition that in the first instance the
subject has a justified belief and in the second case the belief is intuitively not justified, or
else one belief is more justified than the other. They also contend that the best explanation
of these apparent epistemic differences is that there are “internal” differences in their pre-
ferred sense of internal to the subject’s states of mind. Reflecting on these cases and those
like them, Feldman and Conee argue that epistemic internalism, understood as Mentalism, is
true: justification supervenes on the mental; there can be no justificatory difference without
a mental difference.
Having introduced the two broad families of thought experiment that tend to divide episte-
mologists in the internalism / externalism debate, we can now appreciate a broad spectrum of
possible views. Depending on whether external factors, such as the reliability of the process that
gives rise to the belief, are held to be necessary and / or sufficient for justification, the following
possibilities emerge:
286
Table 22.1 The reliability table
Reliability Reliability Possible view:

necessary? sufficient?
YES YES Strong Forms of Process Reliabilism: a belief is justified if and only if it is the
product of a reliable belief-forming process.
YES NO Weaker Form of Process Reliabilism: a belief is justified only if it is the
product of a reliable belief-forming process, and no undefeated defeaters are
present. However, no positive evidence in support of the belief is required
(e.g., Goldman 1979, 1986).
Forms of Evidential Externalism: Indicator Reliabilism / Internalist
Externalism: a belief is justified only if it is based on a truth-conducive
ground (e.g., Alston 1989a).
Epistemological Disjunctivism: a belief is justified only if it is based on a
factive ground, and this ground must be accessible to the subject upon
reflection (e.g., Pritchard 2008, 2011, 2012; and arguably McDowell 1998a,
1998b, 1998c).
NO YES No known advocates.
NO NO Strong Forms of Internalism
Mentalism: justification supervenes on the mental in that there can be no
difference in justification without a mental difference (e.g., Feldman and
Conee 2001; Wedgwood 2002).
Traditional Versions of Access Internalism: the conjunction of Mentalism +
an awareness requirement, such as one must have reflective access to
the adequacy of one’s grounds (e.g., Chisholm 1989; Bonjour 1985;
Steup 1999).
Internalist Coherentism: a belief is justified just in case it coheres with other
beliefs in one’s noetic structure. Authors vary on how to understand the
notion of coherence, but it is often taken to include some kind of explanatory
relation between doxastic states (e.g., Bonjour 1985; Lehrer 2000; Poston
2014). For more on the issue of coherentism in the epistemology of
memory, see Chapter 24, this volume.
Phenomenal Conservatism: a belief is justified only if it based on a seeming,
e.g., seeming that P provides prima facie justification to believe that P
(see Huemer 2006, 2007, 2013. For criticism see Littlejohn 2011; as well as
a number of papers in Tucker 2013).
So understood, Phenomenal Conservatism is a form of Mentalism, and
Phenomenal Conservatives can differ on whether the subject must be
aware of the seeming, or merely require the justified belief to be based on
the seeming. For more on the issue of Phenomenal Conservatism in the
epistemology of memory, see Chapter 23, this volume, on foundationalism.
Theoretical motivations
Besides the cases of clairvoyance and their ilk and New Evil Demon considerations, what other
kind of arguments can be given for and against different forms of internalism and externalism?
As part of a theoretical defense of internalism, one might appeal to considerations of epistemic
value. From the premise that epistemic justification is of value, one might argue that justifica-
tion has to have a certain nature to account for such value. For example, if part of the value of
justification comes from allowing a certain kind of rational defensibility, then this might require
287
Brent J. C. Madison
conscious awareness of grounds or reasons that can be appealed to in defense of one’s beliefs (for
the view that epistemic rationality is linked to a kind of idealized rational defensibility from the
first-person perspective, see Foley 1987, 2001).
On the other hand, externalists might also draw on considerations of what they take to be of
epistemic value in motivating their theory of justification. It is a widely held assumption among
philosophers—with very different accounts of the nature of justification—that what individuates
epistemic justification from other kinds of justification is a connection to truth. Truth is com-
monly taken to be a fundamental epistemic value. For example, reporting on the views of Alvin
Goldman, Richard Fumerton writes the following:
The fundamental idea behind Goldman’s reliabilism is straightforward enough. When

a belief is justified it has a virtue. There is something good about it. From the epistemic
perspective, virtue has to do with truth. The reason epistemologists want epistemically
justified beliefs . . . is that having justified beliefs has something to do with having true
beliefs. At the same time, we can understand justification in such a way that we allow
the possibility of justified false belief . . . The answer is to focus on the processes that
produce beliefs.
(Fumerton 1995: 97; original emphasis)
So an advantage of externalism is clear: externalists can give a straightforward account of the

relationship between justification and truth.
Internalists tend to have much more difficulty on this score. Given a commitment to the
New Evil Demon thesis, it is perfectly consistent with that form of internalism that a subject
could have fully justified but systematically false beliefs. However, a potentially devastating
problem now arises for any such account of epistemic justification: in what way is the internal-
ist’s alleged epistemic justification really epistemic, if it is consistent with having massively false
beliefs? After all, what of the general commitment that epistemic justification must admit of a
connection to the truth? A natural worry is that whatever form the truth-connection might
take, the possibility of massively false, but perfectly justified beliefs, is inconsistent with there
being a genuine and substantial connection between justification and truth. In short, the inter-
nalist owes us an account of the connection between justification and truth.
Another strategy taken by some internalists in motivating their view asserts that justifica-
tion is a deontological notion, and that meeting one’s epistemic duties and obligations requires
awareness of one’s justifiers (for discussion, see Plantinga 1993; Alston 1989b; see also the col-
lection of essays in Steup 2001). Relatedly, epistemic justification has been thought to be closely
related to notions of epistemic responsibility, and that such responsibility implies internalism
(e.g., Foley 2005). Finally, if one adopts a guidance conception of justification, in that justifica-
tion is supposed to guide one in what to believe and ultimately what to do, then one might
think that this implies that one must have access to one’s grounds if one is going to be able to
deliberate on what to believe (see Goldman 1980 for a discussion of the regulative vs. the theo-
retical conceptions of justification and its implications for the internalism / externalism debate).
An important obstacle for epistemic internalism, which is most often pressed against forms of
internalism that appeal to conscious awareness or other higher-level requirements, is the charge
that it entails various kinds of skepticism. For example, internalism has been thought to be too
demanding in various ways. Some have charged that internalism over-intellectualizes justifica-
tion in ways that imply that animals, small children, and the intellectually unsophisticated would
lack justification for their beliefs (e.g., Goldman 1986: 62; Alston 1989c: 164; Burge 2003). This
constitutes an objection on the assumption that animals, small children, and the intellectually
288
unsophisticated have the capacity for having beliefs that specifically enjoy epistemic justification
(instead of other epistemic statuses, such as knowledge, warrant, etc.). An externalist might draw
on cases of animals, small children, etc. as the basis of positive arguments in support of their
preferred externalist epistemology (e.g., Kornblith 2002 argues that the study of the science of
animal behavior supports an externalist theory of knowledge and justification). Both sides of the
debate take it as a constraint on an adequate theory of epistemic justification that the account
advanced needs to be psychologically plausible, so the theory of justification advanced should
not presuppose abilities or capacities that normal epistemic agents do not possess.
A related family of objections holds that internalism entails skepticism in another way. In
particular, a common objection is that access internalism specifically generates vicious regresses.
If a justified belief requires awareness of a reason, does that awareness itself need to be justified?
If so, does that not require a further state of awareness, which in turn calls out for justification?
For example, Michael Bergmann (2006) argues that the awareness requirement that some hold
defines epistemic internalism generates a dilemma: either it leads to both a vicious regress of
token mental states, as well as a vicious regress of mental states with increasingly complex, and
hence unthinkable, content, or else it is entirely unmotivated. As a result, Bergmann argues,
epistemic internalism, the awareness requirement, and the intuitions that motivate it, must be
abandoned, despite any initial plausibility. For some responses to Bergmann’s Dilemma, see
Rogers and Matheson (2011) and Moretti and Piazza (2015).
A final general worry about epistemic internalism is that it cannot provide a satisfactory
answer to the problem of radical skepticism. Radical skepticism says that knowledge of the
external world is impossible, but pre-theoretically, we have the deeply rooted common-sense
intuition that we have such knowledge. Skepticism can be seen as a paradox, since seemingly
good arguments can be given from seemingly true premises for the seemingly unpalatable con-
clusion that knowledge is impossible. One might generalize these arguments to target not only
knowledge, but also epistemic justification. John Greco, for one, has argued that given the
form of skeptical arguments, the problem of radical skepticism can only be solved by appeal to
epistemic externalism (Greco 2000). Greco argues that it is the internalist’s insistence on inter-
nally accessible grounds for their beliefs that give traditional skeptical arguments their force. He
argues that we must adopt a form of epistemic externalism if we are going to be able to solve
the problem of radical skepticism.
Depending on one’s views of the methodological role of skepticism in epistemological theo-
rizing, one might follow Greco and take the issue of responding to skepticism as a general
objection against internalism. As an argument against internalism is often therefore an argu-
ment in favor of externalism, appealing to the problem of skepticism can be used directly as an
argument in favor of one’s preferred brand of externalism. On the other hand, one might take
the relative ease in which externalist views handle the problem of skepticism to count against
externalism. The argument would be that philosophical skepticism is a deep and important
problem, but externalist responses trivialize it, and as such should be rejected (e.g., Stroud 1994;
Fumerton 1990). For more on the general issue of skepticism as applied to memory, see Chapter
26 in this volume by Andrew Moon.4
2. The justification of memory belief

With a taxonomy of some of the possible positions in the epistemic internalism / externalism
debate, as well as a survey of some of the motivations for and against such accounts, we can
now turn to the specific issue of the justification of beliefs stored in memory. Our primary focus
here shall be on what has been called propositional, factual, or semantic memory (for discussion
289
Brent J. C. Madison
of kinds of memory, see Chapter 1 in this volume by Markus Werning and Sen Cheng). Here
what one remembers is a fact or a proposition: for example, I remember that my car broke down
in the desert (for discussion of the objects of memory, see Chapter 7 in this volume by Jordi
Fernández). While I can consciously entertain my belief that my car broke down in the desert,
most of the time such beliefs are retained in memory. Common sense assures us that such beliefs
are justified not only when they are occurrent, but also when stored in memory. But what
accounts for the justification of memory belief?5
Internalist accounts of memorial justification

Epistemic internalists who insist on a reflective accessibility requirement on justification might
begin by appealing to what has been called episodic or experiential memory. Unlike propo-
sitional memory where what one remembers is a fact, the object of episodic memory is an
experience. For example, if one’s car broke down in the desert last weekend, one might be able
to remember the breakdown: one is able to call to mind the experience of seeing or being in the
car, an experience that might be accompanied by sensuous imagery. If an internalist holds that
conscious perceptual experiences justify belief, here memory retains both the belief and its jus-
tifying experiential ground. Here one is able to recall the experience of the breakdown, which
in turn might be thought to justify one’s belief that the car broke down.
Such an approach might fit most naturally within an Evidentialist framework. In its most
general form, Feldman and Conee, champions of both Mentalism and Evidentialism, define
Evidentialism as follows: “ES: The epistemic justification of anyone’s doxastic attitude toward
any proposition at any time strongly supervenes on the evidence that the person has at the time”
(Conee and Feldman 2004: 101).
As perceptual experience is normally understood by Evidentialists as evidence, they might
similarly hold that episodic memory can serve as evidence for the propositional memory
based upon it.
It has been widely noted, however, that as time passes and memory fades, often subjects
retain their beliefs, but they lose the experiential or episodic grounds on which they were based.
Perhaps most of our memory beliefs are like this; I have a dizzyingly wide array of factual beliefs
stored in memory, but if pressed, I often cannot recall how, when, or on what basis I formed
these beliefs. On pain of skepticism, we nevertheless take at least most of these beliefs as epis-
temically justified. For example, I am quite sure both that Napoleon was born in Corsica, and
that if I reflect on why I think that is true, there is very little, if anything I could call to mind
in its support. My belief that Napoleon was born in Corsica is justified nonetheless. Epistemic
internalists face the problem of explaining how such justification is possible. The so-called
Problem of Forgotten Evidence is one of the major challenges facing internalists (for recent sam-
ple presentations of the problem, see Goldman 2001; Williamson 2007; Bernecker 2010: 72).
At this point, internalists within the Evidentialist tradition might broaden their conception of
what evidence is available in support of our beliefs. Recall that Phenomenal Conservatism holds
that, roughly, a seeming that P confers prima facie justification to believe that P. Some have
argued that in cases of forgotten evidence, one still often seems to remember that P (Feldman and
Conee 2001; Fumerton 1995; Pollock 1986; Madison 2014). To return to our earlier example,
while I might not be able to recall any direct ground for my belief that Napoleon was born
in Corsica, I might seem to remember that I learned this fact. Seeming to remember having
learned something might be thought of as some evidence that one did indeed learn it. Even if I
cannot seem to remember that I learned a particular fact, I often have the experience of seeming
to remember it when it is called to mind. Seeming to remember that P has a distinctive feel; it
290
is not like wishing, hoping, or seeing that P (for discussion of the phenomenology of memory,
see Chapter 2 in this volume by Fabrice Teroni).
A worry for phenomenal conservative responses to the Problem of Forgotten Evidence is
that this approach has the consequence that forgetting one’s evidence, which seems like a kind
of epistemic shortcoming, can actually improve one’s epistemic position in objectionable ways
(see e.g., Huemer 1999: sec. 2; Senor 1993). Suppose that one’s original ground for one’s belief
that Napoleon surrendered at Waterloo was that one heard it in an Abba song. At that time,
one’s belief was unjustified. Over time, however, suppose that one forgets that one acquired this
belief in this way. Even though one has forgotten one’s original basis, suppose that one now has
the experience of seeming to remember that the belief is true. Does merely seeming to remem-
ber that Napoleon surrendered at Waterloo now justify that belief?
Internalists of some stripes will argue yes: just as defeating defeaters can positively affect jus-
tification, losing defeaters by forgetting them can have the same positive effect (e.g., Madison
2014). On the other hand, in response to this kind of case, many externalists will argue that
memory cannot improve, or especially generate justification, where formerly there was none.
For more on the general issue of whether memory can generate positive epistemic status,
or whether it merely preserves it, see Chapter 25 in this volume by Thomas D. Senor, on
Preservationism vs. Generativism in the epistemology of memory.
A closely related worry has been called the “epistemic boost” problem (e.g., Huemer 1999;
McGrath 2007 discusses a version of this problem for conservative approaches in epistemol-
ogy more generally). The worry is another way of expressing the concern that Phenomenal
Conservative approaches to memory would allow the counterintuitive result that the process
of remembering can “boost” the justification a belief had, over and above the justification one
originally had for it. The worry is this: one might think it implausible that each and every time a
belief is retrieved from memory it receives an extra epistemic boost due to the epistemic import
of the experience of seeming-to-remember, over and above the belief’s initial good grounds.
Sven Bernecker expresses the objection thus:
Suppose that S initially comes to believe that P by means of an a priori proof. The next
day S still remembers P and the proof of it. But since he also has the experience of
seeming to remember that P, he now has two reasons for holding P true, an inferential
and a foundational one. Thus S has more justification for P now than he had at the
original learning.
Even if one does not allow that memory can generate justification where previously there was
none, the question is now whether memory can enhance or raise justification, or whether the
epistemic role of memory is purely preservative. For a possible response to this problem in terms
of explaining away the potential oddness by appealing to different senses in which justification
can be increased, see Madison (2014: 50–52) (for a sample of two other possible responses to the
epistemic boost problem, see McGrath 2007: 20–21).
Externalist accounts of memorial justification

With one’s preferred externalist theory of epistemic justification in hand, its application to the
case of memory belief is quite straightforward. Take a case of forgotten evidence, such as one’s
belief that Napoleon was born in Corsica. This kind of case poses the greatest difficulties for
internalist theories of justification. But these cases on the face of it can be handled quite easily
291
Brent J. C. Madison
by externalist epistemologies. To see this, suppose that one learned the fact that Napoleon was
born in Corsica from a highly reliable textbook as a child, with the help of an equally reliable
teacher. But as decades have passed, one can no longer call to mind how one formed this belief,
or any reason to think it true. We might judge that the belief may well be epistemically justified.
How is this to be explained?
To take one form of epistemic externalism, the process reliabilist can hold that a belief is
justified just in case it is the product of a reliable belief-forming method (e.g., Goldman 1979,
1986). If the textbook and competent teacher are part of a reliable method in the right kind of
way, then one’s belief about the birthplace of France’s first emperor is justified, regardless of
whether or not one now has access to one’s original grounds.
Or if an externalist rejects simple forms of process reliabilism, they might opt for a form of
Virtue Epistemology, and generalize it to the case of memorial justification. For example, one
might hold that justification arises as a result of the operation of reliable epistemic virtues or
cognitive faculties (for a survey on various forms of virtue epistemology, see e.g., Battaly 2008).
So assuming that the faculty of memory is by and large reliable on matters like this, and assum-
ing it is virtuously deployed when one recalls that Napoleon was born in Corsica, then one
might argue that that belief thereby enjoys epistemic justification. The application of different
externalist accounts to the case of memorial justification will share the virtues and vices of such
approaches generally.
In addition to specific externalist accounts of epistemic justification applied to the case of
memory belief, it is worth noting a family of externalist views particular to the case of mem-
ory that have been called Preservationism. According to such views, just as memory preserves
beliefs, memory also preserves whatever justification a subject originally had for those beliefs.
So, for example, if a belief was formed by a reliable belief-forming process, one might argue
that, in the absence of defeaters, that epistemic goodness can be preserved by memory, even
if one cannot now recall on what basis this belief was held (if one ever could). For defenses of
Preservationism in the epistemology of memory, see Annis (1980); Malcolm (1963); Naylor
(1983); Burge (1993); Owens (2000).
New Evil Demon-style cases can be presented against Preservationism, and so it is incom-
patible with those forms of internalism that embrace the New Evil Demon intuition. Take a
subject S and her recently envatted counterpart S*; both seem to remember that P, and on
that basis believe that P. Suppose however that one subject is enjoying a veridical memory
experience, whereas the other is merely seeming to remember that P, but this experience is
entirely illusory thanks to the demon’s intervention. Do the two subjects share sameness of
justification? Internalists with their commitment to the New Evil Demon case will maintain
yes, and so conclude that memory justification cannot just be a matter of preserving whatever
justification the subject originally had. Michael Huemer makes a similar point in terms of
Bertrand Russell’s well-known five-minute hypothesis, i.e., the hypothesis that the world was
created five minutes ago, replete with all of one’s apparent memories (1999: 350). Internalists
will hold that the subjects of the five-minute hypothesis will be just as justified in beliefs as
their subjectively indistinguishable counterparts who enjoy a genuine past. Whether this con-
stitutes an objection to Preservationism will depend on how one responds to the New Evil
Demon problem more generally.
While externalists might have more straightforward ways of accounting for the justifica-
tion of memory belief, hopefully it has been shown that internalists are not without resources
to respond to this challenge. What still remains at an impasse, however, is how to resolve
the deep general debate between internalists and externalists on the fundamental nature of
epistemic justification.
292
Acknowledgment
Thanks to Sven Bernecker, Andrew Moon, and especially Rhiannon James for helpful written
comments on earlier drafts of this chapter.
Notes
1 Epistemologists often refer to the chicken-sexer case, but almost never attribute it to anyone. To the best
of my knowledge, the earliest mention of the case is Goldman (1975). See especially pp. 114–16. Very
interestingly given his later process reliabilist views (e.g., 1979), Goldman (1975) argues that the case is
one of knowledge, but no justification.
2 See Madison (2010) for these options and others on how exactly to construe the internalist’s awareness
requirement.
3 A factive state is one that entails the truth of its content. For example, most philosophers hold that
knowledge is a factive state: if one knows that P, then P is true. Non-factive states are those whose content
can be false. For example, belief is not factive: one can believe that P, even if P is false. In response to New
Evil Demon cases, most Mentalists hold that justification is determined by kinds of mental states that can
have false content.
4 For further general reading on the internalism / externalism debate, see Kornblith (2001); Pappas (2005);
Poston (2008); Madison (2010). In addition, an up-to-date bibliography on the epistemic internalism /
externalism distinction is maintained on PhilPapers at http://philpapers.org/browse/epistemic-internalism-
and-externalism.
5 For further general reading on the epistemology of memory, see Senor (2014); Frise (2015).
References
Alston, William (1989a) “An Internalist Externalism,” in Epistemic Justification. Ithaca, NY: Cornell
Alston, William (1989b) “The Deontological Conception of Epistemic Justification,” in Epistemic
Justification. Ithaca, NY: Cornell University Press, pp. 115–52.
Alston, William (1989c) “Level Confusions in Epistemology,” in Epistemic Justification. Ithaca, NY: Cornell
Annis, David B. (1980) “Memory and Justification,” Philosophy and Phenomenological Research 40: 324–33.
Battaly, Heather (2008) “Virtue Epistemology,” Philosophy Compass 3: 639–63.
Bergmann, Michael (2006) Justification Without Awareness. Oxford: Clarendon Press.
Bernecker, Sven (2008) The Metaphysics of Memory. New York: Springer.
Bernecker, Sven (2010) Memory: A Philosophical Study. Oxford: Oxford University Press.
Bonjour, Laurence (1980) “Externalist Theories of Empirical Knowledge,” Midwest Studies in Philosophy
5: 53–73.
Bonjour, Laurence (1985) The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press.
Burge, Tyler (1993) “Content Preservation,” The Philosophical Review 102: 457–88.
Burge, Tyler (2003) “Perceptual Entitlement,” Philosophy and Phenomenological Research 67: 503–48.
Chisholm, Roderick (1989) Theory of Knowledge. 3rd edn. Englewood, NJ: Prentice-Hall.
Cohen, Stewart (1984) “Justification and Truth,” Philosophical Studies 46: 279–95.
Conee, Earl and Feldman, Richard (2004) Evidentialism. Oxford: Oxford University Press.
Feldman, Richard and Conee, Earl (2001) “Internalism Defended,” in Epistemology: Internalism and
Externalism, H. Kornblith (ed.). Oxford: Blackwell Publishing, pp. 231–60.
Foley, Richard (1987) The Theory of Epistemic Rationality. Cambridge, MA: Harvard University Press.
Foley, Richard (2001) Working Without a Net. Oxford: Oxford University Press.
Foley, Richard (2005) “Justified Belief as Responsible Belief,” in Contemporary Debates in Epistemology,
M. Steup and E. Sosa (eds.). Oxford: Blackwell Publishing.
Frise, Matthew (2015) “Epistemology of Memory,” in The Internet Encyclopedia of Philosophy, James Fieser
and Bradley Dowden (eds.). www.iep.utm.edu/epis-mem/.
Fumerton, Richard (1990) “Metaepistemology and Skepticism,” Doubting: Contemporary Perspectives on
Skepticism, M. D. Roth and G. Ross (eds.). Dordrecht: Kluwer Academic Publishers, pp. 57–68.
Fumerton, Richard (1995) Metaepistemology and Skepticism. Boston, MA: Rowman and Littlefield.
293
Brent J. C. Madison
Goldman, Alvin (1975) “Innate Knowledge,” in Innate Ideas, Stephen P. Stich (ed.). Berkeley, CA:
University of California Press, pp. 111–20.
Goldman, Alvin (1979) “What Is Justified Belief?” in Justification and Knowledge, G. Pappas (ed.). Dordrecht:
D. Reidel.
Goldman, Alvin (1980) “The Internalist Conception of Justification,” Midwest Studies in Philosophy 5:
27–51.
Goldman, Alvin (1986) Epistemology and Cognition. Cambridge, MA: Harvard University Press.
Goldman, Alvin (2001) “Internalism Exposed,” reprinted in Epistemology: Internalism and Externalism,
Hilary Kornblith (ed.). Oxford: Blackwell Publishing.
Greco, John (2000) Putting Skeptics in Their Place. Cambridge: Cambridge University Press.
Huemer, Michael (1999) “The Problem of Memory Knowledge,” Pacific Philosophical Quarterly 80: 346–57.
Huemer, Michael (2006) “Phenomenal Conservatism and the Internalist Intuition,” American Philosophical
Quarterly 43(2): 147–58.
Huemer, Michael (2007) “Compassionate Phenomenal Conservatism,” Philosophy and Phenomenological
Research 74(1): 30–55.
Huemer, Michael (2013) “Phenomenal Conservatism,” in The Internet Encyclopedia of Philosophy, www.iep.
utm.edu/phen-con/.
Kornblith, Hilary (ed.). (2001) Epistemology: Internalism and Externalism. Oxford: Blackwell.
Kornblith, Hilary (2002) Knowledge and Its Place in Nature. Oxford: Oxford University Press.
Lehrer, Keith (2000) Theory of Knowledge, 2nd edn. Boulder, CO: Westview Press.
Lehrer, K. and S. Cohen (1983) “Justification, Truth, and Coherence,” Synthese 55(2): 191–207.
Littlejohn, Clayton (2009) “The New Evil Demon Problem,” The Internet Encyclopedia of Philosophy, www.
iep.utm.edu/evil-new/.
Littlejohn, Clayton (2011) “Defeating Phenomenal Conservatism,” Analytic Philosophy 52: 35–48.
Littlejohn, Clayton (2012) Justification and the Truth-Connection. Cambridge: Cambridge University Press.
McDowell, John (1998a) “Criteria, Defeasibility, and Knowledge,” in Meaning, Knowledge, and Reality,
John McDowell (ed.). Cambridge, MA: Harvard University Press, pp. 369–94.
McDowell, John (1998b) “Knowledge by Hearsay,” in Meaning, Knowledge, and Reality, John McDowell
(ed.). Cambridge, MA: Harvard University Press, pp. 414–44.
McDowell, John (1998c) “Knowledge and the Internal,” in Meaning, Knowledge and Reality, John McDowell
(ed.). Cambridge, MA: Harvard University Press, pp. 395–413.
McGrath, Matthew (2007) “Memory and Epistemic Conservatism,” Synthese 157: 19–22.
Madison, B.J.C. (2010) “Epistemic Internalism,” Philosophy Compass 5(10): 840–53.
Madison, B.J.C. (2014) “Epistemic Internalism, Justification, and Memory,” Logos & Episteme V(1): 33–62.
Malcolm, Norman (1963) Knowledge and Certainty. Englewood Cliffs, NJ: Prentice-Hall.
Moretti, Luca and Piazza, Tommaso (2015) “Phenomenal Conservatism and Bergmann’s Dilemma,”
Erkenntnis 80: 1271–90.
Naylor, Andrew (1983) “Justification in Memory Knowledge,” Synthese 55: 269–86.
Owens, David (2000) Reason Without Freedom. London: Routledge.
Pappas, George (2005) “Internalist and Externalist Theories of Justification,” Stanford Encyclopedia of
Philosophy. http://plato.stanford.edu/entries/justep-intext/.
Plantinga, Alvin (1993) Warrant: The Current Debate. Oxford: Oxford University Press.
Pollock, John (1986) Contemporary Theories of Knowledge. Lanham, MD: Rowman and Littlefield.
Poston, Ted (2008) “Internalism and Externalism in Epistemology,” The Internet Encyclopedia of Philosophy.
www.iep.utm.edu/int-ext/.
Poston, Ted (2014) Reason and Explanation: A Defense of Explanatory Coherentism. Basingstoke: Palgrave
Macmillan.
Pritchard, Duncan (2008) “McDowellian Neo-Mooreanism,” in Disjunctivism: Perception, Action, Knowledge,
Adrian Haddock and Fiona Macpherson (eds.). Oxford: Oxford University Press, pp. 283–310.
Pritchard, Duncan (2011) “Evidentialism, Internalism, Disjunctivism,” in Evidentialism and Its Discontents,
Trent Dougherty (ed.). Oxford: Oxford University Press, pp. 235–53.
Pritchard, Duncan (2012) Epistemological Disjunctivism. Oxford: Oxford University Press.
Pryor, James (2001) “Highlights of Recent Epistemology,” British Journal for the Philosophy of Science 52:
95–124.
Rogers, Jason and Matheson, Jonathan (2011) “Bergmann’s Dilemma: Exit Strategies for Internalists,”
Philosophical Studies 152: 55–80.
294
Senor, Thomas D. (1993) “Internalistic Foundationalism and the Justification of Memory Belief,” Synthese
94: 453–76.
Senor, Thomas D. (2014) “Epistemological Problems of Memory,” The Stanford Encyclopedia of Philosophy
(Fall 2014 edn.), Edward N. Zalta (ed.). http://plato.stanford.edu/archives/fall2014/entries/memory-
episprob/.
Smithies, Declan (2014) “The Phenomenal Basis of Epistemic Justification,” in New Waves in Philosophy of
Mind, M. Sprevak and J. Kallestrup (eds.). London: Palgrave Macmillan, pp. 98–124.
Steup, Matthias (1999) “A Defense of Internalism,” in The Theory of Knowledge: Classical and Contemporary
Readings, 2nd edn. M. Steup (ed.). Belmont, CA: Wadsworth, pp. 373–84.
Steup, Matthias (ed.). (2001) Knowledge, Truth, and Duty. New York: Oxford University Press.
Stroud, Barry (1994) “Skepticism, ‘Externalism’, and the Goal of Epistemology,” Proceedings of the
Aristotelian Society (supplementary vol.) 68: 290–307.
Tucker, Chris (ed.). (2013) Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism.
Wedgwood, Ralph (2002) “Internalism Explained,” Philosophy and Phenomenological Research 65: 349–69.
Wedgwood, Ralph (forthcoming) “Internalism Re-explained,” in The New Evil Demon, Julian Dutant
and Fabian Dorsch (eds.). Oxford: Oxford University Press.
Williamson, Timothy (2007) “On Being Justified in One’s Head,” in Rationality and the Good, Mark
Timmons, John Greco, Alfred R. Mele (eds.). Oxford: Oxford University Press.
295
23
FOUNDATIONALISM
Berit Brogaard
1. Introduction
Memory has been a mystery to philosophers for millennia. Part of the reason for this is that they
seem to encompass very disparate mental states. You might remember: how to ride a bicycle,
that Bill Clinton was impeached by the US House of Representatives but acquitted by the
Senate, that 119 is a composite number, that your wife forgot your anniversary two years ago,
that you were feeling joyous and elated while dancing salsa at Ball and Chain last Saturday night.
These mental states that we refer to as ‘memories’ seem to be intrinsically different. Your mem-
ory that 119 is a composite number seems more like a belief, whereas your vivid memory of
feeling joyous and elated while dancing salsa seems more like an experience or an affective state.
Nowadays, it’s common to recognize that these states really are very different. They are
encoded and retrieved differently in the brain, they have a different phenomenology, and they
have different types of contents (McKoon et al. 1986). It is not hard to see why philosophers
of the past felt that providing a unified philosophical account of memory was a tall order.
Presumably no unified account can be given (see Chapter 1, this volume).
Cognitive psychologists commonly distinguish among procedural memory, semantic memory
and episodic memory.1 Procedural memory is memory of how to do things, such as how to ride a
bicycle, semantic memory is memory of propositions or facts and episodic memory is memory of
yourself partaking in an event. Episodic memory can include laboratory tasks, such as remember-
ing long lists of words as an experiment. A particular subclass of episodic memories are known
as ‘autobiographical memories’. They refer to a system depending on both episodic and semantic
memory, such as my memory of your wearing a yellow shirt at breakfast on 15 March 2015.
This memory contains an episodic memory, viz. an episodic memory of the breakfast, which you
remember from a distinctly first-person perspective and a semantic memory, viz. that the breakfast
took place on 15 March 2015. This chapter will focus exclusively on autobiographical memories.
For simplicity’s sake, I shall often simply refer to these as ‘memories’.
One of the big debates in philosophy concerning autobiographical memories is that of
whether they can provide us with direct awareness of the past or whether they are represen-
tations, or constructions, of the past. Philosophers have had a special interest in this question
because on the one hand it seems that our memories only rarely, if ever, could provide us with
direct awareness of the past; on the other hand, if they are massively mistaken and don’t track
296
Foundationalism
the truth at all, how could our memories ever be justified? If our memories of the past are
constructions, they are unlikely to reveal what actually happened, and it’s not hard to see how
realizing this fact might lead to the radically sceptical view that we can never be justified in
believing what happened in the past (Bernecker 2008, 2010).
In recent years, psychologists have gathered numerous pieces of evidence for the view that
our memories are in fact constructions of what happened in the past (Eichenbaum and Cohen
2001). We don’t experience something and then store it in the brain as a unit that can be taken
out and put on display when convenient. Our memories are stored in the brain in a distributed
pattern in the outer layer of the cortex. As a rule, experiences are stored in the area of the brain
that initially processed them (Squire 1992; Danker and Anderson 2010). So, a visual aspect of
an experience is normally stored in the visual cortex, an auditory aspect of an experience is
stored in the auditory cortex and a motion element of an experience is stored in the sensory-
motor cortex. When we retrieve a memory, we need to put all those pieces and fragments back
together again, and empirical studies also reveal that we rarely put them back together again to
form a whole that is exactly like the initial experience (Schacter 1989).
Although there is no one-one correlation between brain processing and the nature of mental
states, the empirical data strongly suggest that memories are representations, and indeed repre-
sentations in a strong constructivist sense. They are models of what happened in the past, and
often not very good models.2
Although one can be justified in holding a false belief, it is a prima facie attractive view that
one cannot be justified in holding a type of belief that by its very nature is unlikely to track
the truth (e.g. beliefs based on imagined clairvoyance). This raises the question of how we can
ever come to have justified beliefs about the past. A hardcore externalist might dismiss this
question as a hard one (see Chapter 22, this volume). A reliabilist, for instance, can maintain
that when our memory retrieval processes function optimally, we know what happened in the
past, even if we don’t (internally) know that we know this. But what is the internalist supposed
to say? Here I will argue that some of our memory seemings have a special property that quali-
fies them as immediate and full justifiers of our memories, regardless of how wildly mistaken
our brain was when putting the stored fragments back together. This view is also known as
‘phenomenal dogmatism’. Phenomenal dogmatism says that a seeming can immediately justify
a mental state (e.g. a belief), if the agent has no mental states that defeat the justification. The
view has been defended by numerous authors for visual experience (Tolhurst 1998; Pryor 2000,
2005; Huemer 2007; Tucker 2010; Brogaard 2013a; Chudnoff 2013), higher-level perception
(Brogaard forthcoming a, forthcoming b) and memory (Audi 1995: 37; Huemer 1999; Conee
and Feldman 2004; Ginet 1975; Pollock 1974: 193; and, with reservations, Schroer 2008).
2. Phenomenal dogmatism for memory

Suppose you have a memory as of dancing salsa with your boyfriend last Saturday at the Latin-
American club Ball and Chain. It’s a nice memory. Next time you talk to your boyfriend you
say ‘remember how much fun we had dancing last Saturday?’ With a big smile he replies: ‘yeah,
I remember.’ In ordinary language, ‘I remember that p’ is typically factive. This means that it
entails that p. So, if you say ‘I remember that p’, you are also saying ‘p’. This does not mean that
you are right, only that you are saying that you are right. Here I am not going to use ‘remem-
ber’ the way it is typically used in ordinary language. When I say ‘you remember dancing salsa’,
I shall use that to mean that you have a memory as of dancing salsa. This (autobiographical)
memory may be very similar to a mental image, a vivid imagination, or a vivid dream, but it
297
Berit Brogaard
need not be. You might have a (autobiographical) memory that feels more like a thought or
a belief (as in ‘I believe I danced salsa last Saturday’). In addition to the notion of a (retrieved)
memory I shall introduce another notion relating to retrieved memories, which I shall call a
‘memory seeming’, ‘memory appearance’, or ‘memory image’ (see Chapter 11, this volume). I
take memory seemings to be experience-like in ways I will detail below. A memory seeming
just is a memory of the kind that is like a mental image, a vivid imagination, or a vivid dream.
But it could (e.g. over time) lead to a memory that is not a memory seeming – one of those
memories that are thought- or belief-like.
Given these distinctions we can articulate phenomenal dogmatism with respect to memory
as follows:
If it (memory-wise) appears to S as if p, then, in the absence of defeaters, S thereby has

immediately and full justification for her memory belief that p.
Phenomenal dogmatism holds that memory seemings or appearances can confer immediate
and full justification on memories. The memories that are justified in this way are founda-
tional, that is, they do not depend on any other memories, beliefs, or background information
for their justification. Phenomenal dogmatism is thus a version of foundationalism in episte-
mology, which is the view that some mental states (e.g. beliefs or experiences) are properly
basic and that (the rest of) our beliefs are justified in virtue of receiving proper support from
the basic mental states. This form of foundationalism can be differentiated from the view that
every memory belief is prima facie justified simply in virtue of being a memory belief (Audi
1998: 68–9).
As noted above, we can refer to memories that are phenomenally like a mental image, a vivid
imagination, or a vivid dream as memory seemings. For memories of this sort, phenomenal
dogmatism can be articulated as follows:
If it (memory-wise) appears to S as if p, then, in the absence of defeaters, that memory

appearance that p is immediately and fully justified.
Having a memory (or memory seeming) then may simply all by itself be sufficient for that
memory (or memory seeming) to be justified, depending on its phenomenology.
Phenomenal dogmatism is consistent with the possibility that memory can be the source
of new justification for belief, a view also known as ‘generativism’ (Bernecker, 2010). For
example, you may have a number of stored memory images that would provide justification
for a new belief, were they retrieved together as a single memory seeming. For instance, I may
have a memory seeming of you being super-sensitive to criticism at time t1, a second memory
seeming of you bragging about your accomplishments at time t2, and a third memory seem-
ing of you belittling your peers at time t3. The three memory seemings, if retrieved together,
may form a single memory seeming that confers prima facie justification on a memory belief
about your personality, for instance, the memory belief that you have a narcissistic personality
(Brogaard forthcoming a).
The view defended here is a radical form of internalism with respect to justification for
memory. But it is not as radical as it could be. Let’s distinguish among three views: strong access
internalism, weak access internalism, and mentalism. Strong access internalism is the view that
if you are justified in remembering that p, then you can tell by introspection alone that you
are justified in remembering that p. Weak access internalism is the view that if you are justified
298
Foundationalism
in remembering that p, then you have conscious access to the full justifier of your memory.
Mentalism, finally, is the view that what justifies your retrieved memories is a mental state but
you need not have conscious access to all parts of the mental state. As we will see, the view I am
defending here is a version of weak access internalism.
3. Which seemings can serve as justifiers?

On my view, memory seemings can serve as immediate justifiers either of belief-like memo-
ries or of themselves. The hard question that needs to be addressed is what it is about memory
seemings that make them justifiers. In ordinary language, we speak of seemings, looks, and
appearances in a number of different ways. Consider the following statements:
1(a) I just realized that Lisa looks exactly like her sister.
1(b) I can’t quite put my finger on what’s wrong with it, but the first premise of the argument
just seems wrong.
1(c) I just heard on the radio that a hurricane is coming our way. It seems to me that it would
be wise to evacuate.
1(d) That looks like a dog behind the tree.
1(e) It still seems to me that we were dancing salsa that night, even though I know we didn’t.
These are some of the ways that we can express seemings, looks and appearances in ordinary
language. Not all of these expressions stand in a one-one correspondence to mental states.
1(a), for example, compares Lisa to her sister. On the most plausible reading, it states that
Lisa’s physical appearance is very similar to the physical appearance of her sister. When you
make this statement, it is likely based on your visual experience (either current or past), but
the statement itself does not make it clear how things appear to you. It merely makes it evi-
dent that there are certain similarities, in your view, between Lisa and her sister. Roderick
Chisholm (1957) called this sort of use of ‘appear’ words ‘comparative’.
1(b) and 1(c) appear to express degrees of belief rather than how things sensorily appear to
you. On a likely reading of 1(b), you are stating that you have a high degree of belief that the
first premise of the argument is wrong, even though you cannot quite explain why you have
that belief. And on a plausible reading of 1(c), you are stating that you have a high degree of
belief that you should evacuate the area, given what you heard on the radio. Chisholm (1957)
called this use of ‘appear’ words ‘epistemic’.
The use of ‘look’ and ‘seem’ in 1(d) and 1(e) is non-comparative and non-epistemic.
Following Frank Jackson (1977), let’s call this use of ‘appear’ words ‘phenomenal’. It’s the
phenomenal uses of ‘appear’ words that correspond to the sorts of mental states phenomenal
dogmatists think can serve as immediate justifiers. But if they are right, what exactly is it about
them that makes them suitable for this role?
Elijah Chudnoff (2014, 2016a) has proposed that what makes these seemings (or in his ter-
minology ‘experiences’) the sorts of entities that can serve as immediate justifiers is that they
have what he calls a ‘presentational phenomenology’. To a first approximation, seemings have
a presentational character only when their accuracy conditions ‘include both p and awareness
of a truth-maker for p’ (Chudnoff 2016a). Suppose you are looking at the coffee mug in front
of you. It seems to you that there is a coffee mug in front of you. If someone were to ask you
‘why does it seem to you that there is a coffee mug in front of you?’, one of the most plausible
answers is to point out that the coffee mug (whether it’s actually there or not) makes it seem that
299
Berit Brogaard
Figure 23.1 Despite the fact that some of the parts of the coffee mug are occluded, the whole mug
‘pops out’ and your experience of it has presentational phenomenology, on Chudnoff’s
view.
way to you. You are thereby making a reference to your (alleged) awareness of the truthmaker
for ‘there is a coffee mug in front of me.’
On Chudnoff’s view, a seeming that a thing is in front of you can have presentational
phenomenology, even when you cannot see every part of that thing. Consider the above illus-
tration (Figure 23.1).
Although the coffee mug is partially occluded from the scene, it ‘pops out’ in your per-
ceptual experience of the scene, owing to what is known as ‘amodal completion’. Chudnoff
would say that this gives us reason to think that your experience represents the coffee mug and
all its parts. However, the occluded parts of the coffee mug that you cannot see have a different
phenomenology than the non-occluded parts. The non-occluded parts have a presentational
300
Foundationalism
phenomenology, whereas the occluded parts do not. While the occluded parts of the mug
do not have a presentational phenomenology, the coffee mug does. That makes sense. You
clearly cannot see the occluded parts of the coffee mug but you can clearly see that there is a
coffee mug behind the Mexican candy and in front of the book. So, despite the mug being
partly invisible, the parts we do see suffice to make it ‘pop out’ in a way that gives rise to an
experience that has presentational phenomenology. The presentational phenomenology of the
experience is, in Chudnoff’s view, what makes the experience the sort of thing that can confer
immediate and full justification upon belief. His view thus satisfies what he calls ‘epistemic elitism’
(Chudnoff 2016b):
Epistemic Elitism – If a perceptual experience can immediately justify believing that p,

then p needs to meet some condition over and above being part of its representational
content.
Part of what qualifies experiences as immediate and full justifiers in many cases, on Chudnoff’s
view, is that in veridical cases they literally make us aware of the truth-maker for the content
of the experience. Non-veridical cases that are indistinguishable from the equivalent veridical
cases make it seem to us as if we are in direct conscious contact with the truth-maker for the
content of the experience.
While the view seems rather plausible for standard cases of visual experience, the question arises
whether it can be extended to memory? Although memory seemings are image-like, they are very
different from experience. They typically are less vivid, they lack in brightness, saturation, and hue
(in comparison), and details may be missing from the ‘image’. When having a memory seeming
of you dancing salsa last Saturday, it is not at all like having the experience itself. It is not at all as
if you are in direct conscious contact with the truth-maker of the seeming. And here we are even
just focusing on the visual element of memory. Once we move onto other sensory modalities, the
contrast is even greater. Memories of olfactory, gustatory, tactile, and auditory experiences can
be image-like but they are not at all like the original experience. While there may be a resem-
blance between the phenomenology of the visual experience of the cheese you ate and that of
your visual memory of the cheese, it is not as if your memory of the smell of the cheese has any
smell phenomenology. So, in these cases, (apparent) awareness of the truth-maker of the content
of your memory is partially or completely lacking.
4. The evidence resistance of phenomenal seemings

If phenomenal dogmatism about memory is true, then what is it about memory seemings that
makes them suitable as immediate and full justifiers of themselves or their propositional memory
equivalents? My proposal is that those memory seemings that can function in this way are evi-
dence resistant. I have defended this view for visual experience, appearances of personality, and
speech comprehension elsewhere (Brogaard 2013a, 2016a, 2016b). Here I will defend it for
memories.
Phenomenal seemings are quite unlike epistemic seemings in being evidence resistant. By
‘evidence resistant’, I mean that they endure even when we are faced with strong evidence that
they are inaccurate, at least assuming a rational agent. Let’s consider a non-controversial case:
the Müller-Lyer illusion (Figure 23.2).
In the Müller-Lyer illusion, the line segments have exactly the same length but because of
the fish hooks, they appear as if they have different lengths. This illusion persists even when you
measure the line segments and come to the conclusion that they do have the same length. This is
301
Berit Brogaard
Figure 23.2 The Müller-Lyer illusion – even when you learn that the line segments on the left have the
same length, they continue to appear as if they have different length.
a mark of phenomenal seemings. Epistemic seemings are not evidence resistant in this way. When
you hear on the radio that there will be a hurricane, and it comes to seem wise to you to evacu-
ate, this seeming will dissipate if you come to believe that the radio announcement was a hoax.
It is this evidence resistancy of phenomenal seemings that makes them the sort of thing that
can serve as an immediate and full justifier of other mental states. The seemings themselves
are not subject to defeat, even if the mental state they immediately and fully justify is. In the
Müller-Lyer illusion, the seeming persists after you measure the two line segments and come
to the conclusion that they have the same length. After your measurement you have a defeater,
viz. your belief that the line segments have the same length, so the immediate justification your
seeming confers on your belief (if you still have it) is defeated, and your belief that the two line
segments have the same length is no longer justified. This view satisfies a version of epistemic
elitism, viz. the following:
Epistemic Elitism (Seem) – If a seeming can immediately justify believing that p, then p
needs to meet some condition over and above being part of its representational content.
It may be argued that the view defended here does not satisfy epistemic elitism, as formulated by
Chudnoff. However, this only follows if all experiences are evidence insensitive. This, however,
need not be the case. We can imagine a scenario in which an evil demon makes your experience
fade away whenever you receive evidence that your experience is inaccurate. For example, if
you are looking at the Müller-Lyer illusion and measure the line segments, your experience of
them having different lengths fades away, and you immediately come to see them as having the
same length.
On my view, the reason that evidence-insensitive seemings can confer immediate justifica-
tion upon belief is that, in normal circumstances, they are likely to lead to true belief in the
absence of further background information. In the Müller-Lyer illusion, circumstances are not
normal. The illusion likely occurs as a result of the way the brain processes depth (Gregory
1968; Howe and Purves 2005; Brogaard and Gatzia in press). Depth perception involves gen-
erating an internal three-dimensional model of the environment. Part of the mechanism that
produces the three-dimensional model adjusts for the sameness in size of objects located at
different distances from us. This is also known as ‘size constancy’. This mechanism ensures
that objects are not perceived as shrinking when we move away from them. As a result of this
process, the brain projects the retinal image of the outward arrowheads to what would normally
be its correct distance in our internal model, thus making the line segment with the outward
arrowheads seem longer (Figure 23.3).
Evidence resistance of experience thus is a mark of accuracy of the beliefs that are directly
generated by the brain on the basis of the experience.
The memory seemings that are evidence resistant in the same way as our experience of the
Müller-Lyer illusion are those seemings that can serve as immediate justifiers of themselves or
302
Foundationalism
Figure 23.3 The Müller-Lyer illusion – illustration of how the outside corners generate the appearance
of the object being further away from us, whereas the inside corners generate the
appearance of the object being closer to us.
their propositional counterpart. Suppose after reminding your boyfriend of the great time you
had dancing salsa at the salsa club last Saturday, he denies that you and he were dancing salsa.
He even shows you evidence (e.g. text messages and pictures) that you left before the salsa band
came on stage. You come to believe he was right. You memory was a mere construction with
no grounds in reality. Suppose, however, that when you think back to that Saturday night, it
still seems to you very vividly that you were dancing salsa with him (see Mazzoni et al. 2010).
In this case, you have a memory seeming that can serve as an immediate justifier of itself. It’s
just that in this particular case, you have a defeater, viz. your belief that you weren’t dancing
salsa, so the original memory (or memory seeming) is not in fact justified despite having the
right justificatory marks.
If, on the other hand, your memory seeming of dancing salsa fades away after seeing counter-
evidence and you suddenly remember that you never were dancing salsa, then your memory
seeming was not the sort of memory state that can serve as an immediate and full justifier in the
first place. It was an epistemic memory seeming, perhaps with a sensory phenomenology, but
not a sensory phenomenology that can make it play the role of an immediate and full justifier.3
There are several objections that can be raised against this version of phenomenal dogma-
tism.4 One turns on the idea that evidence resistance is a disposition. It is notoriously difficult to
provide an adequate account of dispositions. For instance, an evil demon may make you resist
evidence for all of the seemings that you generate.
This objection, however, raises no special problem for the present account, as we are assum-
ing a certain level of rationality on the part of the agent. As outside forces influence your
psychology, this is not a case in which you are acting as a rational agent.
Another, potentially more devastating, objection turns on the new evil demon problem in
epistemology (Cohen 1984; Pollock 1984). The new evil demon problem was originally for-
mulated as a challenge for a version of reliabilism that defined justification in terms of reliability.
Imagine a world in which you have all the cognitive dispositions you have in the actual world
but in which a demon hinders their reliability. Since you have the same cognitive dispositions
in the actual world and the evil demon world, it seems that you ought to be equally justified in
both worlds. But this is not what the reliabilist predicts. Since your dispositions are not reliable
in the evil demon world, you are not justified.
One might raise a similar problem for the version of phenomenal dogmatism defended here.
Imagine two agents who, owing to their psychology, generate different memory seemings.
One agent forms an evidence-insensitive seeming about the past, whereas the other forms a
303
Berit Brogaard
evidence-sensitive seeming about the past. Despite having different dispositions toward poten-
tial evidence, however, they feel the same way internally. So, it might be argued, their memory
seemings have the same justificatory status.
My reply to this objection is that it is not analogous to the new evil demon problem. The
latter provides an objection to the implication that agents that are internally the same can be in
a different epistemic standing with respect to their justification. Agents in the scenario outlined
here are not on a par internally, as they have different dispositions toward potential evidence.
It may be argued, however, that this reply is unsatisfactory, as it may be thought that agents
who feel the same should be in the same epistemic standing with respect to the justification
they possess.
This objection, however, seems to me to presuppose that strong access internalism is a viable
position. Strong access internalism, recall, is the view that you can tell by introspection alone
whether you are justified in holding a given belief. But not even the defender of the view
that only experiences with presentational phenomenology can serve as immediate justifiers can
plausibly maintain a view of this sort. Imagine a case where two perceivers are both looking
at the same coffee mug but where the mug is partially occluded for just one of the perceivers
(Figure 23.1). Imagine that both perceivers have an experience that p, where p is that the coffee
mug is regular-shaped. Let’s furthermore suppose this experience is accurate. As a matter of an
abnormal psychology, however, perceiver S1, who is looking at the mug behind the occluders,
has an experience with the same presentational phenomenology-like feel to it as S2. In that case,
the two perceivers feel exactly the same internally but only S2 is immediately justified in her
belief about p. Because the two perceivers are internally on a par, neither perceiver can tell by
introspection alone that they are immediately justified in holding the belief in question. So, the
view in question is inconsistent with strong access internalism.
It may be replied that S1’s experience is just like S2’s, Because of partial hallucination. S2
genuinely sees a truth-maker for p. S1 just seems to see a truth-maker for p. So absent defeaters
both have experiences that immediately justify believing p. This, however, requires not only
that the content of their experiences is: p, and it seems that I am aware of a truth-maker for p, but
also that this highly intellectual content is reflected in the phenomenology of experience, which
seems rather controversial.
5. Metacognition and the problem of forgetting

Phenomenal dogmatism about memory has been subject to various criticism, the best known
of which is the problem of forgotten evidence (Harman 1986; Bernecker 2008). We often
forget our original evidence for our beliefs. Yet it would seem that beliefs for which we have
forgotten our original evidence remain equally justified. This has been taken to be a problem
for the internalist view that holds that your beliefs can be justified only if you have access to the
evidential ground of your belief.
A related problem is that most of our memory beliefs and memory seemings are stored rather
than occurrent. Yet it would seem that a memory should not lose its justification even if I hap-
pen not to entertain it. Accordingly, most of my beliefs are only intermittently or potentially
justified. This problem is also sometimes referred to as the ‘problem of stored beliefs’.
Conee and Feldman (2001) propose to solve the first problem by pointing out that our beliefs
may still be justified even if we no longer have access to the original evidence for our beliefs.
Even if I forget how I first learned that your name is ‘Sven’, I have multiple other sources of
evidence for this belief, for instance, the fact that people consistently address you in this way.
304
Foundationalism
This sort of reply can be extended to memory beliefs. Let’s say that I believe we have met
in the past, but that I have no memory seeming of the event. In this case, my memory belief is
not justified by any memory seemings. However, this does not mean that my memory belief
is unjustified, only that it is no longer justified by the original memory seeming. It might be
justified by a new memory seeming, for instance, by the memory seeming that you said ‘nice to
see you again’ earlier today.
This sort of reply, however, doesn’t ward off the problem of stored memory. Preservationism
offers a potentially promising solution to both problems. According to this view, our beliefs,
once justified, remain justified in the absence of defeaters (see e.g. Bernecker 2008).
One problem with preservationism is that it is rather strange to think that beliefs retain their
justification after being dismantled and put back together again upon retrieval. This form of radi-
cal reconstruction often introduces new mistakes that may undermine the original justification.
Suppose, for instance, that I have ‘stored’ a memory image of you wearing a yellow shirt at a party
ten years ago, and that this image epistemically supports a memory belief that you were wearing
a yellow shirt at that party. Although the memory image represents you as wearing a yellow shirt,
retrieval of the image might result in a memory seeming of you wearing a red shirt – perhaps
owing to a flaw in your memory retrieval system. It seems odd in these circumstances to maintain
that the stored belief of you wearing a yellow shirt is justified by the stored memory image. One
way to avoid problematic cases of this sort is to modify preservationism as follows:
Dispositional Preservationism – If memory image M would confer prima facie justification

on memory belief B at time t upon retrieval, then memory image M confers primary
facie justification on memory belief B at time t, even if neither is retrieved at time t.
Unlike the original version of preservationism, dispositional preservationism does not solve the
problem of forgotten evidence. So, it would need to be accompanied by a different solution, for
instance, the one proposed by Conee and Feldman (2001).
Another promising way to deal with the problem of forgotten evidence is to expand the set
of memory seemings to include metacognitive feelings of familiarity (Proust 2008; Bernecker
2008; Arango-Muñoz and Michaelian 2014). If asked what President Obama’s first name is,
I can readily provide this information. Like most people, I would be at a loss to explain where I
first learned this fact. However, semantic memory beliefs are often accompanied by a metacog-
nitive feeling of familiarity that appears to be evidence insensitive and hence should be able to
confer prima facie justification on the belief much like memory images. Feelings of familiarity is
a degree notion. So, the degree of justification that a feeling of familiarity can confer on belief
will defend on its strength, which can be partly captured in terms of retrieval fluency (Proust
2008). Retrieval fluency depends, among other things, on how readily the information can be
retrieved, and how many related details can be retrieved in addition to the requested informa-
tion. For example, if you can cite all sorts of facts about President Obama, then your fluency is
greater than if you only know that his first name is Barack.
6. Transfer of epistemic badness

For the case of visual experience, phenomenal dogmatism has met with criticism that it leads
to a kind of unwarranted bootstrapping. Susanna Siegel (2012) offers the following case. Jill
dogmatically believes that Jack is mad at her. Jill’s belief is irrational and not even minimally
justified. But because she holds this belief, she sees his neutral face as expressing anger the
305
Berit Brogaard
following day. According to phenomenal dogmatism, it would seem, this experience of Jack
expressing anger can now confer immediate and full justification on the belief she already
holds, viz. the belief that Jack is mad at her. This seems less than satisfactory. All she did was see
Jack, and then – as if by magic – her unjustified belief turns into a justified belief.
There is a way of blocking this sort of argument against phenomenal dogmatism. One pos-
sibility is to deny that bad epistemic properties can transfer from belief to experience, or from
experience to belief. A phenomenal dogmatist may say that just because the initial belief was bad,
this doesn’t mean that the belief that follows the experience is bad too. After all, she might con-
tinue, the experience with all its experiential phenomenology intervened and made a difference.
In a recent symposium (2013), Siegel offers an argument for why we should believe that bad
epistemic properties can transfer across chains of belief-experience-belief. She recognizes that
experiences and beliefs are different in very many ways. Most evidently, experiences are not the
sorts of things that can be rational or irrational. But, she notes, beliefs and experiences have an
important property in common. Beliefs need not be irrational, in order for them to generate a
subsequent ill-founded belief formed on the basis of it. If I rationally believe I am going to Ned’s
talk today but irrationally believe he is going to talk about philosophy of mathematics, then I
may end up with the irrational belief that I will be going to a philosophy of mathematics talk.
This irrational belief is based in part on my rational belief about where I am going later today.
If beliefs don’t need to be irrational in order for beliefs based on them to be irrational, then
experiences (which Siegel claims are never irrational) should also be able to give rise to beliefs
with bad epistemic properties.
So far, so good. But notice that while this argument does not rule out that transfer can occur,
it also does not establish it. The argument cannot simply be that beliefs and experiences both
have property p that allows for transfer of epistemic badness. Hence, since beliefs allow for trans-
fer, experiences do too.5 This argument is fallacious. If you and I are both hit by John, it may
be that John did something bad to you but not to me, because unlike you I had agreed to be
in a boxing match with John. In this case, there is a further property I possess which makes the
two cases disanalogous. But the phenomenal dogmatist might say that there is also a disanalogy
between belief and experience even if they both have property p in common that might make
it look like they can both transfer epistemic badness. The disanalogy might consist in the special
phenomenology that experience has. That may be what prevents the transfer of the epistemic
badness from the original bad belief to the belief Jill has after seeing Jack.6
Although I realize that this response is available to the phenomenal dogmatist in some cases,
I don’t think it works in all cases. In another of Siegel’s cases, John fears that there is a gun in
the fridge and subsequently comes to have an experience of a gun in the fridge. What he is see-
ing as a gun is really a banana. On the basis of this experience, John forms the belief that there
is a gun in the fridge. Now, in this case, we have two options. We can either allow that fear
can be subject to rationality constraints, or we can deny it. Emotions such as fear, sadness, and
disgust are arguably kinds of perceptual experience (Prinz 2006; Brogaard 2015). So, if we say
that emotions can be meaningfully said to be rational or irrational, then we should probably also
allow that experiences in general can be subject to rationality constraints, in which case transfer
of epistemic badness is less controversial. If, on the other hand, we deny that emotions can be
subject to any badness constraints, then there is no epistemic badness to transfer from John’s fear
via his experience to John’s belief. So, finding a way to argue against the transfer of epistemic
badness is not going to block Siegel’s argument. In fact, it is not clear that there is an argument
to be made in that case.
Siegel’s argument relies on the assumption that one’s experience can inherit the content of
one’s belief. This is also what is known as ‘cognitive penetration’ (Pylyshyn 1999). Whether
306
Foundationalism
our experiences are subject to cognitive penetration or not has been subject to much debate
(Raftopoulos 2001; Macpherson 2012; Siegel 2012; Siegel in press; Brogaard and Gatzia in
press). Here I shall set aside that debate and instead focus on how one can adequately address
similar apparent bootstrapping cases for memory, if indeed there are any.
Suppose you remember your girlfriend as sometimes being moody and aloof. Your memory
turns out to be mistaken. Furthermore, the memory seemings upon which it is based do not
have the marks that make them qualified for being an immediate and full justifier of your
memory. Your memory, let’s say, is not even prima facie justified. Today you meet up with
your girlfriend, and owing to your mistaken memory, you experience her as moody and aloof.
Later that evening, your memory seeming of her being moody and aloof earlier, which does
indeed have the marks that make it qualified for conferring immediately justification on your
memory, adds immediate justification to your earlier memory that your girlfriend is sometimes
moody and aloof.
In this case, blocking the transfer of epistemic badness does not seem very compelling, as it
should be rather uncontroversial that memories can transfer epistemic badness to other memo-
ries. The case also seems plausible, as it doesn’t involve a case of cognitive penetration of visual
experience, which has been the main controversy in the debate about cognitive penetration.
I think the right response in this case is that, upon further scrutiny, the case is non-worrisome.
Given the assumption that your memory seeming of your girlfriend today is of the kind that can
confer justification on memories, this memory seeming can and does indeed confer immediate
justification on your memory that your girlfriend was moody and aloof today. This latter mem-
ory can now justify the generalized memory that your girlfriend is sometimes moody and aloof.
If, on the other hand, we had a case on our hands where your memory seeming today did not
have the properties that made it a candidate for conferring immediate justification on memories,
then the case could not be used to challenge phenomenal dogmatism of the sort defended here.
7. Conclusion
I have argued that even if retrieved autobiographical memories are constructions put together
by our brains out of stored pieces of experiences, as empirical data strongly indicate, this does
not imply that we cannot come to have internal justification for what we remember. A version
of phenomenal dogmatism is defensible for memory. On this view, memory seemings – which
we can take to be akin to mental images, imaginations, or even sometimes vivid dreams – are
self-justifying in the absence of defeaters, but can also justify propositional counterparts of
these seemings. A hard question is what it is about memory seemings that makes them imme-
diate justifiers. One suggestion is that they have a special presentational phenomenology that,
at least in veridical cases, puts us in direct contact with the truth-makers for their content.
This view, however, is not very plausible for the case of memory, as it is doubtful that we
ever have any direct awareness of events in the past – certainly not when the events are far
enough back in the past. I have proposed instead that what makes a memory seeming the sort
of mental state that can self-justify or confer justification on memories is their evidence insen-
sitivity. In the case of visual experience, visual illusions are remarkably resistant to defeat. In
the Müller-Lyer illusion, for example, it continues to appear to us that the two line segments
have different lengths, even after gaining knowledge that this is not so. Many of our memory
seemings possess this same property: they continue to persist in spite of knowledge that the
seemings are highly inaccurate. At the end of this chapter, I offered what I consider the most
plausible response to a recent challenge to phenomenal dogmatism, viz. the challenge that the
position gives rise to bootstrapping cases.7
307
Berit Brogaard
Notes
1 Many other distinctions can be drawn here, such as that between declarative and nondeclarative memory,
that between implicit and explicit memory retrieval and that between short-term memory (also known
as ‘working memory’) and long-term memory. Here my primary focus will be on explicit memory
retrieval and long-term autobiographical memory.
2 This view is compatible with the view that our memories are causally connected to the past and also
with the view that the past could partially constitute the content of the representation (Bernecker 2008,
2010). I shall not take a stance on these questions here.
3 Here it may be helpful to distinguish between noetic and autonoetic awareness associated with memory
(see e.g. Klein 2014). Autonoetic awareness provides the subject with the ability to perform mental time
travel, whereas noetic awareness does not. If you know that you saw a Jimi Hendrix concert when you
were in high school, but cannot recollect being there, this memory is associated with noetic awareness
but not with autonoetic awareness. Memories associated with noetic awareness but not with autonoetic
awareness are less likely to be able to serve as immediate justifiers of belief. However, see the section
below on metacognitive feelings.
4 Thanks to Elijah Chudnoff here.
5 Elijah Chudnoff (personal communication).
6 Elijah Chudnoff (personal communication).
7 I am grateful to Sven Bernecker, Elijah Chudnoff, and Casey Landers for comments on an earlier version
of this chapter.
Related topics
• Taxonomy and the unity of memory
• Memory images
• Internalism and externalism
References
Arango-Muñoz, S. and Michaelian, K. (2014). ‘Epistemic Feelings, Epistemic Emotions’, Philosophical
Inquiries 2(1): 97–122.
Audi, R. (1995). ‘Memorial Justification’, Philosophical Topics 23: 31–45.
Audi, R. (1998). Epistemology. London: Routledge.
Bernecker, S. (2008). The Metaphysics of Memory. New York: Springer.
Bernecker, S. (2010). Memory: A Philosophical Study. New York: Oxford University Press.
Brogaard, B. (2013a). ‘Phenomenal Seemings and Sensible Dogmatism’. In C. Tucker (ed.), Seemings and
Justification. New York: Oxford University Press, pp. 270–89.
Brogaard, B. (2013b). ‘It’s Not What It Seems. A Semantic Account of “Seems” and Seemings’, Inquiry
56(2–3): 210–39.
Brogaard, B. (2015). On Romantic Love: Simple Truths about a Complex Emotion. New York: Oxford
University Press.
Brogaard, B. (2016a). ‘Perceptual Appearances of Personality’, Philosophical Topics 44(2): 83–103.
Brogaard, B (2016b). ‘In Defense of Hearing Meanings’, Synthese, doi: 10.1007/s11229-016-1178-x.
Brogaard, B. and Gatzia, D. E. (2017). ‘Is Color Experience Cognitively Penetrable?’, Topics in Cognitive
Science 9(1): 193–214.
Chisholm, R. (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press.
Chudnoff, E. (2013). Intuition. Oxford: Oxford University Press.
Chudnoff, E. (2014). Review of C. Tucker (ed.) Seemings and Justification, Notre Dame Philosophical Reviews.
Chudnoff, E. (2016a). ‘Moral Perception: High Level Perception or Low Level Intuition?’ In T. Breyer
and C. Gutland (eds.), Phenomenology of Thinking. New York: Routledge.
Chudnoff, E. (2016b). ‘Epistemic Elitism and Other Minds’, Philosophy and Phenomenological Research,
DOI: 10.1111/phpr.12308.
Cohen, S. (1984). ‘Justification and Truth’, Philosophical Studies 46: 279–96.
Conee, E. and Feldman, R. (2001). ‘Internalism Defended’, American Philosophical Quarterly 38(1): 1–18.
Conee, E. and Feldman, R. (2004). Evidentialism: Essays in Epistemology. New York: Oxford University Press.
308
Foundationalism
Danker, J. F. and Anderson, J. R. (2010). ‘The Ghosts of Brain States Past: Remembering Reactivates the
Brain Regions Engaged during Encoding’, Psychological Bulletin 136: 87–102.
Eichenbaum, H. and Cohen, N. J. (2001). From Conditioning to Conscious Recollection: Memory Systems of the
Brain. New York: Oxford University Press.
Ginet, C. (1975). ‘Memory Knowledge’, Knowledge, Perception and Memory, Philosophical Studies Series in
Gregory, R. L. (1968). ‘Visual Illusions’, Image, Object, and Illusion, Readings from Scientific American. San
Francisco, CA: W. H. Freeman and Company.
Harman, G. (1986). Change in View: Principles of Reasoning. Cambridge, MA: MIT Press.
Howe C. Q. and Purves D. (2005). ‘The Müller-Lyer Illusion Explained by the Statistics of Image–Source
Relationships’, PNAS 102(4): 1234–9.
Huemer, M. (1999). ‘The Problem of Memory Knowledge’, Pacific Philosophical Quarterly 80(4): 346–57.
Huemer, M. (2007). ‘Compassionate Phenomenal Conservatism’, Philosophy and Phenomenological Research
74: 30–55.
Jackson, F. (1977). Perception: A Representative Theory. Cambridge: Cambridge University Press.
Klein, S. (2014). ‘Autonoesis and Belief in a Personal Past: An Evolutionary Theory of Episodic Memory
Indices’, Review of Philosophy and Psychology 5(3): 427–47.
McKoon, G., Ratcliff, R. and Dell, G. S. (1986). ‘A Critical Evaluation of the Semantic-Episodic
Distinction’, J. Exp. Psychol. Learn. Mem. Cogn. 12(2): 295–306.
Macpherson, F. (2012). ‘Cognitive Penetration of Colour Experience: Rethinking the Issue in Light of an
Indirect Mechanism’, Philosophy and Phenomenological Research 84(1): 24–62.
Mazzoni, G., Scoboria A. and Harvey, L. (2010). ‘Nonbelieved Memories’, Psychological Science 21(9):
1334–40.
Pollock, J. (1974). Knowledge and Justification. Princeton, NJ: Princeton University Press.
Pollock, J. (1984). ‘Reliability and Justified Belief’, Canadian Journal of Philosophy 14: 103–14.
Prinz, J. (2006). Gut Reactions A Perceptual Theory of Emotion. New York: Oxford University Press.
Proust, J. (2008). ‘Epistemic Agency and Metacognition: An Externalist View’, Proceedings of the Aristotelian
Society 108(1): 241–68.
Pryor, J. (2000). ‘The Skeptic and the Dogmatist’, Noûs 34: 517–49.
Pryor, J. (2005). ‘There Is Immediate Justification’. In M. Steup and E. Sosa (eds.), Contemporary Debates in
Epistemology. Malden, MA: Blackwell, pp. 181–201.
Pylyshyn, Z. W. (1999). ‘Is Vision Continuous with Cognition? The Case for Cognitive Impenetrability
of Visual Perception’, Behavioral and Brain Sciences 22: 341–423.
Raftopoulos, A. (2001). ‘Is Perception Informationally Encapsulated? The Issue of the Theory-Ladenness
of Perception’, Cognitive Science 25: 423–51.
Schacter, D. L. (1989). ‘Memory’. In M. I. Posner (ed.), Foundations of Cognitive Science. Cambridge, MA:
MIT Press, pp. 683–725.
Schroer, R. (2008). ‘Memory Foundationalism and the Problem of Unforgotten Carelessness’, Pacific
Philosophical Quarterly, 89: 74–85.
Siegel, S. (2012). ‘Cognitive Penetrability and Perceptual Justification’, Noûs 46(2): 201–22.
Siegel, S. (2013). ‘The Epistemic Impact of the Etiology of Experience’, Philosophical Studies 162: 697–722.
Siegel, S. (in press). ‘Rational Evaluability and Perceptual Farce’. Afterword to J. Zeimbekis and A. Raftopoulos
(eds.), Cognitive Effects on Perception: New Philosophical Perspectives. Oxford: Oxford University Press.
Squire, L. R. (1992). ‘Memory and the Hippocampus: A Synthesis from Findings with Rats, Monkeys, and
Humans’, Psychological Review 99: 195–231.
Tolhurst, W. (1998). ‘Seemings’, American Philosophical Quarterly 35: 293–302.
Tucker, T. (2010). ‘Why Open-Minded People Should Endorse Dogmatism’, Philosophical Perspectives
24(1): 529–45.
309
24
COHERENTISM
Erik J. Olsson
1. Introduction
It is a fact of life that many of the information sources we consult in our daily occupations are
not very reliable in themselves. This goes for what we experience through our senses, what
other tell us, what we read in the news and, notoriously, on the Internet. In these cases, all we
have to go on may be the extent to which the reports that we get from various sources agree
or cohere. If they do cohere to a large extent we tend to think that the information is credible
and that what is reported is probably true. So, if we read on a webpage of dubious credibility
that there has been a robbery outside the Opera House in Stockholm and we overhear someone
saying just this on the metro, only to be told so by a distant acquaintance, then we may start to
think that there is probably something to it.
Impressed by examples like these, some theorists of knowledge have concluded that coher-
ence is the “holy grail” of epistemology, that one thing separating justified from unjustified
beliefs. Roughly speaking, beliefs that hang well together, agree, or exhibit mutual support
are thereby justifiably held and, if true, items of knowledge. This goes in particular for beliefs
based on memory: they are, on this view, justifiably held and, if true, qualify as knowledge if
they cohere with each other or with other beliefs that we have. Here as in the following we
use “memory,” “recollection” and synonymous terms in their non-veridical senses without any
commitment as to the truth of what is being remembered or recollected. In other words, we
view memory as an information system producing reports to the effect that this or that is the
case whether or not those reports be true or false.
Thus, the fact that Karen remembers receiving a lot of Christmas presents when she was
10 years old may not be of any significance by itself, nor the fact that she remembers a lot of
relatives gathering for Christmas that same year. Karen could very well be completely wrong
on both accounts. However, these two memories exhibit a high degree of mutual support: if
Karen received a lot of gifts, this indicates the presence of many relatives, and, conversely, many
relatives would normally mean many gifts for each child, including Karen. In the light of this
coherence, the most plausible explanation may very well be that both recollections are true.
Others have been more modest in their philosophical extrapolations from such pieces of
common-sense reasoning. For C. I. Lewis, for example, coherence cannot create justification
from scratch; it can only amplify an already existing positive degree of credibility. Thus beliefs
310
Coherentism
based on memory are not justified in virtue of coherence unless there is some initial credibility
attaching to those beliefs, taken individually. If there were no presumption in favor of what is
recollected because recollected, observing the coherence between many recollections would
not be epistemically significant. If recollections have such initial credibility, by contrast, coher-
ence may, in favorable circumstances, yield a probability “sufficient for rational and practical
reliance” (Lewis 1946: 357). Lewis would say that if we are impressed with the coherence of
Karen’s Christmas recollections, this is because we attach some initial credibility already to each
recollection considered in isolation.
Referring to Lewis in the connection of coherence justification for memories is particularly
fitting since the justification of our reliance on memory was central to his epistemological pro-
ject. If we cannot even rely on our memories, there is little hope that we can put our trust in
anything else. Lewis was ultimately interested in the grand project of justification memory as a
faculty without reference to that faculty itself. Thus, his main goal was to provide a justification
of memory as such, in a way that would satisfy even a seriously skeptical objector. Nevertheless,
his starting point was to examine more mundane situations like our Christmas case.
Lewis, as we just saw, suggested that we may think of coherence and truth in terms of
probabilities: coherence implies truth if it yields a probability sufficient of rational and practical
reliance. But what exactly is the role of coherence, and what are the probabilistic facts here? In
their search for answers, a number of contemporary authors, starting with Klein and Warfield
(1994), have focused on the question whether more coherence implies a higher likelihood of
truth. Let us say that coherence is truth conducive if it has that property.
As our focus is on coherence and truth of memories or recollections, the question we will ask
is: if a set of recollections S is more coherent than another set of recollections S’, are we then
allowed to conclude that S is more likely than S’ to be true as a whole? For instance, if Joseph’s
recollections on World War I are more coherent than his recollections of his son’s marriage,
does that meant that the former are more likely to be true and, if so, under what conditions?
There are good reasons to pay attention to this particular account of truth conduciveness. First,
it intuitively asks for a minimal sense in which coherence among recollections could imply
truth. It would seem difficult to maintain that coherence implies truth without also maintain-
ing that more coherence implies a higher likelihood of truth. Second, it is relatively clear and
unambiguous.
In the next section, the task will be to get clearer on what kind of property coherence is. I
will then, in Section 3, turn to the question of how to understand truth conduciveness. It is, as
already hinted, implausible to think that coherence is truth conducive in the absence of further
conditions: a well-composed piece of fiction may after all be more coherent than reality itself.1
Those further conditions will be scrutinized in Section 4. In Section 5, I briefly examine the
psychological plausibility of these conditions in the context of memories. In the final section, I
discuss the recent so-called impossibility results for coherence and their significance.
2. Coherence
C. I. Lewis proposed the following definition of coherence, or “congruence,” to use his own term:
A set of statements, or a set of supposed facts asserted, will be said to be congruent if

and only if they are so related that the antecedent probability of any one of them will
be increased if the remainder of the set can be assumed as given premises.2
(Lewis 1946: 338)
311
Erik J. Olsson
Thus, a set S consisting of propositions A1, . . . An is congruent relative to a probability distribu-

tion P just in case P(Ai | Bi) > P(Ai) for i = 1, . . .,n, where Bi is a conjunction of all elements
of S except Ai.
This definition makes coherence or congruence a matter of all or nothing, rather than
a concept admitting of more fine-grained degrees. A fine-grained concept would be more
suitable for our purposes given our account of truth conduciveness. Moreover, as the fol-
lowing example shows, congruence is questionable even as an explication of coherence
in the all-or-nothing sense. Suppose that there are a reasonable number of students and a
reasonable number of octogenarians (80–89-year-olds). Suppose that all and only students
like to party and that all and only octogenarians are bird watchers, and that there are some,
but very few, octogenarian students. Suppose that you seem to recall the following about
John, a distant relative:
A1 = “John is a student”
A2 = “John likes to party”
A3 = “John is an octogenarian”
A4 = “John likes to watch birds”
The set S = {A1,A2,A3,A4} is congruent in Lewis’s sense: the probability of each proposition in
the set is raised if the other propositions are assumed as given premises. Intuitively, however, the
set is anything but coherent: two of the recollections (about John being a partying student) are
very unlikely given the other two (about John being a bird-watching octogenarian).
In the example the joint probability of the propositions in the set is very low; the probability
that John is both a partying student and an octogenarian bird watcher is close to zero. This sug-
gests identifying the degree of coherence of a set with its joint probability:
C0(A,B) = P(A∧B)
The measure can be trivially extended to an arbitrary (finite) number of propositions:
C0(A1, . . .,An) = P(A1∧ . . . ∧An)
However, there can be situations where two sets have the same joint probability and yet, intui-
tively, one set is more coherent than the other, although such examples involving memory tend
to be a little contrived. Let us consider a two-person case involving Lena and Thomas. Lena
recalls that John is a student and also that John likes to party, whereas Thomas’s recollections
concerning John are less definite. He cannot remember whether John is a student or in the
army, only that John is the one or the other, and he cannot remember whether John likes to
party or to play tennis, only that John likes the one or the other. Assuming that being in the
army and playing tennis are statistically unrelated properties and, as before, that all and only
students like to party, Lena’s recollections seem more coherent than Thomas’s. After all, the
former but not the latter are in perfect agreement. And yet, as the reader can assure herself, the
joint probability is the same in both cases.
Rather than measuring the overlap of propositions, we might measure the extent to which
they agree. The more the propositions agree, the more coherent they are. A simple way to
312
Coherentism
measure the extent of agreement was introduced in Olsson (2002) and, independently, in
Glass (2002):
P( A ∧ B )
C1(A,B) =
P( A ∨ B )
C1(A,B) measures the relative overlap of the propositions A and B. This measure, too, is trivially
generalizable:
P ( A1 ∧ ... ∧ An )
C1(A1, . . .,An) =
P ( A1 ∨ ... ∨ An )
An alternative measure of coherence was introduced in Shogenji (1999):
P ( A|B ) P( A ∧ B )
C2(A,B) = =
P ( A) P ( A ) × P (B )
While C1 takes on its maximum value when the propositions coincide regardless of how specific
those propositions are, this is not so for C2. For suppose that A and B coincide. Letting x =
P(A) we get C2(A,B) = P(A∧B)/P(A)×P(B) = x/x2 = 1/x. Thus agreement on something more
specific, i.e., something with low probability, will give rise to a higher degree of coherence.3
Shogenji suggested the following straightforward generalization:
P ( A1 ∧ ... ∧ An )
C2(A1, . . .,An) =
P ( A1 ) × ... × P ( An )
Since C1 and C2 were introduced, the literature on coherence measures has virtually exploded,
and there are now a great many alternative proposals for how to think about coherence in rela-
tion to probability (see, for instance, Olsson and Schubert 2007). Nevertheless, the measures we
have introduced suffice for the purpose of fixing ideas.
3. Truth conduciveness
Although there are some exceptions, most philosophers who have discussed coherence and
likelihood of truth of sets of propositions have thought of the relevant likelihood as the likeli-
hood that the whole set be true, i.e., the likelihood of joint truth. In other words, the relevant
probability is the probability of the conjunction of all propositions in the relevant set.
This leads me to my first attempt to explicate what it means for coherence to be truth
conducive in the context of mnemic beliefs. A measure of coherence C will be said to be propo-
sitionally truth conducive if and only if a higher degree of coherence implies a higher probability
of joint truth, that is, if and only if the following holds: if C(A1, . . .,An) > C(B1, . . .,Bm), then
P(A1∧ . . . ∧An) > P(B1∧ . . . ∧Bm).
Are there any (non-trivial) propositionally truth conducive coherence measures? A measure
is truth conducive, in this sense, if more coherence means higher joint probability. According
to C0, coherence is the joint probability, and so this measure, trivially, comes out as truth con-
ducive in the propositional sense. On the other hand, neither C1 nor C2, two initially attractive
measures, turns out to be truth conducive in the sense of this definition. For example, we saw
before that among pairs of coinciding propositions C2 yields a higher coherence value for more
specific pairs, whereas more specific coinciding propositions have a lower (joint) probability.
313
Erik J. Olsson
One possible reaction is to conclude that the failure of reasonable accounts of coherence
to be propositionally truth conducive teaches us an important lesson about the connection
between coherence and truth, or rather about the lack of such a connection. This is, in effect,
the path taken by Klein and Warfield in their seminal 1994 paper. They conclude that coher-
ence theories of knowledge that rely on coherence to be truth conducive are profoundly
mistaken, taking the theory advocated in Bonjour (1985) to be a case in point. There is also
the option of blaming the troubles on our preliminary measures of coherence. But whether or
not one ultimately finds these measures unsatisfactory, there is a fundamental reason to be dis-
satisfied with our explication of truth conduciveness. The crucial observation is that what we
are primarily interested in here is the coherence and truth of memories, not the coherence and
truth of sets of propositions considered in the abstract.
Let us state the alternative proposal with more precision. We let RemSA stand for “The
subject S remembers that A”: A coherence measure C is mnemically truth conducive (for S) if
and only if: if C(A1, . . .,An) > C(B1, . . .,Bm), then P(A1∧ . . . ∧An|RemSA1, . . .,RemSAn) >
P(B1∧ . . . ∧Bm|RemSB1, . . .,RemSBm). The difference between the two conceptions of truth
conduciveness is that the mnemic conception, unlike its propositional counterpart, is a
conditional notion: it conditionalizes on the assumption that the propositions in question
correspond to recollections (of a given subject). In the next section, I will put forward some
conditions that seem required for coherence to be mnemically truth conducive. The remain-
der of this section is devoted to the elucidation of the definition.
When refuting the propositional truth conduciveness of C2, we referred to cases of two
pairs of coinciding propositions, one pair more specific than the other. We noted, on the one
hand, that C2 yields a higher coherence value for the more specific case and, on the other hand,
that the (joint) probability will in this case be lower. By probability is here meant the antecedent
joint probability. But from the fact that the antecedent joint probability is lower for the more
coherent case, it does not follow that the posterior joint probability—the joint probability con-
ditional on the propositions being the contents of recollections—is also lower. Generally, the
move from propositional to doxastic truth conduciveness blocks the inference from a lower
antecedent joint probability to a lower likelihood of truth (Olsson 2001; Olsson 2002; Bovens
and Olsson 2002).4
Although we are primarily interested in the truth conduciveness of coherence as applied to
memories, the main issues do not pertain essentially to memory reports, but to reports in general.
This point is stressed by Lewis:
It may also serve to emphasize the importance of congruence in the confirmation of

empirical beliefs if we observe in how large a measure the final basis of credibility must
be found in evidence having the character of ‘reports’ of one kind or another—reports
of the senses, reports of memory, reports of other persons—and this label ‘report’ is
appropriate just because such items do not fully authenticate what is ‘reported.’
Lewis (1946: 357)
A statement to the fact that a person remembers a given proposition can be taken as a report on
that proposition, in the sense of (putatively) indicating its truth. Memory reports are just one
type of report. Smith’s saying that A or Jones’s believing that B are examples of other types.
In general, we are presented with a number of reports, and we are interested in the relation
between the coherence of the report contents and their (conditional) joint probability. The
most general question, then, is whether more coherent propositions are more likely jointly to
be true among propositions that have been reported individually to be so.
314
Coherentism
4. Independence and partial reliability

According to C. I. Lewis, as we saw, coherence cannot do useful justificatory work unless there is
already some credibility attaching to the individual memories as such. Lewis requires the reports
to be, as he puts it, “relatively unreliable” (Lewis 1946: 346) for coherence to have any effect on
the credibility of the reported information. His choice of term indicates that he wants to rule out,
among other things, full reliability. Lewis also intends to disqualify the possibility of a report’s
being entirely disconnected from its content. According to Lewis, “For any of the reports taken
singly, the extent to which it confirms what is reported may be slight” (Lewis 1946: 346). But it
must not be zero; he writes, in his discussion of memory reports, that “If . . . there were no initial
presumption attaching to the mnemically presented . . . then no extent of congruity with other
such items would give rise to any eventual credibility” (Lewis 1946: 357). Surely Lewis is here
giving expression to the common-sense view: coherence of completely useless information does
not do a thing for the credibility of that information. Coherence cannot create credibility out of
nothing, but can at best amplify credibility that is already there from the start. Applying this to
memory reports, we should require that 1 > P(A|RemA) > P(A); that is, we should require that
the presence of a memory support its content, though without fully authenticating it. We will
refer to this condition as that of partial reliability rather than relative unreliability.
A further necessary condition for coherence to raise posterior joint probability, noted by
Lewis in connection with witness cases, is that the reporters tell their story independently
(e.g., Lewis 1946: 346). For if a coherent set is
fabricated out of whole cloth, the way a novelist writes a novel, or if it should be set up
as an elaborate hypothesis ad hoc by some theorist whose enthusiasm runs away with
his judgment, such congruence would be no evidence of fact.
(Lewis 1946: 352)
If it were such evidence then “unreliable reporters would be working in the interest of truth if
they got together and fudged their stories into agreement” (ibid.), a notion which Lewis, quite
rightly, dismisses as absurd. The point carries over to coherence among recollections although
Lewis himself sometimes lost sight of this fact: if memories have not been independently formed
but formed with the aim of coherence in mind, we would be unsurprised to discover them to
be in considerable agreement.
This said, it is implausible to require full independence for coherence to have the desirable
effect; intuitively, a tiny influence of the one memory report on the other does not cancel out
the effect of coherence entirely, although it does make that effect less pronounced. Thus, some
degree of dependence is compatible with coherence raising the joint probability, but the raise
will be less significant than it would have been, had the memory reports been less dependent.
The psychological realism of assuming memories to be independent is an interesting issue in
itself which will be briefly discussed in the next section.
As we will see, a formal argument can be given vindicating one of Lewis’s main claims which
Shogenji and Olsson (2004) refer to as his negative thesis: that coherence of useless independent
reports does not do a thing for the posterior joint probability of the set in question. This can be
shown by arguing that full independence should be given a particular exact interpretation and
then showing that coherence fails to be truth conducive under the assumptions of full independ-
ence and complete unreliability.
I have argued elsewhere (e.g., Bovens and Olsson 2000; Olsson 2002) that independence, in
this context, should be explicated as a form of conditional independence. The idea is that different
315
Erik J. Olsson
recollections, viewed as reports from memory, are independent just in case they are directly
influenced only by the respective purported facts they are about: they are not directly influenced
by other recollections, nor are they directly influenced by other putative facts. It will be helpful
for purposes of fixing ideas to refer back to our Christmas gift example throughout this section,
although nothing hinges on the peculiarities of the example. Consider the following propositions:
A1 = “Many of Karen’s relatives gathered for Christmas when she was 10”
A2 = “Karen received a lot of Christmas gifts as a 10 year old”
Rem1 = “Karen remembers A1, i.e., that many of her relatives gathered for Christmas when
she was 10”
Rem2 = “Karen remembers A2, i.e., that she received a lot of Christmas gifts as a 10 year old”
Rem1 and Rem2 are independent memory reports on A1 and A2, respectively, just in case Rem1 is
directly influenced only by A1, and Rem2 is directly influenced only by A2. How do we translate
this into probability theory? The trick is to block or, to use the standard term, “screen off” the
influence of A1 on Rem1 via A1 and see if there are still any influences on Rem1. There are two
ways to block that influence: assuming A1 true or assuming A1 false (or both), i.e., conditioning
on A1 or conditioning on ¬A1 (or both). Similarly for Rem,2 and A2. Conditional independence
is satisfied if there are no residual influences. In the interest of avoiding unnecessary technicali-
ties, I will confine myself to the two report case. The material in the paragraph that follows is
still rather technical; it will suffice for the purposes of the points I wish to make to have a general
grasp of what is going on.
We will say that Rem1 and Rem2 are independent memories of A1 and A2, respectively, if and
only if the following hold:
P(Rem2|A1,A2) = P(Rem2|A1,A2,Rem1) P(Rem2|A1,A2) = P(Rem2|A2)

P(Rem1|A1,A2) = P(Rem1|A1) P(Rem2|¬A1,A2) = P(Rem2|¬A1,A2,Rem1)
P(Rem2|¬A1,A2) = P(Rem2|A2) P(Rem1|¬A1,A2) = P(Rem1|¬A1)
P(Rem2|A1,¬A2) = P(Rem2 |¬A1,¬A2) =
P(Rem2|A1,¬A2,Rem1) P(Rem2|¬A1,¬A2,Rem1)
P(Rem2|A1,¬A2) = P(Rem2|¬A2) P(Rem2|¬A1,¬A2) = P(Rem2|¬A2)
P(Rem1|A1,¬A2) = P(Rem1|A1) P(Rem1|¬A1,¬A2) = P(Rem1|¬A1)
Collectively, these conditions express, in a precise way, that Karen’s individual memories have
an independent basis in her actual experience of the corresponding facts rather than being
directly influenced by other facts or memories. According to the second equation on the left,
for instance, if it is assumed that Karen received a lot of gifts (A2), then the probability that
Karin will remember this, as usual in the non-veridical sense of “remember,” is not affected by
learning in addition that there were a lot of relatives present (A1). This condition rules out, for
instance, Karen’s memory of the gifts being a reconstruction, in the sense of being caused by the
fact that there were a lot of relatives there.
There are two other concepts of independence that should be distinguished from conditional
independence in the sense just defined. First, one could by independence mean that the contents of
the recollections are independent, in the sense of P(A1|A2) = P(A1). In the Karen case, the contents
316
Coherentism
would be independent if learning that many of Karen’s relatives came over for Christmas would
neither add to, nor subtract from, the probability that she received a lot of gifts. Independence in
this sense interferes with coherence and results generally in a low coherence value. If such content
independence were the relevant notion of independence for the purposes of coherence and truth,
it would be impossible for two independent memories to cohere, which is absurd.
Second, there is independence in the sense of P(Rem1|Rem2) = P(Rem1): learning that Karen
remembers receiving a lot of gifts does not provide any information whatsoever as of whether
she also remembers there being many relatives present. Normally, however, learning what a
person remembers should raise our confidence, if ever so little, that she will remember other
things that are in agreement with the first memory (given the condition of partial reliability).5
This kind of dependence is quite compatible with the memories being independent in the sense
relevant for the purposes of coherence and truth.
It can now be proved (Olsson 2002) that Lewis was right in his negative claim: coher-
ence among memories that are independent in the relevant conditional sense but, taken singly,
completely unreliable will not have any effect on the posterior probability of those memories.
Interestingly, this holds regardless of whatever plausible account of coherence one plugs in. If
there is no reliability pertaining to individual independent memories, then nothing is gained
epistemically by combining them, however great their mutual coherence might be.6
There is a further issue that we need to consider regarding the conditions under which
coherence could be truth conducive. We recall that truth conduciveness here means the prop-
erty which coherence has if the memories in a more coherent set are more likely to be true than
memories in a less coherent set. Even if the conditions of independence and partial reliability are
in place, it is pretty obvious that no measure of coherence can be truth conducive unless certain
aspects of the situation are fixed. To see this, let us return to an earlier example: Joseph’s memo-
ries from World War I as contrasted with his memories of his son’s wedding. We assumed that
the former are more coherent than the latter. Let us suppose, moreover, that in both cases the
memories satisfy, in sufficient approximation, the conditions of independence and partial reli-
ability. Yet since Joseph’s memories from the wedding are much more recent, they are plausibly
individually more reliable than his old memories from the war. We would expect, therefore, the
wedding memories to be also more likely to be true, collectively speaking, despite the fact that
they are less coherent than the war memories.
The bottom line is that the beneficial effect of coherence on posterior joint probability can be
offset by the greater reliability of the memories in the less coherent set. In order for a comparison
regarding coherence to be epistemically significant, we must make sure that we compare sets of
memories that do not differ in partial reliability. More generally, coherence can at best be truth
conducive only in a ceteris paribus sense, i.e., if everything else is equal. Exactly what the ceteris
paribus clause involves in the case of coherence has been, and to some extent still is, the topic of
an intense debate (Douven and Meijs 2007; Schupbach 2008; Huemer 2011; Schubert 2012b),
but one thing seems clear: the partial reliability of memories is one of the things that must be held
fixed in the evaluation of the mnemic truth conduciveness of measures of coherence.
Examples like these, in which the sets in question differ with respect to partial reliability,
also show that coherence cannot be truth conducive in the intuitively more ambitious sense of
high coherence implying high posterior joint probability. The posterior joint probability of a
set of memories is underdetermined by coherence alone, even under the favorable conditions of
independence and partial reliability. As shown in Shogenji and Olsson (2004), there is no degree
of coherence which suffices to achieve a posterior joint probability over a certain threshold,
regardless of the degree of partial reliability of the memories. In other words, any claim about
a particular level of coherence being thus sufficient can always be undermined by exhibiting a
317
Erik J. Olsson
suitable case of very low reliability. Thus, what Shogenji and Olsson refer to as C. I. Lewis’s
strong positive thesis about coherence is demonstrably false.7
Yet, for any degree of partial reliability and any posterior joint probability, there is plausibly
a degree of coherence sufficient to attain that posterior. In other words, if the reliability of the
reports is fixed at a particular level, which can be low, and there is a posterior joint probability
sufficient for acceptance, then there is a degree of coherence such that the posterior can be
attained, so that C. I. Lewis’s weak positive thesis is true. Olsson and Shogenji give an example
showing that the posterior can be very close to one even if the independent reports are individu-
ally almost completely unreliable. To take a criminal case, as more dubious witness reports come
in incriminating a given suspect there is, for even the most cautious judge, a point at which it
becomes practically certain that the suspect is guilty.8
The problem of accounting for the relation, if there is one, between coherence among
memories and truth, or at least one clear and manageable sub-problem, is finding a (non-trivial)
measure of coherence that is mnemically truth conducive ceteris paribus under (essentially) the
conditions of partial reliability and independence. Before we proceed to examine recent nega-
tive results in relation to this problem, we will take a brief look at the psychological realism of
assuming independence and partial reliability in the context of memories.
5. Issues of psychological realism

As we have seen, the conditions of independence and partial reliability (the former within
plausible approximation) are necessary for coherence to have any connection with truth or
high posterior joint probability of a set of memories. But how realistic is it to assume that our
memories have these properties? Influential research in cognitive psychology suggests that while
the assumption of independence is highly problematic for common kinds of memory, we may
safely assume that our recollections generally satisfy the condition of partial reliability.
As Conway explains, referring to seminal work by Greenwald (1980):9
Coherence is a strong force in human memory that acts at encoding, post-encoding

remembering, and re-encoding, to shape both the accessibility of memories and the
accessibility of their content. This is done in such a way as to make memory consistent
with an individual’s current goals, self-images, and self-beliefs . . . thus, memory and
central aspects of the self form a coherent system in which, in the healthy individual,
beliefs about, and knowledge of, the self are confirmed and supported by memories of
specific experiences. By way of illustration consider an example of perhaps not uncom-
mon occurrence: supporters of a football team recalled events from an important match
that confirmed their belief in the sportsmanship and high skill level of the players, despite
the fact that their team had clearly played a very physical and unsportsman-like match.
(Conway 2005: 595)
Conversely, extreme violations of coherence, whereby memories undermine or contradict

important aspects of the self, are usually only present in psychological illness or following brain
damage. Conway concludes:
It seems then that autobiographical memory is dominated by the “‘force’ or ‘demand’

of coherence” since “[a] stable, integrated, self with a confirmatory past that yields a
consistent and rich life story . . . constitutes a self that is able to operate effectively,
achieve goals, and relate to others in productive ways.”
(Conway 2005: 595–6)
318
Coherentism
Memory, then, to a considerable extent creates its own coherence. This is done, for instance,
by lowering the accessibility of memories of events which challenges the person’s goal structure
or by distorting memories of such events in order to maintain coherence and delay or avoid a
cognitively costly goal change (Conway 2005: 597). Thus, we can expect violations of inde-
pendence, in the sense explained in the previous section, especially among memories that are
about a person’s own life history. Those memories will not simply be a function of reality mak-
ing an independent impression on the person’s mind. Rather, there will be a cognitive pressure
to make them form a coherent whole. Furthermore, since the self may distort memories for this
purpose, they cannot generally be assumed to be fully reliable representations of the past.
None of this implies, of course, that even autobiographical memory is nothing but a fabrication
of the self:
Set against the force of coherence is the demand of correspondence. Conway, Meares
et al. (2004) argued that from an evolutionary perspective a memory system that did
not maintain an accurate record for goal processing and the effects of goal processing
would be unlikely to survive. Memory then should correspond to experience. On
the other hand, a system that maintained literal or even highly detailed records of
moment-by-moment experience would be faced with insurmountable problems of
storage and retrieval. The memory system is, therefore, faced with several mutually
contradictory demands. One is to represent reality as it is experienced, but in cogni-
tively efficient ways, and another is to retain knowledge in such a way as to support a
coherent and effective self . . . In autobiographical remembering there is a great deal
that people can accurately remember, i.e., that certain events occurred, without recall-
ing many or, in some instances, any further details.
(Conway 2005: 596)
We can conclude that there are certain parts of memory closely tied to a person’s self-image
for which the assumption of independence is not very plausible. Observing the great coher-
ence of those memories is not a significant epistemological fact because coherence was only
to be expected. On the other hand, even autobiographical memory is not completely out of
touch with reality.10 So, while there are reasons to dismiss the idea that our rememberings are
generally completely independent and fully reliable, the demand of correspondence should
guarantee a positive degree of these qualities in many cases and more so for some types of
memories than for others.
Let us return to Karen and her Christmas memories for an illustration of how independ-
ence could fail. Karen has an image of herself, we postulate, of being a positive and happy
individual coming from a loving and peaceful family. According to Conway, Karen’s self
would now operate in such a fashion as to force her autographical memories to cohere with
her self-image and therefore also amongst each other. Suppose there were in fact many
relatives coming over for Christmas that year but due to severe disagreements among fam-
ily members the children were, disappointingly, left without Christmas gifts. This scenario
would plausibly lead to violations of independence in Karen’s case. Given Karen’s self-image
and her memory of the relatives, there is a positive chance that her childhood memory
that she did not receive any gifts will be distorted into its opposite for the purposes of
attaining coherence. But this is in violation of one of our independence equations, namely
P(Rem2|A1,¬A2) = P(Rem2|A1,¬A2,Rem1). This equation states that, in the circumstances as
described, the probability of Karen remembering receiving a lot of gifts would not be affected
by her remembering there being a lot of relatives around.
319
Erik J. Olsson
6. Impossibility results and what they mean

One conclusion of the previous section is that it is not entirely unproblematic to apply our
conception of truth conduciveness to coherence in the realm of memories: the required
background conditions need not be satisfied. But there are more general concerns about the
possibility of coherence to be truth conducive, concerns that apply not only to memories but
to all forms of reports.
We are now in a position to state the impossibility theorems for coherence that have been
the subject of considerable discussion in recent years. What they show is that no measure of
coherence is truth conducive even in a weak ceteris paribus sense, under the favorable condi-
tions of (conditional) independence and partial reliability. The theorems were proved for
coherence of reports in general, regardless of their sources, and they hold in particular for
reports from memory.
The first result of this nature was presented in Bovens and Hartmann (2003). What they
show is that every possible measure of coherence is vulnerable to a certain kind of counterex-
ample. More precisely, by manipulating the level at which the reliability of the reports is held
fixed, one can create a situation in which the more coherent set has a lower posterior joint
probability than the less coherent set. Olsson (2005) proves a similar impossible result in a dif-
ferent setting that is closer to the kind of scenario that C. I. Lewis was concerned with in his
1946 book. The approach is the same, though: to construct counterexamples to a measure of
coherence being truth conducive by strategically manipulating the reliability of the reports in
the sets being compared.
These impossibility results give rise to a perplexing paradox. Our starting point in this chap-
ter was the common use in everyday life of coherence reasoning based on the reports from
less than reliable sources. But how can it be that we trust and rely on coherence reasoning, in
everyday life and in science, when in fact coherence is not truth conducive? Unsurprisingly, this
question has been the topic of much recent discussion in the literature on coherence.
Some commentators accept the conclusion that the impossibility results show that coher-
ence is not truth conducive, adding that this does not prevent coherence from being valuable
and important in other ways. Others doubt that the impossibility results show that coherence
is not truth conducive, typically because they doubt at least one premise used in the proofs.
The approaches of the latter group range from questioning the way in which independence is
formalized (Huemer 2007, 2011) to questioning the ceteris paribus conditions used in the proofs
(Douven and Meijs 2007; Schupbach 2008; Huemer 2011). Here I will focus on the responses
of the first camp and in particular those responses that can be of interest in understanding coher-
ence and truth in the context of memories.
One such study is Olsson and Schubert (2007) in which it is observed that, while coherence
falls short of being truth conducive, it can still be “reliability conducive,” i.e., more coherence,
according to some measures, entails a higher probability that the sources are reliable, at least in
a paradigmatic case (cf. Schubert 2012a, 2011). This would mean, for memories, that higher
coherence can, in some situations, be indicative of the underlying memory-forming processes
being more reliable. Nevertheless, Schubert has proved an impossibility theorem to the effect
that there is a similar impossibility result for reliability conduciveness: no coherence measure is
reliability conducive in general (Schubert 2012b).
For another example, Staffan Angere (2007, 2008) has argued, on the basis of computer
simulations, that the fact that coherence fails to be truth conducive, as defined, does not prevent
it from being connected with truth in a weaker, defeasible sense. In fact, as he shows, many
coherence measures that have an independent standing in the literature satisfy the condition
320
Coherentism
that most cases of higher coherence are also cases of higher probability, although they do so to
different degrees. If this is true, we can rely on coherence as a heuristic guide to truth. As such,
coherence may not yield the kind of justification required for knowledge but only justified
beliefs in a defeasible sense.
Finally, it has been noted that coherence plays an important negative role in our thinking.
Applied to the case of memories, the proposal is that if our memories show signs of incoher-
ence, this is often a good reason for contemplating a closer examination of the purported facts
seemingly recalled. A longer discussion of this point can be found in Olsson (2005: Chapter 10).
While the debate has shown that coherence among our memories may not be a very strong sign
of their truth, their failure vis-à-vis coherence may still be an excellent reason for thinking that
some may be false.
Notes
1 The following sections draw on material in Olsson (2002) and Olsson (2012), where the focus is on
coherence of beliefs rather than of memories. More extended treatments of various aspects of Lewis’s
coherence justification of memory can be found in Olsson (2002: Section 5), Shogenji and Olsson
(2004), and Olsson (2005), especially Chapter 3.
2 Lewis prefers “congruence” to “coherence” because he wants to dissociate himself from British post-
Kantian idealism and its “coherence theory of truth.”
3 For examples illustrating the differences between these measures, see Olsson (2002).
4 For a related conditional account of truth conduciveness, see Cross (1999).
5 The fact that the person remembers something (e.g., receiving a lot of gifts) raises the probability of the
corresponding fact, which raises the probability that there are other coherent facts (e.g., a lot of relatives
being present), which raises the probability that the person will remember those facts as well.
6 A similar but less general theorem was proved in Huemer (1997). For a recent discussion, see Cleve (2011).
7 More precisely, Olsson and Shogenji show that Lewis’s strong positive thesis is contingent on assuming
the Principle of Indifference, a principle which Lewis explicitly rejected.
8 The example presupposes that the degree of coherence is raised when adding more fully agreeing
reported propositions.
9 Interestingly, Conway reports that Bertrand Russell’s discussion of coherence vs. correspondence theo-
ries of truth in The Problems of Philosophy (Russell 1912) was a starting point for thinking about the
nature of autobiographical memory in psychological theory (Conway 2005: 595).
10 Even the modest claim that memory is in part designed to represent reality as it is experienced may be
challenged by recent accounts that view our ability to imagine the past as being derivative of our ability
to imagine the future. See Michaelian (2016: Chapter 11).
References
Angere, S. (2007) “The Defeasible Nature of Coherentist Justification,” Synthese 157(3): 321–35.
Angere, S. (2008) “Coherence as a Heuristic,” Mind 117: 1–26.
Bonjour, L. (1985) The Structure of Empirical Knowledge, Cambridge, MA: Harvard University Press.
Bovens, L. and Hartmann, S. (2003) Bayesian Epistemology, Oxford: Clarendon Press.
Bovens, L. and Olsson, E. J. (2000) “Coherentism, Reliability and Bayesian Networks,” Mind 109: 685–719.
Bovens, L. and Olsson, E. J. (2002) “Believing More, Risking Less: On Coherence, Truth and Non-trivial
Extensions,” Erkenntnis 57: 137–50.
Cleve, J. V. (2011) “Can Coherence Generate Warrant Ex Nihilo? Probability and the Logic of Concurring
Witnesses,” Philosophy and Phenomenological Research 82(2): 337–80.
Conway, M. A. (2005) “Memory and the Self,” Journal of Memory and Language 53: 594–628.
Cross, C. B. (1999) “Coherence and Truth Conducive Justification,” Analysis 59: 186–93.
Douven, I. and Meijs, W. (2007) “Measuring Coherence,” Synthese 156(3): 405–25.
Glass, D. H. (2002) “Coherence, Explanation and Bayesian Networks,” in Artificial Intelligence and Cognitive
Science, M. O’Neill and R. F. E. Sutcliffe et al. (eds.) (Lecture Notes in Artificial Intelligence, Volume 2464),
Berlin: Springer-Verlag, pp. 177–82.
321
Erik J. Olsson
Greenwald, A. G. (1980) “The Totalitarian Ego: Fabrication and Revision of Personal History,” American
Psychologist 35: 603–18.
Huemer, M. (1997) “Probability and Coherence Justification,” Southern Journal of Philosophy 35: 463–72.
Huemer, M. (2007) “Weak Bayesian Coherentism,” Synthese 157(3): 337–46.
Huemer, M. (2011) “Does Probability Theory Refute Coherentism?” Journal of Philosophy 108(1): 35–54.
Klein, P. and Warfield, T. A. (1994) “What Price Coherence?” Analysis 54: 129–32.
Lewis, C. I. (1946) An Analysis of Knowledge and Valuation, LaSalle, IL: Open Court.
Michaelian, K. (2016) Episodic Memory and Our Knowledge of the Personal Past, Cambridge, MA: MIT Press.
Olsson, E. J. (2001) “Why Coherence Is Not Truth-Conducive,” Analysis 61: 236–41.
Olsson, E. J. (2002) “What Is the Problem of Coherence and Truth?” The Journal of Philosophy 99: 246–72.
Olsson, E. J. (2005) Against Coherence: Truth, Probability, and Justification, Oxford: Clarendon Press.
Olsson, E. J. (2012) “Coherentist Theories of Epistemic Justification,” The Stanford Encyclopedia of
Philosophy (Spring 2014 edn.), Edward N. Zalta (ed.), http://plato.stanford.edu/archives/spr2014/
entries/justep-coherence/.
Olsson, E. J. and Schubert, S. (2007) “Reliability Conducive Measures of Coherence,” Synthese 157(3):
297–308.
Russell, B. (1912) The Problems of Philosophy, London: Williams and Norgate.
Schubert, S. (2011) “Coherence and Reliability: The Case of Overlapping Testimonies,” Erkenntnis 74:
263–75.
Schubert, S. (2012a) “Coherence Reasoning and Reliability: A Defense of the Shogenji Measure,” Synthese
187(2): 305–19.
Schubert, S. (2012b) “Is Coherence Conducive to Reliability?” Synthese 187(2): 607–61.
Schupbach, J. N. (2008) “On the Alleged Impossibility of Bayesian Coherentism,” Philosophical Studies
141(3): 323–31.
Shogenji, T. (1999) “Is Coherence Truth-conducive?” Analysis 59: 338–45.
Shogenji, T. and Olsson, E. J. (2004) “Can We Trust Our Memories? C. I. Lewis’s Coherence Argument,”
Synthese 142: 21–41.
322
25
PRESERVATION AND
GENERATION
Thomas D. Senor
Introduction
Preservationism is the thesis that, generally speaking, a memory belief’s justification is a function
of the epistemic status it had when it was initially formed. Closely related to preservationism is
the claim that memory is not an epistemically generative process. In this chapter, I shall argue for
preservationism and for the view that memory is not typically epistemically generative. I begin
by considering the nature of epistemic generation and its relation to memory, and proceed to
argue that preservationism is supported by these theoretical considerations. While our discussion
will reveal that memory can be generative both doxastically and epistemically, it is nevertheless
primarily best conceived as an epistemically preservative psychological process.
1. Laying the groundwork

At any given time the vast majority of what we believe is not in consciousness but is stored in
memory. If these beliefs are largely unjustified, then a serious variety of skepticism follows. Yet
if they are justified, upon what does their justification depend? There are two broad approaches
in the epistemology of memory—one synchronic and the other diachronic. According to the
former, the justification of a memory belief must be present at the time of the epistemic evalu-
ation. S’s memory belief that P is justified at t only if there is some present, non-historical fact
about S at t that provides the justification. On the other hand, diachronic theories hold that the
history of one’s memory belief is relevant to its present epistemic evaluation. We’ll now exam-
ine these two types of theories to see what subtypes are possible for each.
According to synchronic theories, the justification of a memory belief at time t is strictly a
matter of what the current, non-historical facts are at t. But what kinds of facts might those be?
There would seem to be two possible kinds of answers. First, one might have or be in a state
that is a source of non-inferential justification; second, the relevant fact about one might be that
her evidential state at t provides her with inferential support for her belief that P. Let’s examine
these in turn.
There is a trio of possibilities for non-inferential, synchronic justification for a memory belief.
First, and most radically, it might be held that all memory beliefs are prima facie justified. This
view has been defended under the monikers of “conservatism” and “negative coherentism.”1
323
Thomas D. Senor
As long as the subject does not have a defeater, the belief is ultima facie justified. Alternatively,
one might argue that memory beliefs are justified in virtue of an experiential ground that is
non-inferential in nature. A traditional view is that memory beliefs are justified in virtue of their
being associated with an image or some other distinctive experiential state that served as their
epistemic ground.2 Much like my seeing the tree provides me with immediate justification for
believing there is a tree in front of me, my recalling the image of my cereal bowl this morn-
ing provides me with prima facie justification for believing that I had cereal for breakfast. The
third possibility for non-inferential justification of memory belief is the position that any time
it seems to one that p, one is prima facie justified in believing that p. This perspective, known
as “phenomenal conservatism,” differs from standard conservatism in putting the experiential
seeming in the epistemic front and center. So it is an experiential position in a way that the
conservatism neat is not. But the relevant experience need not be a recollected image of a past
experience but can be simply the fact that it seems to the subject that her memory belief is true.
In addition to these varieties of non-inferential accounts, there are inferential synchronic
theories of the justification of memory belief.3 On these positions, a memory belief is justified in
virtue of an inference that is available to the subject at the time of the evaluation of the belief.
In principle, there are two ways such accounts could go. First, one might insist that a memory
belief is justified only if the subject has a good argument that directly supports the truth of p. So,
for example, if I have a stored belief that the Cubs will be better next year than they are this year,
my belief will be justified only if I possess an inferential justification that would be sufficient for
now forming the belief for the first time.
I say that such a position is a way that inferential theories could go because, as far as I know,
no one has advocated it. And it is easy to see why: such a position would entail that many long-
held memory beliefs that we take to be epistemically above reproach are, in fact, unjustified. For
example, my belief that Lincoln was assassinated in 1865 is not accompanied by any such direct
inference, yet we tend to think such beliefs are justified. So, synchronic theorists are wise to
allow beliefs, the evidence for which has been forgotten, to nevertheless count as justified. Still,
it is curious why a synchronic theory of justification would make such a significant distinction
between the conditions necessary and sufficient for justifiably coming to believe a proposition and
justifying continuing to believe one.
The second variety of synchronic inferential theory allows for a memory belief to be justified
indirectly; that is, the inference is grounded in a conviction that the belief is a memory belief
together with a belief that memory beliefs (at least memory beliefs of the relevant sort) are reli-
able. This synchronic inferential theory of memory justification will allow that a belief can be
justified even when we’ve lost our initial evidence for it, since another justification can be found
among one’s doxastic and experiential states.4
In contrast to synchronic theories, a diachronic theory will allow that the history of the
belief is relevant to its epistemic evaluation. Just as with the former, there is no single way for
diachronic theory to be developed. The most straightforward theory is preservationism. The
preservationist holds that the justification of a memory belief is largely a matter of its prior
epistemic status. More to the point, she claims that other things being equal, a memory belief
at t is prima facie justified if and only if the belief was justified when it was originally formed.
The clause “other things being equal” is necessary because one might be unjustified in forming
a belief and then later become justified by acquiring new evidence. Obviously, in such a case
what matters is not the original justificatory state but rather the improved, later state. This is
consistent with the primary idea behind diachronic theories, however, because when the belief
is remembered sometime after the new evidence has been acquired (and appreciated), its justifi-
cation then is a function of its justification at the earlier, intermediate time.
324
Preservation and generation
There are other ways of formulating diachronic theories. It is helpful to see that whereas
synchronic accounts say that only non-historical facts are relevant for justification, diachronic theories
say only that historical facts are relevant, and not that only historical facts are relevant. So a diachronic
account can incorporate synchronic elements. For instance, one might think that while the way
a belief has been formed and maintained is relevant to its justification, its epistemic past needs a
boost from the present in order for it to be prima facie justified. For example, one could hold
that, other things being equal, a memory belief is justified only if it was justified when formed
and it now seems to one that the belief is true (this would be to add a diachronic element to the
phenomenal conservatist view).5 Or one might hold that one needs evidence that memory is
reliable (or reliable regarding the matter at hand) in addition to a belief’s having the right kind
of pedigree. One motivation for adding a diachronic feature to either the phenomenal con-
servative or the evidentialist position is that it blocks a standard objection to purely synchronic
accounts memory belief, to wit: one can be unjustified in coming to believe a proposition but
once it is believed and in memory, it might well satisfy the synchronic criteria for memorial
justification. For example, suppose I come to believe Obama’s favorite food is lasagna because a
character on a sitcom asserted it. My belief is unjustified. But tomorrow (or next week, month,
or year) when my friend wonders aloud what Obama’s favorite food is and I recall believing that
it is lasagna, my belief will now be justified on standard synchronic views. But why should the
mere passage of time alter the epistemic status of that belief? One who accepts a mixed view of
either sort mentioned above would be able to say both that my belief was unjustified and that
one’s current evidence is crucial in generating prima facie justification.
As in so many disputes in epistemology, the question of how best to adjudicate between
competing theories of memorial justification is not so clear. Examining cases to see how our
theories fare against the intuitions we have about them is the standard approach. But, notori-
ously, judgments about cases clash, and when they do, this methodology quickly leads to a
stalemate. So I think that, when possible, it is better to look for theoretical reasons for preferring
a view to its competitors. Regarding the justification of memory belief, I believe that broader
considerations can help us decide which of these theories is to be adopted. The issue between
a purely diachronic position like preservationism and various synchronic alternatives, I shall
argue, turns on whether memory is epistemically generative. For if memory isn’t (typically)
generative,6 then its role in the justification of belief would seem to be preservative, and we’d
have a reason to accept preservationism. On the other hand, if memory is (typically) generative,
then the preservationist’s position is untenable. So if we can get clear on what it is for a process
to be generative, and then evaluate memory and the prospects of its generating justification,
we will have a theoretically grounded reason to pick between the alternative perspectives. As
will become clear in due course, I will argue that memory is not (typically) generative and that
because of this, preservationism has a leg up on its competitors.
Before making my case, I will need to say a bit more about what preservationism is. We will
then flesh out the nature of epistemic generation, making clear what makes a cognitive process
generative. This turns out to be a somewhat complicated business.
2. Preservationism
As explained above, preservationism is the thesis that the justificatory status of a memory belief
is primarily a function of its earlier epistemic state. Another way of characterizing preservation-
ism is as the thesis that a belief gets no justification by simply being retained in memory, being
recalled, or seeming to be true.7 If, at t1, I believe that P, and then at a later time, t2, recall my
belief that P, then provided that I have not acquired new (non-mnemonic) justifiers from t1 to
325
Thomas D. Senor
t2 regarding P, my justification for believing P at t2 can be no greater than the justification that
the belief had at t1. Of course, if memory is hazy or unreliable or I’ve gained defeaters for P,
then my belief may not have as much justification as it had originally. But that is consistent with
preservationism. For preservationism is the thesis that in the absence of a source of new justifi-
cation, a belief can’t have more justification than it had when it was first believed, and not that it
must have the same measure of justification when recalled.8 Putting the point slightly differently,
a belief’s justification is not enhanced by simply being maintained or recalled.
The motivation for preservationism is clear: memory serves us by preserving what we put
into it; that is its function in our cognitive system. We don’t ask it to get us truth; we hope only
for memory to retain it. Given the role memory plays in our cognitive lives, the preservation-
ist believes, memory preserves rather than generates justification. Being a vehicle of epistemic
preservation is importantly distinct from being a source of epistemic generation. And it is this
generation-thesis that the preservationist denies. So the justification one has for holding a belief
at the earlier time can be preserved via memory, but one’s maintaining and recalling the belief
does nothing to generate justification.
At this point, one might object as follows: thinking of memory as preservative is akin to
thinking of it as a storehouse. The picture is a natural one. When you experience something or
come to learn a new fact, you store it away until it is needed later. Recollection is then a mat-
ter of pulling out the right item from storage. Of course, we all recognize that we are fallible in
what we recall, but that can be easily enough explained on the storehouse model: sometimes we
can’t find what we are looking for and we come up empty. Alternatively, sometimes we misi-
dentify items and retrieve the wrong item (e.g., I’m asked where I went on vacation last year
and recall being in Jackson, WY; the problem is that I was there two years ago and not last year).
But as natural as it is, the storehouse model of memory has been pretty well discredited in the
psychology of memory. Memory is constructive at the time of learning in that what gets put into
long-term memory is a matter of the features of the episode that get processed and how they
are encoded; it is not a passive process of storing away an image or video/audio clip. Memory
is also reconstructive in that what one recalls at the later time is, as it were, put back together (or
maybe together for the first time) when the remembered event is activated. So instead of being
a passive storage receptacle that houses items that one later removes without the item being
changed (much as one opens a jpeg from one’s hard drive), memory is a complicated process,
both constructing the item initially at encoding and reconstructing it later at retrieval.9
If memory is both constructive and reconstructive, then one might wonder why it isn’t
epistemically generative as well. In the same way that, say, perceptual processes produce beliefs
from percepts, memories produce beliefs via a combination of storied items and background
knowledge. So why not think that perception and memory are equally doxastically and epis-
temically generative?
This is an important issue and one that I do not have the space to thoroughly explore.
As this chapter progresses, we will see the ways that, on my position, memory is and isn’t
generative. For now, though, let me say that I am not convinced that the constructive nature
of memory entails that perception and memory are equally generative or that there is even
a strong parallel. First, cases in which memory is clearly reconstructive are cases of episodic
memory. The phenomenon of “false memory,” in which one seems to remember being
witness to something that one wasn’t witness to, concerns the apparent recollection of past
experiences. And while, of course, much of what we take ourselves to remember is episodic,
much of what we remember is what psychologists call “semantic memory.” Semantic memory
is not just memory of concepts and word meanings, but includes general information about
326
both the world at large and one’s life. And it is much less clear that construction and recon-
struction is prevalent in this domain. There isn’t reconstruction going on when one simply
recalls that Lincoln was assassinated in 1865 or that Wrigley Field is on Chicago’s North Side.
Second, while constructive/reconstructive models of memory alter the way we might be ini-
tially inclined to think about how memory works, those models are consistent with the idea that
we have beliefs (and lots of them!) stored in long-term memory. If we have stored beliefs, they
presumably each have a justificatory status. So even if episodic memory is shot through with
construction and reconstruction, we’ll still need an epistemology that accounts for the epistemic
status of stored beliefs. And, by my lights, preservationism will be the preferred theory because
all we can expect from memory with respect to the belief that P is that it be retained; remem-
bering the belief from one time to the next provides that belief with no extra epistemic boost.
3. What is it for a process to be generative?

In order to make my case that preservationism is to be preferred over its rivals, I will give an
account of what it is for a process to be generative and then argue that memory (generally)
doesn’t qualify. Before getting into my positive account, however, I want to briefly discuss an
argument that Jennifer Lackey makes for thinking that memory is epistemically generative.10
Because I’ve responded to her elsewhere, I will be very brief here.11 After discussing Lackey’s
cases, we’ll explore the crucial issue of what it is that makes a process epistemically generative.
Lackey has a two-fold goal in her paper. She wants to show that memory can be epistemically
generative and, because of this, that preservationism is false (since the preservationist claims that,
all things being equal, a belief can’t have a higher epistemic state simply in virtue of memory).
To accomplish these goals, she presents cases in which a subject has a true belief that is suffi-
ciently well justified that, in the absence of defeat, it would count as knowledge. But, alas, defeat
is not absent and the subject does not know. For instance, Arthur has excellent reason from an
inside source to believe that the mayor has been taking bribes; and indeed the mayor is on the
take. However, there are widely circulated (although false) reports that the accusations against
the mayor are mistaken. If Arthur were paying attention to any news sources, he would have
seen these reports. So he has a normative defeater that prevents him from knowing that the mayor
is taking bribes. Eventually, it comes out that these defensives of the mayor are wrong, and that
the mayor is guilty. Arthur is oblivious to all of these reports and rebuttals, and has never quit
believing what his inside source told him. According to Lackey, although Arthur didn’t know
that the mayor was taking bribes when he first learned it (because of his normative defeater), he
does know later when it is widely reported that the claims of the mayor’s innocence were false
(because there is then no normative defeater).
On Lackey’s view, memory has been epistemically generative in this case. For at the earlier
time, Arthur had a true belief that didn’t count as knowledge, and at the later time, this true
belief is knowledge. Since he has acquired no new evidence for the belief, memory has gener-
ated knowledge. But neither of the lessons Lackey takes the case to show are clearly right. First,
it’s not obvious that Arthur knows at the later time. Lackey thinks that because there is no
longer a defeater, Arthur’s knowledge isn’t then defeated. A preservationist might well claim,
however, that just as evidence can justify at a later time after it is forgotten, a defeater can defeat
at a later time even if it is no longer available. But, second and more importantly, even if we
grant that Arthur knows at the later time and didn’t at the earlier time, that is not an indica-
tion that memory is epistemically generative in any substantial sense. This will, I hope, become
apparent as this chapter progresses.
327
Thomas D. Senor
4. Varieties of justification and belief-forming processes

In order to become clearer on what’s required for epistemic generation, I’ll begin by marking
the distinction between initial justification, preserved justification, and transmitted justification.
Let’s call newly minted justification derived from generational sources original justification. This
contrasts with the ongoing justification a belief has at a later time; call this preserved justification.
Furthermore, when I justifiably believe P, justifiably believe Q, and reasonably infer R from
them, my newly formed belief has neither original justification (because the process of infer-
ence doesn’t generate new justification) nor preserved justification (since the justification is new
for R). So we need a third category. Let’s say that a belief has transmitted justification when it
is properly epistemically grounded in at least one justified belief, the content of the ground-
ing belief is distinct from the belief it grounds, and the latter belief is justified (at least in part)
because of the justification of the earlier belief.12
So much for the trifold distinction of flavors of justification. Let’s now consider belief
formation. Some of our cognitive processes are aimed at acquiring information. That is, their
role in our cognitive economy is to add true beliefs to our doxastic corpus. The most obvi-
ous such process is perception. In standard circumstances, my seeing the dog on the sofa has
the result of introducing a new belief into my cognitive system; not only that, but the newly
formed belief is prima facie justified. Perception, then, is both a belief-generating process and a
source of original justification, which is to say it produces not only beliefs but justification
that does not come via transmission from other beliefs or preservation from earlier beliefs.
Introspection is another process that provides us with new beliefs that are not derived infer-
entially from other beliefs, and hence with initial justification.13 Unlike perception, however,
introspection is aimed at representing the consciously available mental life of the subject.
Even if the objects of introspection are already part of the subject’s mental inventory, the
goal of the process is to represent the objects of introspection as items in the subject’s mind.
Rational intuition is another means by which beliefs are formed and original justification is
generated. My “seeing” the validity of modus ponens is my ground for forming the belief that
there are no possible cases in which arguments with that logical structure have true premises
and a false conclusion; rational intuition is both the producer of the belief and the generator
of initial justification.
When these processes function as they typically do, new beliefs are formed and are prima
facie justified. The justification is generated rather than transmitted in these cases because it is not
beholden to the epistemic status of other beliefs. That is, their epistemic grounds are not them-
selves justified. In the case of perception and rational intuition, the epistemic ground of the
belief will be either the associated experience (the literal percept for perception and the “seeing”
of rational intuition) or a virtue of the process that produced it (e.g., being reliable or function-
ing properly). Introspection is a bit different because in some cases what is introspected is a
belief; however (as noted in Note 12), the justification of the introspective belief is still original
because it is not a function of the justification of the belief being introspected.
When beliefs are formed on the basis of inference, their prima facie justification is not
generated but transmitted. Transmission of justification occurs when a belief becomes prima
facie justified in virtue of having been inferred from other justified beliefs. Unlike perception,
introspection, and rational intuition, inferential belief-forming processes are doxastically but
not epistemically generative. That is, while they produce beliefs, they don’t generate justifica-
tion. For consider: if S infers Q from her beliefs P and (If P, then Q), the justification for Q
is a function of her justification for P and (If P, then Q); the justification for Q cannot be any
greater than that of the least justified premise. That is, inference, when done well, can fully
328
transmit justification, but it can neither produce justification where there is none nor increase
the justification had by basing beliefs.
Memory, as we have noted, is importantly different from perception, introspection, and
rational intuition in that its primary function in the cognitive system is preservative. When
things go optimally, the inputs to a mnemonic process and the outputs are just the same; the
best we can possibly hope for concerning a memory belief is that no content is lost or added.
Memory seeks to preserve belief; yet often, when a new doxastic state is produced by memory,
it is unreliably formed.14 If I seem to remember seeing Mary at the office party last Christmas
when in fact I neither saw nor thought I saw her there, my memory is playing me tricks and any
belief formed by this apparent memory is the result of mnemonic malfunction. Or consider a
case in which I’ve been told of an event in my childhood so often that now I seem to remember
witnessing it even though the belief that I witnessed it is purely a product of what I’ve been told.
In such cases, memory produces belief in an unreliable way.
Because of its role as a doxastic preservative, memory might be thought incapable of generat-
ing justification, or of producing original justification; for as we have seen, original justification
is a product of belief-producing rather than belief-sustaining processes. If preservationism is
equivalent to the thesis that memory is never epistemically generative, then preservationism
itself will entail that memory never produces prima facie justification. If this were true it would
explain why, ceteris paribus, no memory belief can have more justification than it had origi-
nally. We would then have a reason to think preservationism is theoretically preferable since if
memory never produces prima facie justification, then the epistemic status of the belief would
presumably never be strengthened by memory.
Natural as this thought is, we will come to see that it isn’t right. Indeed, we’ll see that
memory turns out to be epistemically generative in pretty much the same way as perception,
introspection, and rational intuition. All of this, I shall argue, is not only consistent with preser-
vationism, but when understood aright, is a reason to think that preservation is true.
5. More on epistemic generation

What is it, then, for a process to be epistemically generative? The obvious thing to say is this:
EG1: A process is epistemically generative iff it is able to produce beliefs that have
original justification.
The problem with this simple answer is that it conflates (or maybe compounds) epistemic and
doxastic generation. Inasmuch as the cognitive process in question produces beliefs, it will of
course also be doxastically generative. But an account of epistemic generation need not include
any reference to belief production. So, let’s cut that from (EG), which gives us:
EG2: A process is epistemically generative iff it is capable of conferring original

justification.
As we’ve noted, an originally justified belief is a belief that is justified but which has not had its
justification transmitted from other justified beliefs.15
This is all right as far as it goes but it doesn’t go very far. To be more exact, we need to
be explicit that what is crucial for epistemic generation is the ability to produce original prima
facie justification. A belief is prima facie justified if, in the absence of defeat, it is ultima facie justi-
fied, or justified simpliciter. A justification-generating process must be able to confer prima facie
329
Thomas D. Senor
justification on a belief, but it is not necessary that it be able to guarantee that any of its beliefs
is ultima facie justified. This is because the ultima facie justification of a belief depends crucially
on what other beliefs a subject has, what sorts of evidence gathering the subject has done (or
failed to do), and on the various vagaries of the epistemic social environment the subject hap-
pens to find herself. For example, if I see what looks to me like a small dog on your sofa, I’ll be
prima facie justified in believing there is a dog on the couch. But if you’ve told me that people
often mistake your stuffed animal for a real dog, then the prima facie justification that percep-
tion has generated in this case will be defeated. Yet the fact that my prima facie justification is
defeated does not suggest that original justification was absent in this case. Perception, after
all, can’t be held accountable for my every background belief. As long as perception (in the
right circumstances) provides me with what it takes for me to be ultima facie justified provided
that there are no defeaters, then it has done all the epistemic work that can be asked of it. It is
enough that it can produce original prima facie justification. That is:
EG3: A process is epistemically generative iff it is capable of conferring original prima

facie justification.
Any process that is incapable of producing original prima facie justification is thereby not an
epistemically generative process.16
6. Is memory generative?
I’ve claimed that to be epistemically generative, a process must be capable of producing
prima facie justification. And if preservationism is right, then memory isn’t capable of pro-
ducing prima facie justification and hence memory isn’t generative. For if a recalled belief
can have no more justification than it had when first formed, then memory adds nothing of
epistemic significance. But that is just to say that it isn’t generative. So from the account
of epistemic generation glossed earlier, we can get a pretty direct argument to the claim that
memory is not epistemically generative.
Although this perspective is initially plausible, it turns out to be wrong. If beliefs were the
only contents of memory, then memory would not be generative. But memory contains other
items too. For example, suppose after an episode of freezing rain, I walk gingerly out to my
driveway to pick up my newspaper. After coming back into the house, my wife asks if the
whole driveway is covered in ice. I think for a minute, and reflect on the visual experience I had
when picking up my paper. I realize that I had seen the entire driveway and that there was one
area under a tree that was not ice covered. I thereby form the belief “The area of the driveway
under the tree is not icy.” My new belief has prima facie justification and has derived this justifica-
tion from memory. So memory can be epistemically generative after all.17
By making a couple of observations, we can begin to see that the recognition that memory
can be generative in such circumstances does not undermine preservationism. First, the icy-
driveway case does not involve a belief that formerly lacked prima facie justification coming to
have it solely in virtue of memory. Rather, the belief is formed for the first time on the basis of
memory. That is to say, in this example, memory is both epistemically and doxastically genera-
tive. In this way, memory does resemble perception, but this is not an instance of a pre-existent
belief’s becoming justified through the working of memory. So this kind of case fails to be a
counterexample to preservationism because the latter claims only that no belief gains justifica-
tion merely by being held.
330
Now a preservationist might want to dig in her heels here and deny that the icy-driveway
example shows that memory is epistemically generative in the same way that perception is (i.e.,
by being epistemically generative when it is doxastically generative). Here is how such a case
could be made. Suppose that when I look at the ice on my driveway as I reach for the paper,
I notice that the ice has not yet formed under the tree, and formed the appropriate belief. So
I have a prima facie perceptual belief that the ice hasn’t formed under the tree. Now, consider
my situation a few moments before my perceptual experience of my driveway. Not only did I
not believe that the ice had yet to form under the tree but I also had no reason to believe it. It
is only in having the perceptual experience and seeing that the cement under the tree was bare
that I came to possess, in any sense, justification for the relevant claim. In the vocabulary of
many epistemologists, prior to looking at the driveway I lacked not only prima facie justification
but also propositional and ex ante justification for the proposition that there is ice under the tree.
But after looking at the driveway but before examining my remembered image, I came to have
both of these. One has propositional justification for a proposition just in case one possesses
evidence that justifies the proposition. One can have propositional justification for propositions
one doesn’t believe (and even for propositions one disbelieves). Similarly, one has ex ante justi-
fication for a proposition just in case this conditional is true: if one were to come to believe the
proposition, one would be justified in his belief. Prior to the perceptual experience I have of my
driveway, I lack both propositional and ex ante justification. But once I have the visual experi-
ence, I have both propositional and ex ante justification for the proposition that there is no ice
under the trees covering my driveway, even if I have not yet attended to that part of my visual
representation. For my total evidence, which includes my stored visual experience, supports the
proposition that there is no ice under the tree; furthermore, I have ex ante justification because
in the nearest worlds in which I form a belief that there is no ice under the trees are worlds in
which I reflect on my experience and come to believe the proposition on that basis.
So the preservationist might insist that the person who sees the driveway but doesn’t imme-
diately form the relevant belief, and later does with the help of an image stored in memory, has
both propositional and ex ante justification for her belief on the basis of her perceptual experi-
ence. So in order for memory to be doxastically and epistemically generative, it must make use
of resources provided for it by another epistemically generative source. But this is not true of
perception. So even in the best case, memory’s generative capacities are beholden to those of
another process.
I don’t propose to make too much of the significance of the propositional and post-ante
justification that are ultimately derived from non-mnemonic sources. Allowing that memory
is generative in cases when it is also doxastically generative doesn’t do anything to weaken the
connection between memory’s epistemically generative capacities and preservationism. But the
fact that even in cases in which memory is both epistemically and doxastically generative, it is
only capable of either because the generative powers of another source does reinforce memory’s
epistemically derivative nature.
So let’s grant that the icy-driveway case shows that there are circumstances in which memory
and perception are similar in that both produce justification when they also produce belief. This
would seem to work against the preservationist view that memory is importantly different from
perception in not being epistemically generative. Inasmuch as memory and perception are epis-
temically alike, there would seem to be good motivation for thinking that memory is generative
too. However, the preservationist might insist that there is still a relevant epistemic difference
between the two. Perception can be epistemically generative even when it is doxastically dor-
mant. That is, one can have a belief that is unjustified that later becomes justified on the basis of
331
Thomas D. Senor
further perceptual experience. So suppose I believe that the cat is on the mat because I always
believe what my lying-friend Ted says. I know Ted lies but I’m an incurable optimist who
believes that this time he’s telling the truth. So my belief that the cat is on the mat is unjustified.
But then I look at the mat and, unmistakably, see the cat. So my formerly unjustified belief has
become justified because of my perceptual experience. Perception is able to convert a previ-
ously unjustified belief and make it into a justified belief. In short, it has the power of epistemic
generation even in cases where it is not doxastically generative. So might the preservationist say
that we’ve found a way that perception is generative and memory is not because memory is only
epistemically generative when it is also doxastically generative?
Unfortunately, no. We can generate parallel cases in which an originally unjustified belief
gains epistemic support from memory. Now of course it will be possible for a previously unjus-
tified belief to have inferential support among other beliefs the person has in memory. But
that will be a more or less standard case of epistemic transmission. In order for a memory case
to parallel perception in the relevant respect, it will need to be a case of original justification
provided by memory. But cases like this are easy to generate. Suppose that my lying-friend Ted
comes to my door and tells me that my driveway is only partly iced over. Ever gullible, I believe
him. I relay this belief to my wife who reminds me that Ted is not to be trusted. I then reflect
on my experience of seeing my driveway just before he arrived and I realize that I had seen
my driveway only partially iced over. Assuming that my visual memory is generally reliable,
my previously unjustified belief that my driveway is only partially iced over is now justified. So
memory, like perception, has the power to convert unjustified belief into justified belief. That
is, memory can be epistemically generative even when it is not doxastically productive.
Let’s take stock. The preservationist understands memory to be, typically, not epistemically
generative. To make the case, memory is contrasted with other processes, like perception, that
clearly do generate prima facie justification. The first thought was that what makes memory not
epistemically generative is that it is not doxastically generative but is instead merely preservative;
in contrast, perception is clearly both. But then the initial icy-driveway case was presented and
memory shown capable of both doxastic and epistemic generation. A second thought was that
whereas memory is only epistemically generative when it is doxastically generative, perception
can generate justification for already acquired beliefs. But then it turns out that memory can do
the same. So it seems that every relevant distinction we’ve been able to find between perception
and memory has, in the end, collapsed, and we are left with the (perhaps surprising) conclusion
that perception and memory are epistemically generative in pretty much the same way.
7. Why preservationism is attractive

What is the upshot of this? In particular, does this show not what I had initially claimed—that
considerations of epistemic generation support preservationism—but rather that preservationism
is wrong? Here’s why someone might think that: if memory is epistemically generative, then it
produces prima facie justification. And if it produces prima facie justification, then the epistemic
status of a given memory belief is not determined by the original justification that the belief had
when it was originally formed. Put slightly differently, memory not only preserves (assuming it
does preserve) but it adds to the justification of a belief. And that is pretty clearly just another
way of saying that preservationism is false.
So is the result of our inquiry the demise of the position I set out to argue for? I’m happy
to say that it is not. What we’ve discovered is this: like perception, memory can be epistemi-
cally generative both when it is and when it isn’t being doxastically generative. That is, both
produce prima facie justification in both kinds of cases. So memory and perception are generative
332
in much the same way.18 However, that is not a problem for preservationism since it is a theory
about the relationship between the epistemic status of a belief at one time and the epistemic of
the same belief at later time. Preservationism has nothing to say about the epistemic status of
either a belief that is newly formed on the basis of memory or a previously unjustified belief that
becomes justified because of a recollected nondoxastic state.
So what do preservationism and the generative powers of memory have to do with each
other? Preservationism requires that a belief does not get an epistemic boost just by being
maintained. If memory were epistemically generative in the sense that it produces prima facie
justification for a belief by sustaining it over time, then preservationism would be false. But the
kinds of cases in which memory is generative are not of this type. Not only that, they are also
exceedingly rare. Far and away, the main function of memory is preservative. How well it lives
up to its role is, of course, a matter of debate. But it cannot be seriously doubted that, for the
most part, memory’s job in our cognitive economy is to allow us to, in some way or another,
make available to us what we’ve learned from perception, rational intuition, introspection,
inference, etc.
My conclusion is clear but still rather modest. I don’t pretend that what I’ve made anything
like an overwhelming case for preservationism. And this is so for (at least!) two reasons. First,
some will not think that considering epistemic generation as property of cognitive processes is
to the point. Internalists of various stripes will have little sympathy for this approach. Second, I
have not argued directly that, for example, memory seemings are epistemically inert, and doing
so would be a necessary part of making a strong overall case for preservationism.
What I’ve attempted to do here is simply to provide a plausible account of what it is for a
cognitive process to be epistemically generative, and argued that while memory is epistemically
generative in the same general way that perception is, it is only very rarely generative (unlike
perception which almost always is). Given that where beliefs are concerned, memory’s role in
our cognitive system is preservative, it is very natural to think that it typically adds nothing to
the epistemic status of our beliefs. When it works as it should, memory preserves the justifica-
tion that was generated when the belief was produced in the first place.
Acknowledgment
Thanks to Jack Lyons and Kirk Michaelian for helpful discussion and comments on an earlier
draft of this chapter.
Notes
1 See McGrath (2007) and Harman (1986).
2 See Russell (1921).
3 Although I’m using the term “inferential” to emphasize that these theories require the epistemic sup-
port of other doxastic states or credences in order for memory beliefs to be justified, one might also
develop such a theory strictly in terms of evidence sets and their relations to the target beliefs.
4 See Feldman and Conee (2001).
5 See Huemer (1999).
6 It will be clear later in the essay why the “generative” is here being modified by “typically.”
7 Throughout this chapter, I will use “justification” to denote the epistemic virtue under discussion. But I
could just as easily talk about warrant or epistemization. The topic of this chapter is general or abstract
enough, however, that the choice of terminology isn’t important.
8 An important, but often neglected, issue concerns the epistemic status of the belief between the time
when it is first formed and when it is later recalled. However, for the purpose of this chapter, I’ll ignore
such beliefs and focus instead on beliefs that have been recalled.
333
Thomas D. Senor
9 From Lampinen and Beike: “Despite the appeal of the storehouse metaphor, modern-day memory
psychologists put very little stock in the idea” (2015: 56). See also Alba and Hasher (1983).
10 See Lackey (2005) and Lackey (2007).
11 See Senor (2007).
12 This is described as such to allow for the possibility of a mixed process that takes both beliefs and expe-
riences and produces new beliefs from them. More standardly, I suppose, is the case where S justifiably
infers that R from justified beliefs P and Q, and S’s belief that R has its justification transmitted from
beliefs P and Q.
13 While it is true that some introspective processes take beliefs as inputs, these processes are not inferential.
So when asked if I believe that the President will be re-elected, I introspect and form the second-order
belief “I believe that ‘I believe that the President will be re-elected.’” While there is some clear sense in
which the latter belief is “based on” the former belief, this is not a sense that concerns epistemology. For
the epistemic status of the second-order belief is independent of the epistemic status of the first-order
belief: even if I am unjustified in believing that the President will be re-elected, I will nevertheless be
justified in believing “I believe that ‘I believe the President will be re-elected.’” I thank Jack Lyons for
this point.
14 As we will see presently, this turns not to be quite right.
15 Technically, for the sake of defining original justification, what is required is that the justification not
be transmitted from anything. That is, if there are other states that have justification and are capable of
transmitting it, then a belief will have original justification only if its justification has not been transmit-
ted from one or more of those states.
16 My insistence that a process be able to produce prima facie justification (rather than merely take one
from prima facie to ultima facie justification) distinguishes my account of epistemic generation from Sven
Bernecker’s “moderate generativism” and also from Lackey’s view. See Bernecker (2010: 96–103) and
Lackey (2005) and Lackey (2007).
17 Cases of this kind can also be found in Lackey (2005).
18 Although the previous discussion of the propositional/ex ante justification had by propositions prior to
the formation of the relevant beliefs from memory, and the dependence of such on perception should
be noted as a non-trivial difference between the two.
References
Alba, J. W. and Hasher, L. (1983) “Is memory schematic?” Psychological Studies, 93: 2003–231.
Feldman, R. and Conee, E. (2001) “Internalism Defended.” American Philosophical Quarterly, 38: 1–18.
Harman, G. (1986) Change in View: Principles of Reasoning. Cambridge, MA: MIT Press.
Huemer, M. (1999) “The problem of memory knowledge.” Pacific Philosophical Quarterly, 80: 346–57.
Lackey, J. (2005) “Memory as a generative epistemic source.” Philosophy and Phenomenological Research, 70:
636–58.
Lackey, J. (2007) “Why memory really is a generative epistemic source: A reply to Senor.” Philosophy and
Phenomenological Research, 74: 209–19.
Lampinen, J. M. and Beike, D. (2015) Memory 101. New York: Springer Publishing Company.
McGrath, M. (2007) “Memory and epistemic conservatism.” Synthese, 157: 1–24.
Russell, B. (1921) The Analysis of Mind. London: Allen & Unwin Limited.
Senor, T. D. (2007) “Preserving preservationism: A reply to Lackey.” Philosophy and Phenomenological
Research, 74: 199–208.
334
26
SKEPTICISM AND MEMORY
Andrew Moon
1. Introduction
Our topic is skepticism and memory. Skepticism is the view that we cannot have knowledge or
rational belief in some important area. I mean by “knowledge” and “rational belief” what they
mean in ordinary English. I will use “warrant” as a technical term for whatever it is that makes
the difference between knowledge and mere true belief. (I assume, with Plato, that there is a
difference between knowing a truth, and merely believing that truth.) Although most of the
skeptical arguments discussed in this chapter could be formulated so as to conclude either that we
lack rational belief or that we lack knowledge (or warrant), for specificity, I will normally just
formulate them so they conclude the latter.
The topic of memory covers many cognitive processes and qualities.1 I will focus on the cog-
nitive qualities attributed to Fred in the following, ordinary English sentences:
(I) Fred remembers that Tammy ate the taco.

(II) Fred remembers Tammy eating the taco.
(III) Fred has memories (or a memory) of Tammy eating the taco.
(IV) Fred’s knowledge that Tammy ate the taco is stored in his memory.
(V) Fred has excellent memory.
In this chapter, I will present and explore various arguments for skepticism that can arise
when considering these qualities.2 My focus will be on the aspects of the arguments that are
unique to memory, which are not shared, for example, by the more often explored skepti-
cal arguments related to perception. I will also point out ways that the two types of skeptical
arguments differ.
This exploration will be opinionated, meaning that I will give my own perspective on the
arguments. However, I do not pretend to be giving the last word or providing an exhaustive
treatment. I will instead often provide some of the key points in the dialectic, leaving references
in the endnotes for further research and study.
335
Andrew Moon
2. Memory as faculty
Global skeptical memory argument

In sentence (V), “Fred has excellent memory,” the word “memory” picks out a faculty or
power of the mind, which allows Fred to have the sort of qualities mentioned in sentences
(I)–(IV). With memory, we store, retain, and recall our knowledge of the world. Suppose
I learn, via the class schedule, that my class begins at 3:00 p.m. It is by memory that I store
this knowledge, retain it, and recall it later in the day when I remember that my class begins
at 3:00 p.m.
Here, a skeptical threat arises. How do I know that my memory reliably performs these func-
tions? And if I don’t know that it does, how can I know anything on the basis of my memory?
I will state these questions more forcefully as an argument:
(1) I can know a proposition on the basis of a faculty only if I know that faculty is reliable.
(2) I cannot know that my memory is reliable.
(3) Therefore, I cannot know a proposition on the basis of my memory.
Call this a global skeptical memory argument. It is global since it involves the reliability of my mem-
ory as a whole. The later local skeptical memory arguments are only about the reliability of memory
in certain circumstances or domains. I will generally present my skeptical arguments using the
pronoun “I,” with the assumption that they can also apply to you, the reader. After presenting
some preliminary reasons to believe premises (1) and (2) in this section, I will engage in deeper
discussion of the argument in the following sections.
In favor of premise (1), suppose someone makes a claim about health, and you ask him where
he got his information. He answers, “A random website on the Internet! And I don’t know
whether it’s reliable.” Perturbed, you would conclude that he does not know that the health
claim is true. More generally, it at least initially seems that you cannot know a proposition on
the basis of a source (or faculty) if you do not know it is reliable.
Regarding premise (2), why think we cannot know that our memory is reliable? I might
try to prove that I can know my memory is reliable by thinking about the many times it
delivered correct information in the past. There was that time I aced my history exam and
that other time that I recited a poem. Aren’t these good grounds for believing my memory
is reliable?
No, an objector might say, such a belief would be epistemically circular. One’s belief that
faculty F is reliable is epistemically circular if one used F to come to believe that F is reliable. In
this case, I used memory to justify my belief in the reliability of my memory. And this circular-
ity, the objector claims, makes the belief unwarranted. (Epistemic circularity is different from
logical circularity, which, roughly, is a property of arguments that include the conclusion as one
of the premises.)
Here is the argument for (2):
(2a) Any reasoning in favor of the conclusion that one’s memory is reliable will
either result in epistemic circularity or not be sufficient to justify the belief.
(2b) If (2a) is true, then I cannot know that my memory is reliable.
(2) Therefore, I cannot know that my memory is reliable.
I will examine this argument in the following sections.
336
Skepticism and memory
Discussion of premise (2a)

We have seen that arguing for the reliability of memory by appealing to its past performances
will result in epistemic circularity. But perhaps I can use present perceptual knowledge to confirm
whether a prediction of my memory is correct; this could provide noncircular justification for
believing that my memory is reliable. In Roy Harrod’s (1942: 65) example, he uses his memory
of past observations of lightning to predict the ending of a presently seen flash of lightning.
As his memory predicted, the lightning ended! His perceptual knowledge—his seeing that the
lightning ended—provided, without any circularity, some justification for believing that his
memory is reliable.
However, a single correct output from memory will not be sufficient to justify the belief
that his memory is generally reliable; this would be a hasty generalization. Suppose, then, that
he tries to provide additional justification for believing that his memory is reliable by way
of a second correct prediction. This will result in epistemic circularity, since now he would
be forced to remember the previous success of his memory with the lightning. Predictions of
memory, therefore, can provide only a small amount of justification for noncircular belief that
one’s memory is reliable.3
Instead of appealing to prediction, perhaps we can appeal to inference to the best explanation.4
We are often justified in believing the best explanation of the data when it is a good one. For
example, I am justified in believing that Timmy ate the cookies because that is a good, and the
best, explanation of the missing cookies and the crumbs on his hands. Now, what is the data in
our case? They are the facts one can know without memory, say, what I know by introspection
and perception. I can introspect that I have many memories and beliefs about the past, and I
can perceive many present facts about the external world. The best, and a good, explanation for
this coherent set of facts is that I have interacted with this world in the past, and these memories
and beliefs are mostly accurate. In other words, the best explanation is that my memory is reliable.
Here is the inference to the best explanation argument:
(1) The hypothesis that my memory is reliable is a good, and the best, explanation of both (a) the
facts about the world that I know via perception, and (b) the facts about my memories and
beliefs that I know via introspection.
(2) Therefore, my memory is probably reliable.
Unfortunately, premise (1) is doubtful. At any moment, the facts that I know only via percep-
tion and introspection are very few. I introspect and “see” that I have a memory of my tenth
birthday. At the same time, I see that there is a computer. It is implausible that my memory is
reliable is a good explanation for there is a computer and I have a memory of my tenth birthday; they
are too unrelated. Now, perhaps there are a few more facts that I know, at that moment, on
the basis of only introspection and perception. Unfortunately, it seems that they will not be suf-
ficient to justify thinking that the proposition that my memory is reliable is a good explanation of
them. There are just too few facts known via perception and introspection in one instant, what
philosophers call the specious present.5
Suppose I did have a powerful enough mind so that I could perceptually and introspectively
know a plenitude of facts all at once, so that my memory is reliable was in fact a good explana-
tion for them. It would still be another step to know that it is the best explanation of those
facts, which would require considering alternative explanations and evaluating them (by standard
criteria such as simplicity and explanatory power). Unfortunately, this process would require
memory. For example, a moment after reflecting, I might have to recall (a) the original facts and
337
Andrew Moon
explanations, (b) which explanation I was considering, or perhaps (c) what I am even doing in
the first place. So, the inference to the best explanation argument against (2a) fails.6
Having blocked two attacks on (2a), I will formulate an argument in favor of it.
(2a′) Any reasoning in favor of the conclusion that one’s memory is reliable will
either be so intellectually sophisticated that it requires the use of memory, or so
simple that it cannot justify believing it.
(2a′′) If (2a′), then (2a).
(2a) Therefore, any reasoning in favor of the conclusion that one’s memory is reliable
will either result in epistemically circularity or not be sufficient to justify the belief.
We saw above that those who use the predictions argument or the inference to the best explanation
argument to attack (2a) only avoided one horn of (2a′) at the cost of getting impaled by the
other. I advise future critics of (2a) to focus on avoiding the two horns mentioned in (2a′).
We can see how (2a) raises different issues than would a premise in an analogous argument
that concludes that we cannot know that our perception is reliable.7 Yes, it is difficult to argue,
in a noncircular way, that one’s perception is reliable, but one can at least engage in intel-
lectually sophisticated chains of reasoning.8 For example, inspired by Descartes, one might
formulate an ontological argument for God’s existence and then argue that since God is not
a deceiver, one’s perception is reliable. Unfortunately, such reasoning would require the use
of memory and thereby be unavailable for those who want a good, noncircular argument that
their memory is reliable.
Many now know just how easy it is to fall prey to epistemic circularity. However, we must
be careful not to over-attribute the use of memory. Consider the following quote by Ted Poston:
How long does it take you to read this sentence? Did you rely on memory at all in
reading that sentence? What is the most complex thought you can entertain with-
out relying on memory at all? These questions raise a fundamental epistemological
issue concerning our ability to justify our extensive reliance on memory. Nearly
every thought relies on memory. Even simple thoughts we entertain in the fleeting
present—e.g., “green here now”—rely on our apparent memory that the mean-
ings of our terms are constant and that the ‘I’ which now thinks is the same ‘I’ that
thought a moment ago.
(Poston 2016: 183)
It is not obvious that having the thought “green here now” relies on memory in the way Poston
suggests. Suppose a boy looks at a green object and has the thought expressed by “green here
now.” Does he also need an apparent memory that the meanings of his terms are constant and
that the “I” which now thinks is the same “I” that thought a moment ago? No. This is overin-
tellectualization; the 6-year-old has no such memories.
The following remarks will help us to make correct attributions of memory and avoid circu-
larity. The sort of memory that is the focus of this chapter, the sort attributed by the sentences
(I)–(IV), are all undergirded by a single faculty, what some psychologists call “declarative mem-
ory.” Some psychologists also use the word “memory”—and more specifically, “nondeclarative
memory”—to pick out the mechanisms responsible for simple classical conditioning, priming,
procedural skills, and more. Perhaps the boy needs some nondeclarative memory to have the
thought expressed by “green here now.” Now, Michaelian (2010) has argued that there is no
single faculty (or natural kind) memory, that unites declarative and nondeclarative memory; they
338
are distinct systems. If he is right, then although one falls into circularity when one uses declara-
tive memory to believe that my declarative memory is reliable, one might avoid circularity if one
only uses nondeclarative memory. (For the rest of this chapter, I will continue to use “memory”
to pick out only declarative memory, that which undergirds the qualities attributed by (I)–(IV).)
How, then, do we know when declarative memory is being used? If the type of quali-
ties attributed in (I)–(IV) can be correctly attributed, then declarative memory is being used.
Unfortunately, exploring this question in further detail is beyond the scope of this chapter. We
can see, then, how discussions of the nature of memory are relevant to discussions of epistemic
circularity, and hence, to our skeptical argument.9
Discussion of premise (2b)

To sum up, the criticisms of (2a) have decent replies, and I have also mounted a positive argu-
ment for (2a). Let us now turn to (2b). It states:
2b) If any reasoning in favor of the conclusion that one’s memory is reliable will
either result in epistemically circularity or not be sufficient to justify the belief,
then I cannot know that my memory is reliable.
To counter it, we must think of how I might know my memory is reliable, even if any reasoning
in favor of this conclusion will either result in epistemic circularity or be insufficient to justify
the belief. Here are two potential ways.
First, perhaps I know that my memory is reliable noninferentially; I need not know it on the
basis of reasoning. Indeed, this option is psychologically plausible. Most people do not reason to the
conclusion that their memory is reliable. It’s something they just believe without any reasoning.
But why think that this noninferential belief is warranted? An epistemological theory could
provide the explanation. According to a simple version of externalist reliabilism, a belief is war-
ranted in virtue of being produced by a reliable process.10 Thus I could say that my noninferential
belief that my memory is reliable is warranted not in virtue of a reasoning process, but in virtue
of the reliability of the process producing the belief. Perhaps we evolved reliable faculties that
produce noninferential beliefs about the reliability of our memory, and none of those faculties
use memory. Although I chose reliabilism for my illustration, other epistemological theories
could also undergird an attack on (2b). For example, according to proper functionalism, a belief
is warranted if and only if it is produced by properly functioning, truth-aimed, reliable faculties
in the right sort of environment. A proper functionalist could tell a similar story of how we have
such a noninferential belief.11
One could argue that even if the belief is noninferentially formed, it is still epistemically
circular. Arguably, even to form the belief that my memory is reliable, I need the concepts of
memory and of reliability, and this will require having some knowledge of what memory
and reliability are. But this knowledge will presumably be stored in my memory. Hence, I
am using memory in the process of forming my noninferential belief after all. Each premise
of this argument could be further explored: that having the belief that memory is reliable
requires concept possession, that concept possession requires background knowledge, and
that memory is involved in this knowledge.12
The second objection to (2b) is one I endorse. It questions whether circularity necessarily
renders a belief unwarranted. Let the term “benign circularity” indicate circularity that does not
make a belief unwarranted, and “malignant circularity” to indicate circularity that does.13 The
objection has two steps. The first step develops a plausible theory of epistemic circularity that
339
Andrew Moon
specifies conditions under which a circular belief is benign.14 The second step shows that the
belief that one’s memory is reliable can meet those conditions.
I will use Bergmann’s (2006) theory for the purpose of illustration. He thinks that if, prior
to the formation of the epistemically circular belief that faculty F is reliable, the believer already
seriously doubted, or should have been seriously doubting, the reliability of F, then this circular-
ity is malignant.15 Otherwise, it is benign (2006: 199). For example, suppose I already seriously
doubt (or should seriously doubt) that my memory is reliable. Then any use of memory to try to
justify my belief that my memory is reliable will result in malignant, circular belief. Such a belief
is unwarranted. One reason we are inclined to think that all circular belief is malignant is because
we are normally focusing on cases where one has, or should have, serious doubt (2006: 200).
On the other hand, suppose I never seriously doubted, nor should have seriously doubted,
that my memory is reliable. Then I can use my memory to recall my track record of suc-
cessful memory performance and form the belief that my memory is reliable. Or I might just
noninferentially believe that my memory is reliable, making use of the concepts of memory
and reliability that are stored in my memory. These ways of forming the belief that my memory
is reliable might be circular, but according to Bergmann’s theory, the circularity is benign.
Furthermore, with circularity no longer an issue, nothing stands in the way of these beliefs
being warranted, so long as the other conditions for warrant are met. (These other conditions
will be determined by whatever is the true theory of warrant, whether it be a reliabilist view,
a proper functionalist view, or something else.)
A number of theories allow for benign circularity.16 Since circularity so easily infects beliefs
about memory’s reliability—more easily than beliefs about perception’s reliability—it is surpris-
ing how so little has been written that applies these theories to discussions about skepticism and
memory.17 I hope I have helped fill that lacuna. Lastly, although I find such theories plausible, I
still find moving the following quote by Richard Fumerton:
All of this will, of course, drive the skeptic crazy. You cannot use perception to jus-
tify the reliability of perception! You cannot use memory to justify the reliability
of memory! . . . Such attempts to respond to the skeptic’s concerns involve blatant,
indeed pathetic, circularity.
(Fumerton 1995: 177)
I confess that I do still feel something illegitimate about epistemic circularity, and there is more
to be said about this issue. I recommend that those who defend skeptical arguments about
memory to make as their targets the theories that allow for benign circularity. If they win that
battle, then I believe that there is little hope for the believer in the reliability of memory.18
Discussion of premise (1)

Can I know a proposition on the basis of a faculty only if I know that faculty is reliable? Not
knowing the reliability of the Internet source seemed to prevent my friend from knowing the
health claim. However, the fact that he couldn’t know in that particular case does not mean that
no believer could ever know on the basis of a source that she does not know is reliable. Think
of children’s knowledge. It seems that little Timmy could know that he sees cookies even if he
doesn’t know that his perception is reliable; he might have never given it a thought.19
Epistemologists have suggested ways to honor the intuitions in both the Internet case and the
Timmy case. A random source from the Internet has a significant chance of being unreliable.
We thereby have reason to doubt its reliability. Now consider:
340
(1′) I can know p on the basis of a faculty only if I don’t have good reason to doubt
that the faculty is reliable.
Since condition (1′) is not met in the Internet case, we cannot know on the basis of the
website. On the other hand, although little Timmy doesn’t know that his perception is reli-
able, he has also never gained a good reason to doubt it. In accordance with (1′), Timmy can
still know that he is receiving cookies. Similarly, unless we have gained good reason to doubt
that our memory is generally reliable, which we have not, then (1′) allows that we could know
on the basis of memory.20
Local skeptical memory arguments

Both premises of the global skeptical memory argument that we discussed are subject to serious
objections. But aren’t there circumstances in which we cannot know on the basis of our mem-
ory because we have good reason to doubt our memory’s reliability in those circumstances?
Suppose you were inebriated at the philosophy party last night. As you remember things, you
were the star, dazzling everybody with your wit and philosophical acumen. However, you also
know from past experience that alcohol affects your memory in such a way that, the next morn-
ing, you end up with an inflated view of your social interactions. You realize that you meet
these conditions. It seems that you cannot know that you were the hit of the party.
In addition to anecdotal evidence, the scientific study of memory can reveal circumstances
in which our memories are unreliable. For example, in one experiment, subjects who studied
a list of words connected by a theme would later falsely recall the theme word being on the
list.21 If the subjects studied words such as “bed,” “rest,” “tired,” and “dream,” then they would
later incorrectly recall “sleep” being on the list. Now, suppose someone finds himself in such a
context, where he is memorizing related words in a scientific study, and then he finds himself
“recalling” the theme word in the list. Given that he knows of such studies, he would not be
warranted in believing that the theme word was on the list.
In these examples, one has reason to doubt that one’s memory is reliable in a particular
circumstance. However, one need not doubt the reliability of all of one’s memory in those
circumstances. Surely, you can still remember that you were at a party, that some people were
there, and that you drank some alcohol. The reliable production of those beliefs is not called into
question. In the scientific study, you can still remember what your name is and that you are in
a lab. So, not only is the unreliability of the faculty in those circumstances relevant, but also its
unreliability within a certain domain.
We can now formulate a schema for local skeptical memory arguments:
(1′′) If I have good reason to doubt the reliability of a faculty’s belief outputs in cer-
tain circumstances in a certain domain, then such a belief is not warranted.
(2) I have good reason to doubt the reliability of memory’s belief outputs in certain
circumstances in a certain domain.
(3) Therefore, such a belief (that mentioned in premise (2)) is unwarranted.
I argued above that the global memory skeptical argument is subject to serious objections. Yet local
memory skeptical arguments are likely successful in some cases. Furthermore, we can now see how
the scientific study of memory can aid us in formulating plausible skeptical arguments. It is very
unlikely that science will ever show that our memory overall is unreliable, but it will certainly be
able to specify the unreliability of memory in certain circumstances and domains.
341
Andrew Moon
It remains to be answered why the relevant “certain circumstances” and “certain domain”
include what they do. In the party case, the relevant circumstance was being inebriated and the
domain was beliefs about my social performance. But why didn’t the domain also include your beliefs
about whether you were at the party? Why didn’t the relevant circumstances also include that
you were breathing air? Exploring the interesting question of how to determine the relevant
domain and circumstances is unfortunately beyond the scope of this chapter. Fortunately, we
do have an intuitive grasp of what the relevant circumstance and domain are in many cases, and
we will have to settle for that.
3. Memory as storage unit

Recall sentence (IV), “Fred’s knowledge is stored in memory,” and sentence (V), “Fred has
excellent memory.” Sentence (V) uses “memory” to denote a faculty. Sentence (IV), on the
other hand, uses “memory” to refer to something like a container in which knowledge and other
information can be kept. I could use my memory (the faculty) to store information in memory
(the container). Of course, memory is not literally a container.22 It does not hold information,
beliefs, or knowledge, like a box holds its contents. In fact, it is not obvious what is going on
when one stores knowledge in one’s memory. Fortunately, we do still understand what people
mean by sentences like (IV), and that is enough to understand this section.
Here is an argument that many of the beliefs in our memory that we think are warranted are
not. Plausibly, Tim, an adult raised in the United States, knows that Columbus sailed in 1492,
even if he has completely forgotten where or how he learned it. The belief is just stored in
his memory. However, it seems that the initial evidence upon which Tim based his belief no
longer exists. But warranted belief requires that the belief be based on evidence. Therefore,
Tim’s stored belief that Columbus sailed in 1492 is not warranted after all. Here is the form of
this argument:
(1) If a belief is not based on good evidence, then it is not warranted.

(2) Tim’s stored belief that Columbus sailed in 1492 is not based on good evidence.
(3) Tim’s stored belief that Columbus sailed in 1492 is not warranted.
Such an argument could be applied to many of our stored beliefs.23

Interestingly, such a skeptical argument has rarely, if ever, been formulated. The reason is that
many find the denial of the conclusion to be far more plausible than any of the premises. In fact,
some use the intuitive plausibility of the denial of (3) to then form an argument to attack (1)!
~(3) Tim’s stored belief that Columbus sailed in 1492 is warranted.

2) Tim’s stored belief that Columbus sailed in 1492 is not based on good evidence.
~(1) It’s not the case that if a belief is not based on good evidence, then it is not war-
ranted.24
Another option is to deny premise (2) by saying that the belief is based on evidence. The chal-
lenge for such a denier would be to find a plausible candidate for what that evidence might be.
Perhaps Tim has other stored beliefs that could serve as his evidence. Or maybe Tim’s belief is
still based on the evidence that exists in the past, even if that evidence no longer exists now. A
detailed exploration of these candidates is beyond the scope of this chapter.25
342
4. Memories and Russellian skepticism
A difference between perceptual beliefs and memory beliefs

Recall sentences (I) “Fred remembers that Tammy ate the taco,” (II) “Fred remembers Tammy
eating the taco,” and (III) “Fred has a memory of Tammy eating the taco.” Sentences like (II)
and (III) typically refer to events the person experienced in the past; this is not so for (I).26 For
example, I can remember that Socrates taught Plato. But I certainly—and unfortunately—do
not remember Socrates teaching Plato, nor do I have any memories of Socrates teaching Plato.
Memories are of events we have experienced or felt that we experienced; they might not be
accurate.27 Suppose Fred has memories of Tammy eating the taco. Even if Tammy didn’t actu-
ally eat the taco, Fred might still have memories of her eating the taco. They would just be false
memories. Similarly, we could say that Fred remembers Tammy eating the taco, even though she
didn’t actually eat it; he just remembered incorrectly. Some might prefer to call such memories
apparent memories and say that Fred only seemed to remember Tammy eating the taco. I believe
that ordinary English does not require adding these qualifiers, so I will not add them, although I
invite those who think they are required to do so.28
We sometimes base our beliefs on our memories. Consider a witness trying to recall a
crime scene:
Detective: “Try to remember the color of the car. What color was it?”
Witness: “Umm . . . Let’s see, I remember it being maroon.”
Detective: “You have a memory of it being maroon?”
Witness: “Yes!”
Plausibly, the witness can come to believe that the car was maroon on the basis of her memory of
the car being maroon. In her mind’s eye, she might picture a memorial image of the car being
maroon and then believe on that basis. However, much knowledge stored in memory is not
based on memories in this way. While he is asleep, Fred knows that Columbus sailed in 1492,
and this knowledge is not based on any of his memories.29 And depending on how one interprets
the arguments of the previous section, the knowledge might not be based on any evidence at all.30
Perceptual beliefs, on the other hand, are typically or always based on sensory experiences.31
My belief that that house is yellow is based on my visual experience, the sort one has when see-
ing a yellow house. Epistemologists then exploit the gap between how things appear and how
things are, the appearance-reality gap, to formulate skeptical arguments.32 For example:
1) We cannot know that a demon is creating sensory experiences in us with no cor-

responding external world.
2) If (1), then we do not know anything about the external world.
3) Therefore, we do not know anything about the external world.
Call this the perceptual skeptical possibility argument. Such an argument is toothless in the case of
memory:
4) We cannot know that a demon implanted false memories into our minds, with no
past that corresponds to those memories.
5) If (4), then we do not know anything about the past.
6) Therefore, we do not know anything about the past.
343
Andrew Moon
The key difference between the two arguments is the justification of their second premises.
Regarding premise (5), we have a way of knowing about the past that does not depend on our
memories. It is our stored knowledge. Fred’s stored knowledge of his phone number and the name
of his hometown, for example, does not depend on his memories. On the other hand, apart from
our sensory experiences, we have little to no knowledge about the external world. So, a difference
between these two arguments is that one is driven by an appearance-reality gap, and the other is not.
Russellian skepticism
This “victory” is short-lived. There is still a skeptical argument about the past that is at least
as challenging as the perceptual skeptical possibility argument. Consider the following famous
quote by Bertrand Russell:
There is no logical impossibility in the hypothesis that the world sprang into being five
minutes ago, exactly as it then was, with a population that “remembered” a wholly
unreal past. There is no logically necessary connection between events at different
times; therefore, nothing that is happening now or will ever happen in the future can
disprove the hypothesis that the world began five minutes ago.
(Russell 1921: 159)
The earlier argument suggested the possibility that false memories were implanted in me by an
evil demon. Russell’s scenario contains the possibility that the beliefs stored in my memory were
also implanted and false. So, the typical skeptical argument about the external world envisages a
scenario in which my perceptual evidence is the same but my beliefs are false. In the case where
the belief is not based on any evidence, Russell challenges us with the possible scenario in which
the belief is false; no mention of evidence is needed because there is none. This is a structural
difference in the perceptual and memory arguments.
Let us turn Russell’s evocative quote into an argument:
7) I cannot know that Russell’s hypothesis did not occur.

8) If (7), then I do not know anything about the past.
9) Therefore, I do not know anything about the past.
Call this the memory skeptical possibility argument.
Responses
Much has been written about the perceptual skeptical possibility argument that easily translates
over to the memory skeptical possibility argument. Since little here is unique to the memory
argument, I will provide only a brief discussion.
One might deny the closure principle that undergirds premise (8). A version of it states,
(Closure Principle): If I know that p, and I know that p entails q, then I am in a position
to know q.
Equivalently, if I am not in a position to know q, and I know that p entails q, then I do not know
that p. Thus, since I cannot know that Russell’s hypothesis did not occur, and I also know that
344
this entails there is no past (as I believed it to be), it follows that I don’t know anything about
the past (as I believed it to be). Although the Closure Principle seems very plausible, a number
of objections have been raised against it.33
Another option is to deny premise (7) by arguing that we can know that Russell’s scenario
did not occur. Perhaps we can know this noninferentially by way of a reliable common sense
mechanism that is responsible for getting us to deny seemingly absurd claims.34 One could also
argue that, contrary to appearances, Russell’s scenario is in fact incoherent or impossible; hence,
we can know that it did not occur.35
Lastly, perhaps one could argue that Russell’s scenario did not occur by appealing to infer-
ence to the best explanation. One could argue that the existence of an ordinary past is a far
better explanation of our current sensory experiences, beliefs, and memories, than Russell’s
hypothesis. It would be better because it contained more virtues, such as simplicity, parsimony,
and explanatory scope. It would be interesting to see both whether this could be successfully
argued and also whether the argument would be substantially different from an argument that
the hypothesis that there is an ordinary external world is a better explanation than the hypothesis
that a demon is beguiling us.36 This project has yet to be explored.
There is also the contextualist response. It states that “knows” means different things in differ-
ent contexts.37 When skeptical scenarios are raised, the evidential standards required for “knows”
to correctly apply are very high. On such a meaning of “knows,” all of sentences (7)–(9) can be
true. But suppose skeptical scenarios are not raised. Then the evidential standards required for
“knows” to correctly apply are lower, and so (9) can be false.38
5. Conclusion
A number of epistemologists have noted the lack of attention to the topic of skepticism and
memory in the philosophical literature.39 I hope that this chapter helps to fill that void and also
provides new avenues of research to explore.
Acknowledgment
Thanks to Sven Bernecker, Brent Madison, Kevin McCain, Matthew Frise, and Matthew
McGrath for helpful comments on earlier drafts.
Notes
1 See Chapter 1.
2 See Frise (2015: Sect. 4) and Locke (1971: 103–37) for additional discussions of skepticism and memory.
3 This was my shorter version of an argument inspired by Harrod (1942: 65). See Locke (1971: 107–11),
and Bernecker (2008: 99–100) for discussion of Harrod’s argument. My criticism is similar to Locke’s
(1971: 110).
4 This is my shorter version of an argument inspired by Brandt (1955).
5 This criticism follows Frise’s (2015: Sect. 4a) response to Brandt’s argument.
6 The problem of having to use memory to recall earlier premises in one’s reasoning has been pointed
out by Plantinga (1993: 61–4), Bernecker (2008: 86), Senor (2009: Sect. 5), and Poston (2016). See the
Plantinga reference for more on the sort of inference to the best explanation argument I am discussing.
7 Cf. Frise (2015: Sect. 4a).
8 See Alston (1993) for why they are still all likely to fail.
9 For more on the nature of memory, see Chapter 1.
10 See Sections 1 and 2 of Steup (2005) for an introduction to reliabilism.
345
Andrew Moon
11 See Plantinga (1993) for a book-length defense of this view.

12 Thanks to Sven Bernecker for helping me clarify this paragraph. The argument is inspired by McCain
(2014: 37–8).
13 This is Bergmann’s (2006: 198) terminology, although he uses “justification” instead of “warrant.”
14 This step will require making it plausible that benign circularity is possible. For two detailed arguments
that it is, see Bergmann (2006: 184–97).
15 Bergmann’s theory entails, as I assume, that there are some doxastic attitudes we epistemically ought to
hold. For example, if I believe that p and I receive evidence against p, then I should seriously doubt p.
Bergmann (2006: 189–200) defends a proper functionalist theory of the epistemic ought, although one
could embrace other theories and still accept his account of malignant circularity.
16 For example, see van Cleve (1984), Alston (1989: Ch. 12), Markie (2005), Bergmann (2006: Ch. 7), and
Sosa (2009: 195–204).
17 An exception is Steup (2013).
18 For more on epistemic circularity, see previous two notes and Lammenranta (2015). For more on the
reliability of memory, see Locke (1971: Ch. 12) and Frise (2015: Sect. 4a).
19 Cf. Alston (1989: 164).
20 Higher-level requirements on knowledge, like the one in premise (1), are no longer very popular. For a
recent defense of a higher-level requirement on justification, see Smithies (2012).
21 See Roediger and McDermott (1995). Thanks to Matthew Frise for helpful discussion.
22 Cf. Frise (forthcoming).
23 For more on arguments involving forgotten evidence, see Frise (2015: Sect. 3.a.i), Frise (forthcoming),
and Chapters 22 and 23 of this volume.
24 See the examples in Section 3.a.iii, “The Problem of Stored Beliefs,” in Frise (2015).
25 I explore such candidates in a similar case in Moon (2012). See also Frise (forthcoming) and McCain’s
(2014: 146–9) reply to Moon.
26 Some use “personal memory,” “experiential memory,” and “episodic memory” for the type of qual-
ity attributed to Fred in (II) and (III), and “factual memory,” “propositional memory,” and “semantic
memory” for the type of quality attributed to him in (I).
27 This is unlike remembering a proposition, as in sentence (I). If Tammy didn’t eat the taco, then Fred
didn’t really remember that Tammy ate it. One remembers that p only if p is true. For a detailed defense,
see Bernecker (2008: Ch. 8).
28 Consider that neither “Old man Nelson remembers the fish being this big, but he’s certainly wrong” nor
“Old man Nelson has a memory of the fish being this big, but he’s certainly wrong” seem inconsistent.
29 See Locke (1971: 37–8) for further defense of this point.
30 For discussion of these points as they relate to phenomenal conservatism, see Chapter 24.
31 Plantinga (1993: 62, 188) explores this contrast between memory and perception.
32 Fumerton (1995: 31) says this is a “recurring pattern” in skeptical arguments.
33 Bernecker (2008: 181–3) objects to it and Frise (2015: sect. 4b) defends it. See also the sections on clo-
sure in Steup (2005) and Klein (2015).
34 Bergmann (2006: 206–11) defends a view like this by drawing from Thomas Reid.
35 See Locke (1971: 120–31) for a few arguments for this.
36 McCain’s (2014: 125–42) response to skepticism could likely be applied here.
37 For more, see the contextualism sections in Steup (2005) and Klein (2015).
38 For a summary of some debates about Russell’s hypothesis in the mid-1900s, see Locke (1971: Ch. 11).
39 See Fumerton (1995: 34) and Frise (2015: sect. 3)
References
Alston, William 1989. Epistemic Justification. Ithaca, NY: Cornell University Press.
Alston, William 1993. The Reliability of Sense Perception. Ithaca, NY: Cornell University.
Bergmann, Michael 2006. Justification without Awareness. New York: Oxford University Press.
Bernecker, Sven 2008. The Metaphysics of Memory. Dordrecht: Springer.
Brandt, Richard 1955. “The Epistemological Status of Memory Beliefs.” Philosophical Review, 64: 78–95.
Frise, Matthew 2015. “The Epistemology of Memory.” In The Internet Encyclopedia of Philosophy, www.
iep.utm.edu/epis-mem/.
Frise, Matthew forthcoming. “Eliminating the Problem of Stored Beliefs.” American Philosopical Quarterly.
346
Fumerton, Richard 1995. Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield Publishers,
Inc.
Harrod, Roy 1942. “Memory.” Mind, 51: 47–68.
Klein, Peter 2015. “Skepticism.” In The Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/
entries/skepticism/#ArgForAcaSkeEmpCloPri (June 2, 2015 version).
Lammenranta, Markus 2015. “Epistemic Circularity.” In The Internet Encyclopedia of Philosophy, www.iep.
utm.edu/ep-circ/.
Locke, Don 1971. Memory. Garden City, NY: Anchor Books.
Markie, Peter 2005. “Easy Knowledge.” Philosophy and Phenomenological Research, 70: 406–16.
McCain, Kevin 2014. Evidentialism and Epistemic Justification. New York: Routledge.
Michaelian, Kourken 2010. “Is Memory a Natural Kind?” Memory Studies, 4: 170–89.
Moon, Andrew 2012. “Knowing Without Evidence.” Mind, 121: 309–31.
Plantinga, Alvin 1993. Warrant and Proper Function. New York: Oxford University Press.
Poston, Ted 2016. “Acquaintance and Skepticism about the Past,” in Intellectual Assurance: Essays
on Traditional Epistemic Internalism, eds. Brett Coppenger and Michael Bergmann. Oxford: Oxford
Roediger, H.L. and McDermott, K.B. 1995. “Creating False Memories: Remembering Words Not
Presented in Lists.” Journal of Experimental Psychology: Learning, Memory, and Cognition, 24: 803–14.
Russell, Bertrand 1921. The Analysis of Mind. London: Allen and Unwin.
Senor, Thomas D. 2009. “Epistemological Problems of Memory.” In The Stanford Encyclopedia of Philosophy,
http://plato.stanford.edu/entries/memory-episprob/ (September 4, 2009 version).
Smithies, Declan 2012. “Moore’s Paradox and the Accessibility of ‘Justification.’” Philosophy and
Phenomenological Research, 85: 275–300.
Sosa, Ernest 2009. Reflective Knowledge: Apt Belief and Reflective Knowledge. Oxford: Oxford University Press.
Steup, Matthias 2005. “Epistemology.” In The Stanford Encyclopedia of Philosophy, http://plato.stanford.
edu/entries/epistemology/#SKE (December 14, 2005 version).
Steup, Matthias 2013. “Is Epistemic Circularity Bad?” Res Philosophica, 90: 215–35.
van Cleve, James 1984. “Reliability, Justification, and the Problem of Induction.” Midwest Studies in
Philosophy, 9: 555–67.
347
PART VIII
Memory and morality

27
A DUTY TO REMEMBER
Jeffrey Blustein
1. Conceptual clarification
The notion of a “duty to remember” has gained currency in the West in recent decades, a
development that can be traced in no small measure to the role it has played in reflections on
how to respond to the Nazi extermination of Jews, Gypsies, and other so-called undesirables
during World War II. It is to the Holocaust that we must inevitably turn when speaking of such
a notion, for it has largely acquired moral and political traction because of it. When a duty to
remember with this antecedent is invoked, as has all too frequently been the case since the end
of that war, it is characterized by the following features: it is incumbent on humanity collec-
tively; it targets the victims of genocide or other atrocity; and its purpose, to a significant degree,
is to put the world on guard against a repetition of crimes of this magnitude.
On further reflection, however, we can identify different sorts of cases that do not fit this
mold but that still under some reasonable construal exemplify a duty to remember, and a discus-
sion of this duty should be capacious enough to include them. This will reveal that a duty to
remember is not confined to extraordinary states of affairs but arises as well in ordinary circum-
stances of public and private life. It will also show historical continuities and shifts of meaning
between different instantiations of the duty that will provide a fuller picture of its nature and
moral basis. To begin to suggest something of this breadth, I will analyze the concept of a duty
to remember by posing several questions.
(i) In what sense is it a duty?

The notion of duty might be used interchangeably with obligation in this context, so that it is a
matter of indifference whether one speaks about a duty to remember or an obligation to remember.
Alternatively, the two might be distinguished on the basis that obligations have their source in
consent whereas duties do not. (I will continue to use the expression “duty to remember” none-
theless.) Thus, it might be preferable to speak about an obligation to remember rather than a duty
to remember, if we have promised someone that we will remember him after his death and keep
his memory alive through activities that honor him. But in some uses of the expression “duty to
remember,” no such voluntary act on our part is plausibly implied. For example, arguably human
beings ought to remember the Holocaust and its victims irrespective of their voluntary acts.
351
Jeffrey Blustein
The common element among all particular conceptions of a duty to remember is that
remembrance is morally imperative, non-optional, or not morally elective, and that omitting
to remember whatever it is that one is obligated to remember is strongly morally criticizable.
Whether or not it is explicitly referred to as such, and whatever the reasons for regarding
remembrance in this manner, there is a “duty” if it meets these conditions.
However, perhaps this is too hasty, for there is an objection that will immediately strike some
as compelling, concerning the very coherence of the idea of a duty to remember. The objection
is that remembering, like forgetting, is just not the sort of phenomenon to which the notion of
duty attaches. This is because it is something that for the most part happens to us, not something
over which we exert control and that is a product of deliberate choice. In this respect, it might
be said, it is like emotions: we cannot summon up sadness or anger at will nor, similarly, can we
remember at will. But since the capacity to exercise voluntary control over our memories is a
necessary condition of being morally responsible for them, we cannot be morally responsible for
our memories, so cannot have duties to remember.
Without entering into the metaphysical debate about whether we are “really” morally
responsible for our memories, it is evident that we ordinarily believe we have considerably
more control over our memories, and incidentally over our emotions as well, than the objec-
tion claims. (Emotions can be subject to various sorts of emotion regulation techniques.) One
way to show this is to examine our reactions to people’s failures and successes with respect to
memory. If, for example, you have promised to do something important and fail to follow
through as appropriate, claiming that you just forgot, we are not likely to accept this at face
value. Whatever excuses you may offer might not be sufficient to exonerate you, for there
are circumstances in which the failure to remember cannot be excused. Our reactions to suc-
cessful remembering are also instructive. If, for example, you remember to attend your son’s
birthday party despite having many competing responsibilities and distractions, you might be
commended for not forgetting when it was so easy and tempting to do so. In short, terms of
blame and praise are commonly used in connection with our memories, and we wouldn’t react
to others in these ways if we didn’t hold them morally responsible for their memories.
(ii) What or whom is there a duty to remember?

If, as is widely believed, there is a duty to remember the victims of injustice and atrocity, then
for related reasons there is a duty to remember those who resisted these crimes and took per-
sonal risks to rescue those who were in harm’s way. However, even when the scope of the duty
is enlarged in this way, a duty to remember seems to encompass much more than a duty to
remember the victims or opponents of wrongdoing.
For example, it is commonly thought that we ought to remember persons who, through
some extraordinary act of dedication or compassion for the suffering of others, have made
significant contributions to the betterment of humankind. In addition, friends are supposed to
have a duty to remember friends who have died and to place them at the center of events in
which they importantly figured; children to remember deceased parents who lovingly cared
for and nurtured them; students to remember teachers who had a profound influence on their
development, and so on. The primary reason offered for the duty in these cases would likely be
that remembering them is just what one ought to do as an integral part of the caring relationship
that does not end with their deaths. On this view, remembering is a manifestation of ongoing
caring. (If they were also the victims of wrongdoing, this would ground an additional reason to
remember them, but under a different description.) We also seem to have a duty to remember
persons to whom we have promised that we will not forget them, as mentioned before, or will
352
A duty to remember
carry out their wishes after they die, even if there were no bonds of caring and affection between
us when they were alive.
The above are examples of individualized duties, by which I mean duties to remember
individuals under descriptions that assign them particular identities. These duties encompass
remembering individuals both living and dead, though they are fulfilled with particular solem-
nity for those who are no longer living. Some victims and rescuers, for example, may still be
alive and, other things being equal, there is a duty to remember them just as there is a duty to
remember those who did not survive. A duty to remember is also thought to encompass espe-
cially important people, such as friends, who are absent but not dead, where the absence is not
due to wrongdoing—this is just what a friendship demands. In contrast to duties to remember
that are individualized in this way, there are duties to remember that include within their
ambit individuals whose specific identities are unknown or who are not given specific identi-
ties. These individuals may just form an aggregate or they may jointly conceive of themselves
as unified in some way. For example, there are said to be duties to remember entire communi-
ties, such as the indigenous Mayan communities of Guatemala destroyed by government-led
genocide in the 1980s. Finally, different from all of the above uses, a duty to remember may
refer primarily to events rather than to people. That is, what is salient in the duty may be that
so-and-so occurred rather than that it happened to so-and-so, despite the fact that what made
those events significant is their human impact. For example, there may be a duty to remember
the debacle of Vietnam or the bombing of Pearl Harbor.
(iii) Who has the duty?

The answers to the former question have a bearing on how we answer this one. If there is a duty
to remember a deceased or absent friend qua friend, then it would only be binding on the person
who had this particular relationship with him. The nature of the duty is such that the identity of
the individuals who owe this duty is completely or perfectly specified. A complete stranger might
have a duty to remember him, but it would be under a different description than that he was or
is my friend. Similarly for other individualized duties. There are complications here, however.
For not all individualized duties are duties that obligate persons individually. A community
(religious, ethnic, national, or other) to which an individual belonged might have a collective
duty to remember him specifically, by name. No one member of the community alone has this
duty, nor do the members have this duty severally, but as an entire community. And then too
there may be non-individualized duties to memorialize the dead—in the sense that they are
not remembered as persons with specific identities—that are incumbent on groups of people
or humanity at large, such as the victims of genocide or gross human rights violations.1 These
may be in addition to duties to remember victims of atrocities that are thought to be especially
imperative for the perpetrators because they are responsible for committing them.
(iv) To whom is the duty owed?

Here the distinction between remembering the dead and remembering the living takes on par-
ticular significance, for while duties to remember the living seem unproblematic, this is not so
for duties to remember the dead. Suppose I say that I have to remember someone who has died
because I promised him that I would do so. Or I promised him while he was alive that I would
do something after his death that required me to remember him and my promise to him. To
whom is this duty owed, once he dies? Does the duty become otiose at his death? There is a vast
philosophical literature on the problem of posthumous duties, and I will not attempt to review
it here (see, e.g., Partridge 1981; Ridge 2003; Meyer 2004; Winter 2010). One approach is to
353
Jeffrey Blustein
distinguish between duties to and duties in regard to: there are no duties to the dead as such but
there are duties toward or concerning them, which are owed to the ante-mortem person and
can remain in force after the person dies. Another explanation is that they are owed to future
generations who are liable to be the victims of renewed abuses if the dead are forgotten.
Further issues are raised by the distinction between individualized and non-individualized
duties. I said earlier that among the non-individualized duties to remember there are thought to
be duties to remember groups of people. Nothing follows from this about to whom the duties
are owed, but it may be that these duties are also understood to be owed to the group and not
just to individuals considered severally. The question this raises is how a duty can be owed to
a group as such, and to answer this we first have to define what we mean by “the group” and
clarify its relationship to the individuals who make it up. If those to whom the duty is owed
are alleged to be the victims of human rights violations, there is the additional problem that the
number of victims vastly exceeds the capacity of anybody to remember them all, except under a
very indefinite description. However, since this opening section only aims to provide an analysis
of the concept of a duty to remember, I simply flag these problems here.
Answering the above four questions will help clarify the conceptual contours of a duty to
remember: what it means to have duty in this context; what the content of the duty is; who
possesses the duty, and to whom the duty is owed. I have not taken a position on whether there
actually are any of the duties to remember that I have used as examples or even whether there is
a duty to remember at all. I have simply presented these as illustrations of when intuitively it
seems appropriate to talk about a duty to remember, and of the range of contexts in which such
talk seems germane, so they are to be taken as provisional, pending a theory (or theories) that
makes sense of and justifies our intuitions about the duties that individuals and groups have.
They are nevertheless a necessary preliminary. For it is not advisable to approach the task of
moral justification without first having a more or less clear understanding of what is being evalu-
ated, and this is especially important when, as in the case of a duty to remember, there is such a
multiplicity of possible meanings and uses to consider.
2. Philosophical antecedents
As I showed in the previous section, a conception of a duty to remember shadowed by the
Holocaust provides only one set of answers to the questions about the duty that I asked there,
and I suggested that we think of the duty in the more capacious sense suggested by the vari-
ous possible answers to these questions. To do this, I believe it is fruitful to begin by asking
whether philosophers of classical antiquity had a concept of a duty to remember and, if so, how
they understood, deployed, and justified it. My aim is not merely to ask whether they had our
modern understanding of a duty to remember, for I don’t want to privilege this as the point of
reference. Rather, I want to let them speak for themselves from within their own moral frame-
works. We can expect the investigation to uncover similarities and dissimilarities between how
they conceived of a duty to remember and how we, influenced by modern moral thinking, do.
All bona fide cases must have this in common, however: remembrance must be morally impera-
tive and omitting it morally criticizable in a particularly strong sense.
What follows are some very brief and selective sketches that are intended to be suggestive
only and to invite further inquiry into the wide range of philosophical expressions of a duty
to remember. Due to limited space, I will confine myself to certain philosophers from classical
antiquity who provide us with materials for thinking about the duty in ways that contrast with
how the duty is predominantly conceptualized and defended today.
354
A duty to remember
Alasdair MacIntyre makes the familiar point that “the concepts of duty and responsibil-
ity in the modern sense appear only in germ or marginally [in Greek philosophical ethics],
those of goodness, virtue, and prudence are central” (MacIntyre 2011: 84).2 Kant’s account
of the imperatives of practical reason is usually thought to typify the “modern sense” of duty.
In Kant’s moral universe, these imperatives are divided into two kinds: those that command
unconditionally simply because we possess rational wills (moral) and those that tell us how we
must act on the condition that we want or desire some end (non-moral). The Greeks, how-
ever, recognized a kind of “must” that “represents neither of Kant’s imperatives” (Williams
1993: 76). It is not that they failed to appreciate the distinction between duty and mere inter-
est. Quite the contrary, they were certainly capable of distinguishing between actions that
are good to do and those that are both good to do and of fundamental human importance, so
imperative or obligatory. It is rather that duty as they understood it does not exactly fit under
the Kantian (i.e., “modern”) rubrics.
Aristotle
Aristotle does not say in so many words that there is a duty to remember, but by drawing
together various remarks in some of his writings a case can be built that he would endorse
it in some circumstances. The first reference is to the discussion in the Rhetoric of the ideal
rhetorician, the phronimos, who plays the role of teacher and persuader in shaping the Athenian
public realm. This ideal rhetorician celebrates public virtue by presenting stories of individu-
als and deeds that serve as models for emulation, and he seeks to draw citizens who are never
free from the passions of the moment or from self-interest together in the collective pursuit of
shared goals. Though his responsibilities in the public realm go beyond simple commemoration,
memorials are nevertheless integral to the promotion of civic virtue and social unity. The duty
to remember, thus conceived, binds both the ideal rhetorician and the citizenry; it is a duty to
remember the illustrious heroes of the past and their glorious deeds; and it is a duty because
remembering them is an essential part of the means of promoting public morality and the for-
mation of a communal ethos. Moreover, since for Aristotle a good or happy life for individuals
can only be realized within the context of a political community of this sort, we can establish
by an indirect route a connection between the duty to remember thus construed and the good
life. Jointly participating in memorializing and celebrating the community’s past is necessary to
create the conditions that are constitutive of and instrumental to individual happiness.
Other passages can be gathered to suggest that a duty to remember has a more private func-
tion within the ethical life of the family. In book I of the Nicomachean Ethics, Aristotle claims it
would be “absurd” (NE, book I, chapter 10) to believe that the dead are not affected at all by
the changing fortunes of living descendants. The happy person, in Aristotle’s view:
is one who succeeds in fulfilling the aims that he sets himself, and if these include, e.g.,
the welfare of his children, then his success, and hence his happiness, depends on what
happens after his death, when his children succeed or fail.
(Irwin 1999: 188)
If this is so, then a child who, for example, acts dishonorably would bring discredit to his father
and would retroactively frustrate the aims that his father had for him. If the child remembers the
investment his father made in his (the child’s) future welfare and understands the effect his actions
would have on his father’s happiness, then this may be a disincentive to acting dishonorably.
355
Jeffrey Blustein
But why is it imperative for the child to remember his father and the hopes his father had
for him? The reason is suggested by Aristotle’s belief that a child is forever in his father’s debt
because of the benefits he has received from him (NE, book VIII, chapter 14). The child goes
some way toward fulfilling this debt by showing his father continual honor and deference, and
this does not end with the father’s death. On the contrary, the child must monitor his actions
even after his father dies to make certain that they do not dishonor him and thus adversely
affect his happiness. The basis of a duty to remember can be found here. The child has a duty to
remember his father and his inferior status vis-à-vis him because the father’s happiness, which is
to some extent in the child’s hands, can be significantly altered by the child’s actions, and filial
duty demands that the child attend to it.
Greek and Roman Epicureans and Stoics

Epicureans placed a very high value on friendship because of its special contribution to the
goal of human happiness, which for them consisted in ataraxia, or tranquility of mind. A
network of friends, in their view, protects each from physical danger and provides help to
each in times of need, thus enabling each to lead a life free of mental anxiety. But Epicureans
thought that friendship has other kinds of benefits as well for the individual. As Tim O’Keefe
notes, “it reinforces proper philosophy, provides models of conduct, and helps prevent one
from developing vain desires” (2001: 278).
Arguably, friendship could continue to provide these benefits even after the friend has died.
There seems no reason why, if the companionship of one’s friends provides a model of conduct
for the living and keeps the living from developing vain desires, memories of them after they die
couldn’t do something similar, even if perhaps somewhat less vividly or forcefully. Furthermore,
if receiving the intellectual and moral benefits of friendship is part of the reason why friend-
ship is valued so highly in virtue of its relation to happiness, then we have here the makings of
another argument for a duty to remember: in this case, it is a duty essentially owed to oneself
to remember one’s friends, or more precisely, the good council and virtuous conduct of one’s
friends. The argument is a plausible extension of their views, I believe, even if these connections
are not actually drawn in the Epicurean sources we have.
We might be on somewhat firmer ground attributing a duty to remember in the core sense
I have identified if we turn to the Roman Stoic Seneca, since memory, and more specifically,
memory of one’s deceased friends and loved ones is a recurring theme in his writings. There is a
problem here, however, for it seems that there can be little room in Stoic ethical thought for a
duty to remember deceased friends and loved ones. Remembering them after they die is a sign
that one is excessively attached to them, the argument goes, and since excessive attachments are
to be spurned, remembering friends and loved ones cannot be either good or morally imperative.
But this argument is mistaken. To understand why, we have to bear in mind the distinction that
Seneca, like other Stoics, draws between grief (or mourning) and remembering. To be sure, a
true Stoic will not allow his attachments to friends or loved ones to become so extreme that he is
consumed by misery and grief when they die. Although it is natural even for him to grieve over
their loss for a time, he will moderate his grief and his grief will be short-lived. But his memory of
friends and loved ones, if they can be truly counted as such, will not last only as long as his grief. His
grief passes, but the memory survives.3 Thus, the Stoic criticism of pathological attachments entails
that one should control one’s memories of the dear departed to the extent that these memories are
accompanied by grief. But it does not entail that one should strive to forget. We can therefore say
at least this much: it is consistent with Stoic ethical thought to maintain that one ought to remember
one’s friends and loved ones after they die. But what is the argument that it is imperative to do so?
356
A duty to remember
Part of the answer is that remembering our friends and loved ones after they die is an attrib-
ute that sets us apart from lesser animals (Seneca Letter 99, para. 23). Remembering “one’s
own” is what the wise person does because it is a condition of living a truly human life, a life
that befits human beings rather than animals. And of course for the Stoic there can be no ques-
tion whether this is the life that one ought to live.
Another part of the answer, perhaps more speculative, concerns the role of memory and,
in particular, memories of friends and loved ones, in enabling a person to live the life of a wise
man. Remembering past pleasures is a resource that protects us from the anxiety caused by the
precariousness of possessions and that brings tranquility to the soul, for memory endures as a
refuge against the loss and the fear of loss of present possessions.4 As Seneca says, “We can be
robbed of ‘having,’ but never of ‘having had’” (Letter 98, para. 5, 11).5 The turn to memory
facilitates the detachment from external contingency that Stoics believe is necessary for happi-
ness and a life of moral virtue. As for how memories of friends and loved ones figure in this
account, the following is a plausible conjecture: the memories of friends and loved ones, of
“one’s own,” both living and dead, are an indispensable source of enduring pleasure. They are a
treasure that those who aspire to wisdom should mine. The wise person will use these memories
as part of a process of psychological discipline, to console himself for his losses and to achieve
the tranquility of mind needed for a life of virtue.6 Memory in general, and memory of friends
and loved ones in particular, serve as a powerful antidote to the fears and anxieties that beset us
in our everyday lives and that one ought to seek to transcend, and passages in Seneca’s writings
suggest that he would endorse this view.
Admittedly, the notion of a duty to remember does not occupy as prominent a place in the
ethical thinking of the ancients as it does in philosophical literature marked by the Holocaust.
But what this brief foray into history of ideas suggests is that it was not foreign to their ethical
thinking to hold that people have obligations with respect to memory and that memory can
be controlled and directed to serve moral ends. The pertinent writings I have cited develop
particular culturally inflected iterations of the duty, different from contemporary iterations in
part because of how they conceive of duty as such, whatever object it happens to be related to.
However, despite theoretical differences dividing contemporaries from ancients, the domains or
spheres of life where I have found evidence of a duty to remember in ancient writings are also
domains where we locate it and where we think remembrance has considerable moral signifi-
cance, namely, the domains of politics, family life, and friendship. This shows that the sorts of
duties the ancients hinted at or gestured towards are not part of a world-view or moral outlook
totally alien to our own.
There are differences too, of course, between ancient ethical thought and the ethical thought
that has dominated contemporary theorizing, even if these differences are occasionally exagger-
ated, and it is important to point these out to get a fuller understanding of the moral bases of
a duty to remember in its various instantiations. The dominant consequentialist and deonto-
logical theories of morality (more below) tend to focus on actions and their consequences, and
the rules or principles to which they conform, in their moral assessments. In contrast, ancient
ethical thought focuses on agents in their assessments, on their states of mind, traits of character,
and virtues, where the conception of a virtuous agent is irreducible to an account of the actions
she typically performs or the principles to which she adheres.7 If my interpretations of passages
from the ancient writers are plausible, they show that they do set forth moral imperatives to
remember and further, that they justify them by drawing close connections between them and
the human flourishing of persons who remember. This agent-centered orientation is generally
missing from contemporary treatments of the duty to remember from a consequentialist or
deontological standpoint.
357
Jeffrey Blustein
The agent-centered perspective is taken up by Nietzsche in his famous discussion of the

values of remembering and forgetting, and in this respect he is a descendant of the ancients.8
Whether, when and what to remember, and to forget, are questions that for Nietzsche can
only be answered by reflecting on what it takes to live a good life, one that is neither enslaved
to the past nor deluded by fantasies of radical self-invention. Remembering and forgetting
are in dynamic tension with one another, and a good and healthy life achieves the right sort
of balance between the two. This is possible because, like Seneca, Nietzsche is a voluntarist
about memory, meaning that he believes that memories, their uses and their effects, can be
governed by the will, and he is a voluntarist about forgetting as well.9 Since remembering
and forgetting are subject to the will, each can be shaped by and adjusted to the other. This
emphasis on leading a good life in turn has implications for the content of duties to remem-
ber. Generally speaking, individuals and peoples must remember the past because this serves
vital human interests, and the duties specify how to do this in the right way, worked out
case-by-case. Throughout, we have to attend critically both to how to remember and what
to forget, for otherwise we cannot develop new habits of living free of the encumbrances of
the past. That is, we must remember it in ways that fit our memories to be part of an admi-
rable or healthy life, for individuals individually as well as collectively. As with the ancients,
duties to remember for Nietzsche are conditioned by their contributions to a good life for
those who remember.
3. Modern ethical theories and a duty to remember

Modern ethical theories are commonly referred to as either consequentialism or deontology,
although recently there has been growing interest in what is called “virtue ethics.” Contemporary
virtue ethicists tend to see themselves as carrying on a tradition of moral philosophy that begins
with Aristotle, and reflection on the possibility of a virtue ethics foundation for the duty to
remember harks back to my previous discussion of this duty among the ancients. In what fol-
lows, however, I will confine myself to the two dominant theories.10 I start by discussing one
familiar version of consequentialism which I will use to identify the conditions under which an
agent can be said to have a duty to remember. I then turn to deontology.
Consequentialism, as a general type of normative ethical theory, holds that the rightness
or wrongness of a course of conduct is solely a function of the goodness or badness of the
consequences of our actions or the rules to which they conform. There are different versions
of consequentialism, so I will confine my attention to what William Shaw calls “standard con-
sequentialism,” according to which “the morally right action for an agent to perform is the
action, of those actions that the agent could perform at the time, that has the best consequences
or results in the most good” (Shaw 2006: 5). Morally right actions are also morally required.
Applying standard consequentialism to the case at hand, there is a duty for x to remember y
just in case x’s remembering y is the action that, among those actions available to x at the time,
can reasonably be anticipated to bring about the best consequences. (Versions of consequential-
ism that do not hold it obligatory to bring about the best consequences might yield a permission
to remember but not a duty.) This purely formal statement of necessary and sufficient conditions
for the duty is given content by answering these questions: what is the good that x remembering
y can be expected to produce? Are there other actions open to x at the time (including neglect-
ing to remember y or taking steps to forget) that can be expected to bring about equally good
or better consequences? Let’s consider a few examples in light of these questions.
For ethicists who are primarily concerned about the good of the agent, empirical psychol-
ogy offers some support for a duty to remember. The duty in question involves memories
358
A duty to remember
of lost loved ones, as in the discussion of Seneca, but it presents an interesting contrast with
the justification I attributed to him. The “good” in terms of which the duty is justified is not
detachment from external contingency (Seneca) but rather the recovery of effective agency,
which involves the renewal of hope and energy for engagement with and immersion in life.
This can only happen if the loss no longer commands as central a place in the survivor’s life as it
once did, and this in turn is only possible if the survivor is able to mourn the loss and place it
in the past: not by forgetting it, for this cannot lead to healing, but by coming to remember
it in a different way.
Arguably there can also be a standard consequentialist duty to remember lost loved ones.
Consequentialists require individuals to be effective maximizers of the good, and only those
who are able to exercise effective moral agency can function this way. If one has suffered a seri-
ous loss and is unable to recover from it, one’s effectiveness as an agent of good, and one’s ability
to fulfill one’s moral responsibilities to others, are undermined. But by mourning the lost loved
one, one’s energies are freed up to promote the good of others as well as oneself. Remembering
the dear departed is imperative for this reason.
I said in Section 1 of this chapter that our intuitive moral judgments about when there is a
duty to remember are quite varied. Among these is that there are duties to remember not only
specific absent or dead persons, but also persons who satisfy a certain generic description, such as
the victims of atrocities like crimes against humanity or genocide, and whose specific identities
are in many cases unknown. There are also duties to remember that give past events, rather than
the individuals who benefited or suffered from them, salience. Consequentialist justifications of
duties of these sorts are also available.
Remembering past atrocities and their victims is widely regarded as a duty for more than just
the perpetrators, and a frequently offered consequentialist justification of this duty is that remem-
bering them is necessary to prevent recurrence. Those who offer this argument do not claim
that remembering the crimes and abuses of the past, and teaching the next generation about
them, are sufficient for their future deterrence. At the same time, neither do they claim that it
is only a way of preventing their repetition, one option among others. It is rather indispensable
for the realization of this goal. This consequentialist justification of a “duty to remember” con-
tinues to have many supporters, even though memory’s significance is diminished by mankind’s
repeated failure to learn from the past.
Still, remembering past atrocities could have a deterrent effect by helping to promote a social
environment in which human rights are widely respected. If remembering past atrocities is
essential to achieving this goal, as it seems to be, there would be grounds for a duty to remember.
Of course, drawing attention to human rights violations, while extremely important, is not
enough to create a climate of respect for them, even if the violations are widely publicized.
Among other things, the wrongs have to be contextualized—their political, social, and institu-
tional causes explained—so that constructive action to prevent recurrence can be undertaken
and respect can be concretely shown. But there are compelling reasons to believe that reveal-
ing, disseminating, and advocating for the truth about events of the past is necessary to prevent
future abuses, and this cannot happen if memory of those events is stifled or repressed.
Relatedly, a duty to remember can be defended for its benefits to the society or commu-
nity that has suffered the traumatic effects of atrocity. Here, therapeutic notions of closure and
healing—not or not only of individuals, but of entire societies and communities—are usually
invoked, and once again remembering the wrongs of the past is said to be essential to the
achievement of these ends. Moreover, remembering past atrocities and those who commit-
ted them is necessary if perpetrators are to be held accountable for their crimes, justice is to be
served and seen to be served, and reparations are to be paid to survivors.
359
Jeffrey Blustein
Three comments about these consequentialist arguments for duties to remember further
clarify the approach. First, as with many consequentialist arguments of this type, there is con-
siderable difficulty both in identifying the best course of action and in arriving at judgments
about what the consequences of one’s actions are likely to be. Second, these duties are not just
binding individually, but collectively as well. The duties are collective because to have the sort
of deterrent and therapeutic effects I have described, the memories of past wrongdoing and its
victims must be shared across entire societies and political communities comprised of perpetra-
tors as well as victims. Third, remembering abuses of the past is held to be a duty because, by
comparison with other options the agent could perform at the time, the most good is likely
to be produced by remembering, but the good that is maximally produced by memorializing
human rights violations and their victims can be characterized in different ways. One suggestion
is that the good in question is that of respect for rights, or for human rights specifically. Thus, it
would be the case that acts of remembering past wrongs are morally obligatory if, of the actions
the agent could perform at the time, only they (of course in combination with other acts) can
be expected to result in the most good, that is to say, the most respect for certain moral rights,
or the least bad, i.e., the fewest violations of them.
I turn next to deontology. The features of a deontological ethic are commonly brought
out by contrasting it with consequentialism. For a deontological ethic, the morality of actions
is assessed by different criteria than whether they produce the best states of affairs. This is not
to say that deontologists hold consequences to be irrelevant—a common misconception about
deontology—but only that they fundamentally evaluate the correctness of actions without refer-
ence to the goodness of the states of affairs they bring about, or can be expected to bring about.
For many deontologists, the rightness of actions is explained by whether they conform to a
moral rule or principle: the rule or principle is simply to be obeyed by the moral agent. Other
deontologists regard the property of rightness as inherently particularistic, so not dependent on
a rule or principle. However, a unifying element of most contemporary deontological theories,
and a sense in which they are all Kantian, is a commitment to respect for persons. Our ethical
response to persons is held to be qualitatively different than our ethical response to things and
animals, and consequentialists are criticized for not recognizing in persons a moral status that sets
them apart from these other valuable objects.
How can a duty to remember be defended from a deontological standpoint? We can illus-
trate this if we conceive of remembrance in some circumstances as a requirement of justice. A
deontological justification of the duty would not be that remembering wrongdoing prevents its
future recurrence; or promotes respect for a culture of respect for human rights; or removes bar-
riers to the reconciliation of perpetrators and victims. These would be important but ancillary
considerations. Instead, a deontological argument would be that fulfilling the duty to remem-
ber is justified on the grounds that it discharges a duty of justice, or more specifically a duty of
corrective justice, which is owed to the victims and survivors and, further, that this is a cat-
egorical requirement that cannot simply be overridden by a calculation of adverse consequences.
From this deontological standpoint, remembering would be a duty because it conforms to a
moral principle, such as “Do not allow injustice to go unacknowledged or un-redressed,” or
“Reparations are owed to the victims of injustice,” or because doing justice in this way is an
intrinsically right-making feature of the action. Respect for rights, on this view, is not a good to
be maximized but a duty to be done.
Consequentialist arguments for a moral duty to remember like those discussed above are
vulnerable to defeat based on reasonable projections of consequences. Thus, if it were the case
that a reasonable and conscientious person would judge it unlikely for remembering to have
the aforementioned results, or if a reasonable and conscientious person would judge other
360
A duty to remember
actions available to him at the time to be more likely to achieve them, the argument for the
duty would fail in those circumstances. Deontological arguments about a duty to remember,
on the other hand, are not defeated in the same way, or for the same reasons. Adverse expectable
consequences of engaging in acts of remembering past wrongdoing, if they pass a certain thresh-
old of gravity, might dissuade even a deontologist from them. But the demands of corrective
justice—reparative, compensatory, and punitive—would still have purchase.
Principles of justice are not the only candidate deontological principles that can play a role
here, although they are particularly relevant in political contexts where societies are transition-
ing from a past of violence and oppression to a new political order founded on peace and respect
for human rights. Examples of other deontological principles that could serve as the basis of
a duty to remember include these: “Honor is owed those who have made great sacrifices in
service to others”; “Participants in associative relations of love and friendship owe each other
loyalty,” and “Gratitude is owed those who have made us who we are.” Of course these and
other principles, plausible as they may be, cannot stand alone as the moral foundations of a duty
to remember, for they are in need of justification themselves. A deontological ethic owes us an
account of why any of these principles give an agent a moral reason to obey them independently
of the consequences of the obedience.
Deontology, I have argued, is one source of an ethics of remembrance. However, this may be
questioned, since at least on some conceptions of deontology, they appear to represent opposing
philosophical perspectives.11 To explain this, consider differences in how they conceive of the
self. This is a large and complicated issue, so I will simplify by presenting the contrast at issue
in deliberately stark terms: the conception of the self and its interactions with others underlying
certain deontological outlooks is not the conception of the self and its interactions with others
that best accords with how memories are constructed and sustained. On the one hand, there
are deontologies that are individualistic because of a core principle of respect for autonomy,
where autonomy is understood to be a property of the self that is modeled on the autonomy
of independent sovereign nations, a property that is causally affected by but not constituted by
personal and social relations in which they stand. This is the conception of the self that lies at the
core of much liberal social and political philosophy. On the other hand, there is a conception
of the self that emphasizes its relational character, relational in the sense that it is understood to
be historically and socially embedded and whose most important features and sense of identity
are social, interpersonal constructions.12 This is a conception of the self that lies at the core of
feminist and care critiques of traditional, that is to say, individualistic, conceptions of autonomy.
These conceptions of the self are at philosophical odds with each other, and reflection on how
memory operates in the lives of individuals provides evidence for the former, not the latter.
Memories, as they are experienced by us in our remembering lives, are constituted by inter-
actions between us and others in our social environment. We rely on various sorts of input from
other persons to help form our memories; to sustain them and strengthen our confidence in their
veracity; to correct them when our “memories” do not jibe with those of others; and to hold
our memories in safekeeping so that they do not wane as our unassisted capacity to retain them
weakens. It is still, according to this account, the individual who remembers. But the activity
of remembering is understood as socially extended, distributed between the individual and her
co-rememberers, which include friends, family members, colleagues, and fellow community
members.13 The remembering self, or the self insofar as it remembers, is in this way a relational
self, the sort of self that is at odds with metaphysical assumptions about the essential separateness
of persons that underlie some deontologies. The solution, I believe, is not to abandon respect
for autonomy as a core moral principle, but to modify their conception of autonomy so that it
better accords with the relational nature of memory.
361
Jeffrey Blustein
Acknowledgment
I want to thank David Konstan and Nick Pappas for helpful comments on the philosophical
antecedents of the notion of a duty to remember.
Notes
1 As Margalit claims (2002: 82–3).

2 According to Williams (1993: 41), “Duty in some abstract modern sense is largely unknown to the
Greeks.”
3 See also Seneca, Consolation to Marcia.
4 Silvia Montiglio (2008: 168–80) argues this point.
5 See also Letter 99 (5) where Seneca maintains that the most secure possessions are those retained in
memory and that the wise person will seek comfort in them.
6 David Konstan (2015) makes a related point in his discussion of grief and friendship in the writings of
the Roman Stoic Cicero.
7 I follow Richard Parry’s characterization in “Ancient ethics,” in the online Stanford Encyclopedia of
Philosophy.
8 There is a longer treatment of Nietzsche’s views on remembering and forgetting in Blustein (2008).
9 On the voluntarism of Seneca, see Montiglio (2008: 169).
10 Additional sources of an ethics of remembrance are discussed in Blustein (2015).
11 I don’t mean to suggest that only deontological ethics raises this problem.
12 For an exhaustive discussion of different accounts of so-called relational autonomy, see Mackenzie and
Stoljar (2000: 3–31).
13 I adopt language used by proponents of the extended mind thesis but want to avoid the controversies
it has generated. There is a useful discussion in Wilson and Lenart (2015); also Chapters 19 (“Extended
Memory”) and 20 (“Collective Memory”) in this volume.
Related topics
• Extended memory
• Collective memory
References
Aristotle (1991) On Rhetoric, G. Kennedy (trans.) (New York: Oxford University Press).
Aristotle (1999) Nicomachean Ethics, 2nd edn., T. Irwin (trans.) (Indianapolis, IN: Hackett Publishing).
Blustein, J. (2008) The Moral Demands of Memory (New York: Cambridge University Press).
Blustein, J. (2015) “How the Past Matters: On the Foundations of an Ethics of Remembrance,” in
K. Neumann and J. Thompson (eds.), Historical Justice and Memory (Madison, WI: University of
Wisconsin Press), pp. 74–92.
Irwin, T. (1999) Notes to Nicomachean Ethics (Indianapolis, IN: Hackett Publishing).
Konstan, D. (2015) “Cicero on Grief and Friendship,” in A. Tutter and L. Wurmser (eds.), Grief and Its
Transcendence: Memory, Identity, Creativity (New York: Routledge).
MacIntyre, A. (2011) A Short History of Ethics, 2nd edn. (Notre Dame, IN: University of Notre Dame
Press).
Mackenzie, C. and Stoljar, N. (2000) Relational Autonomy (Oxford: Oxford University Press).
Margalit, A. (2002) The Ethics of Memory (Cambridge, MA: Harvard University Press).
Meyer, L. (2004) “Surviving Duties and Symbolic Compensation,” in L. Meyer (ed.), Justice in Time:
Responding to Historical Injustice (Baden-Baden: Nomos Verlagsgesellschaft), pp. 173–83.
Montiglio, S. (2008) “Meminisse Iuvabit: Seneca on Controlling Memory,” Rheinisches Museum fur Philologie
Neue Folge, 151 Bd., H. 2: 168–80.
O’Keefe, T. (2001) “Is Epicurean Friendship Altruistic?” Apeiron: A Journal for Ancient Philosophy and
Science, 34(4): 269–305.
362
A duty to remember
Parry, R. (2014) “Ancient Ethical Theory,” Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/

entries/ethics-ancient/.
Partridge, E. (1981) “Posthumous Interests and Posthumous Respect,” Ethics, 91(2): 243–64.
Ridge, M. (2003) “Giving the Dead Their Due,” Ethics, 114(1): 38–59.
Seneca (2008) “Consolation to Marcia,” in J. Davie (trans.), Seneca: Dialogues and Essays (Oxford: Oxford
University Press).
Seneca (2015) Letters on Ethics: Letters 9, 98, 99. M. Graver and A.A. Long (trans.) (Chicago, IL: Chicago
University Press).
Shaw, W. (2006) “The Consequentialist Perspective,” in J. Dreier (ed.), Contemporary Debates in Moral
Theory (Malden, MA: Blackwell Publishing), pp. 5–20.
Williams, B. (1993) Shame and Necessity (Berkeley, CA: University of California Press).
Wilson, R.A. and B.A. Lenart (2015) “Extended Mind and Identity,” in J. Clausen and N. Levy (eds.),
Handbook of Neuroethics (Dordrecht: Springer), pp. 423–39.
Winter, S. (2010) “Against Posthumous Rights,” Journal of Applied Philosophy, 27(2): 186–99.
363
28
AN OBLIGATION
TO FORGET
David Matheson
In this chapter, I argue that as an activity forgetting is susceptible to moral obligation. After
responding to a worry about whether the moral obligation to perform this activity ever realisti-
cally obtains, I present a tentative account of an important species of the obligation, namely, the
moral obligation to forget sensitive information about other agents. I then consider the bearing
of these reflections on two broad criticisms of a recently affirmed “right to be forgotten” online.
1. Morally obligated epistemic activities

The processes of learning, remembering, failing to learn, and forgetting are epistemic in the
sense that they characteristically result in epistemic states. Learning characteristically results in
the positive epistemic state of new knowledge of something, remembering in that of continued
knowledge of something.1 Failing to learn characteristically results in the negative epistemic state
of continued ignorance of something. And forgetting characteristically results in new ignorance
of something.
Very often these processes simply befall agents. But sometimes the processes are activities
that agents perform for the sake of realizing the relevant epistemic states in themselves—what I
will call “epistemic activities.” Thus, my learning something need involve no activity at all; but
where my learning is an activity I perform (e.g., searching carefully, consulting an expert) for the
sake of having new knowledge of the thing in question, my learning is an epistemic activity. I
can remember something without performing an activity; but if my remembering is an activity
I perform (e.g., reviewing notes, rehearsing a chain of thought) for the sake of having contin-
ued knowledge of that thing, my remembering is an epistemic activity. Similarly, although my
failing to learn or my forgetting something may amount to a mere epistemic process (i.e., an
epistemic process that simply befalls me), if my failing to learn or my forgetting is an activity I
perform (e.g., avoiding relevant evidence, repeatedly refusing to think about a matter) in order
to have continued or new ignorance of that thing, my failing to learn or my forgetting is an
epistemic activity.
Unlike mere epistemic processes and the epistemic states in which they characteristically
result, epistemic activities seem clearly susceptible to moral obligation. To illustrate, consider
the following situations:
364
An obligation to forget
Diagnosis
Your young child has recently been diagnosed with a medical condition. You know
little about the condition beyond that it is serious and requires careful, long-term
attention from parents and healthcare professionals alike. You say as much to the
attending physician, and she provides you with some excellent resources for learning
more about the condition, particularly about the best parental practices for dealing
with it. She encourages you to learn the information and promises to do what she can
to help. You correctly assume that with her help you can successfully learn the infor-
mation, and that doing so will not carry significant costs in terms of your ability to care
for yourself or others in morally appropriate ways.
Progress
Having learned the relevant information in Diagnosis, it becomes apparent that it
will take some effort to remember it all in the coming months and years. You know,
however, that so remembering it will be very important to the continued progress
of your child. You correctly assume that, despite the effort it will involve, you can
successfully remember the information without unduly diminishing your ability to
care for yourself or others in morally appropriate ways.
Hypothesis
You are a behavioral psychologist who would love to know whether a certain hypo-
thesis about human action under duress is true. Your interest in knowing this is
fundamentally epistemic: it’s not that you want to know it so that you can help cor-
rect a grave social injustice. But you’re aware that, as things stand, the only ways of
successfully learning whether the hypothesis is true would involve the very unethi-
cal treatment of experimental subjects. Because you would so dearly love to know
whether the hypothesis is true, not giving in to the temptation to learn the truth of
the matter will require some effort on your part; but of course the effort could hardly
be said significantly to diminish your ability to care for yourself or others in morally
appropriate ways.
Plausibly, you have some moral obligation to perform an epistemic activity in each of these
situations. In Diagnosis you have a moral obligation (derived from a more general moral
obligation to look after your dependents) to perform the epistemic activity of learning the
information about your child’s condition. In Progress you have a moral obligation to per-
form the epistemic activity of remembering this information. And in Hypothesis you have a
moral obligation to perform the epistemic activity of failing to learn (i.e., avoiding learning)
whether the hypothesis is true.
But what about the epistemic activity of forgetting? Is it also susceptible to moral obligation?
Consider now a further situation:
Invasion
You recently learned some sensitive information about a colleague (i.e., informa-
tion about her that she strongly prefers to keep to herself). She obviously has a
moral right to privacy with respect to sensitive information about herself of this sort
365
David Matheson
(i.e., a moral right that others not perform the epistemic activity of learning such
information about her), and your conscience is getting the better of you because
you learned the information by violating that right and shamefully invading her
privacy. You know that with appropriate effort you can over time successfully for-
get the information you learned. You also know that the effort would not unduly
diminish your ability to care for yourself or others in morally appropriate ways.
It certainly seems that you have some moral obligation to perform an epistemic activity in
Invasion, and forgetting the information you shamefully learned is the obvious candidate.
Construed as an obligation to perform an epistemic activity, then, it seems on the face of
things that we can be morally obligated to forget, just as we can be morally obligated to learn,
remember, and fail to learn.
2. Realism about the moral obligation to forget

One might worry, however, that this moral obligation to forget never realistically obtains—
not for agents like you and me, at any rate. And one might point to an important difference
between situations like Invasion, on the one hand, and situations like Diagnosis, Progress,
and Hypothesis, on the other, in support of this worry: situations like ‘Invasion’ presuppose
that we can successfully cause ourselves to cease to know information we presently know without
unduly diminishing our ability to care for ourselves or others in morally appropriate ways. But,
the worry goes, that presupposition is unrealistic: in the real world, we cannot successfully cause
ourselves to cease to know information we presently know without unduly diminishing our
ability to care for ourselves or others in morally appropriate ways. For in the real world, we can
successfully cause ourselves to cease to know information we presently know only by doing
such drastic things as inflicting serious brain injuries on ourselves or taking cognitively harmful
drugs; and, obviously, by doing such things we would be unduly diminishing our ability to care
for ourselves or others in morally appropriate ways.
By contrast, situations like Diagnosis, Progress, and Hypothesis presuppose only that we
can successfully cause ourselves to come to know information we don’t presently know, or that
we can successfully cause ourselves to continue to know information we presently know, or that
we can successfully cause ourselves to remain ignorant of information of which we’re presently
ignorant, without unduly diminishing our ability to care for ourselves or others in morally
appropriate ways. And none of these alternative presuppositions is unrealistic. Regularly in the
real world, after all, we successfully cause ourselves to come to know information we don’t
presently know without unduly diminishing our ability to care for ourselves or others in mor-
ally appropriate ways. Regularly in the real world, we successfully cause ourselves to continue
to know information we presently know without unduly diminishing our ability to care for
ourselves or others in morally appropriate ways. Regularly in the real world, we successfully
cause ourselves to remain ignorant of information of which we’re presently ignorant, without
unduly diminishing this ability.
Because situations like Invasion thus presuppose something unrealistic, the worry con-
tinues, they are unrealistic situations. So even if we are morally obligated to perform the
epistemic activity of forgetting in situations like Invasion, it doesn’t follow that there are any
realistic situations in which we are morally obligated to perform this activity. And hence it
is unclear whether the moral obligation to forget ever realistically obtains for agents like you
and me.
366
Surprising though it may be, there is good reason to reject the key thought behind this worry,
to wit, the thought that in the real world we can successfully cause ourselves to cease to know
information only by doing such drastic things as inflicting serious brain injuries on ourselves or
taking cognitively harmful drugs. As I have argued elsewhere (Matheson 2013), evidence from
the literature on self-deception (e.g., Bach 1981, 1997; Sahdra and Thagard 2003; von Hippel
and Trivers 2011) and directed forgetting (e.g., Barnier et al. 2007; Joslyn and Oakes 2005;
MacLeod 1998, 2012; Michaelian 2011) strongly supports the idea that there are in fact various,
nondrastic things we can do successfully to cause ourselves to cease to know information we
presently know—even where this information is emotionally significant to us. These typically
amount to retrieval-inhibition practices, whereby we prevent ourselves from regularly render-
ing occurrent the knowledge we presently have of the information in question. They amount to
nothing like self-infliction of serious brain injuries or taking cognitively harmful drugs.
3. The moral obligation to forget sensitive information

about other agents
Allowing, then, that the moral obligation to forget at least sometimes realistically obtains for
agents like you and me, under what conditions does it obtain? Answering this question would
amount to providing an account of the moral obligation to forget. In this section, I provide an
account of what I take to be one (if not the most) important species of this obligation, namely,
the one that plausibly obtains in situations like Invasion—the moral obligation to forget sen-
sitive information about other agents.2 The account I provide is a tentative one because the
conditions of which the account consists will undoubtedly require further precisification and
supplementation if it is to survive close critical scrutiny.
Alter Diagnosis, Progress, Hypothesis, or Invasion such that you cannot successfully per-
form the relevant epistemic activity without unduly diminishing your ability to care for yourself
or others in morally appropriate ways, and the plausibility of the claim that you are morally
obligated to perform the activity vanishes. This suggests one condition on the moral obligation
to forget sensitive information about other agents that is in keeping with the general deontic
principle that “ought implies can”: if an agent is morally obligated to perform the epistemic
activity of forgetting sensitive information about another agent, she can successfully perform
that activity without unduly diminishing her ability to care for herself or others in morally
appropriate ways.
I think a further, slightly less obvious condition on the moral obligation to forget sensitive
information about other agents can be gleaned from comparing situations like Invasion with
situations like the following:
Sharing
You recently learned some sensitive information about a colleague. Although she obvi-
ously has a moral right to privacy with respect to sensitive information about herself of
this sort, you learned the information because you listened intently as she purposefully
and freely shared it with you. You know that with appropriate effort you can over time
successfully forget the information. You also know that the effort would not unduly
Whereas it is plausible that in Invasion you have some moral obligation to perform the epis-
temic activity of forgetting the sensitive information about your colleague, it is not plausible
367
David Matheson
that in Sharing you have such an obligation. The best explanation of the difference seems to
be that in Sharing, but not in Invasion, your colleague has for you waived a moral right to
privacy that she has with respect to the sensitive information: by purposefully and freely sharing
the information with you (despite her strong preference to keep it to herself ), she has waived
that right. The comparison of situations like Sharing with situations like Invasion, accord-
ingly, provides reason to accept something like the following condition on an agent’s being
morally obligated to forget sensitive information about another agent: for the obligated agent,
the other agent must not have waived a moral right to privacy that she has with respect to that
information.
Consider now a further situation:
Stumbling upon
You recently learned some sensitive information about a colleague. Although she
obviously has a moral right to privacy with respect to sensitive information about
herself of this sort, you learned the information by accident: your colleague did not
purposefully and freely share the information with you, but neither did you invade
her privacy by violating her right to privacy in learning it; you just happened to stum-
ble upon the information. You know that with appropriate effort you can over time
successfully forget the information. You also know that the effort would not unduly
It is not plausible that you have a moral obligation to forget the sensitive information about
your colleague in Stumbling upon—no more plausible, at any rate, than that you have such an
obligation in Sharing. But why is this the case when in Invasion it is plausible that you have
the obligation? The last suggested condition yields no explanation, for in both Stumbling upon
and Invasion, your colleague has not for you waived a moral right to privacy that she has with
respect to the sensitive information.
To explain why it is plausible that you have the moral obligation in Invasion but not in
Stumbling upon, it seems that we must appeal to a point about the reason you know the sensi-
tive information about your colleague: in Invasion, you know the information because you
performed an epistemic activity, namely, the epistemic activity of learning the information. In
Stumbling upon, you do not know the information because you performed an epistemic activity;
you know the information rather as a result of learning it by a mere epistemic process—the mere
epistemic process of accidentally stumbling upon the information.
Reflection on situations like Stumbling upon thus seems to warrant the addition of yet one
more condition on an agent’s being morally obligated to forget sensitive information about
another agent: the obligated agent must know the sensitive information because she performed
the epistemic activity of learning it.
My tentative account of the moral obligation to forget sensitive information about other
agents takes the three conditions articulated above to be jointly sufficient for that obligation.
Thus, an agent S1 is morally obligated to perform the epistemic activity of forgetting sensitive
information i about another agent S2 just in case:
(1) S1 knows i because S1 performed the epistemic activity of learning i;

(2) S2 has not for S1 waived a moral right to privacy that S2 has with respect to i, and
(3) S1 can successfully perform the epistemic activity of forgetting i without unduly diminishing
her ability to care for herself or others in morally appropriate ways.
368
The situations to which I have appealed in support of this account have all involved only
individuals agents—persons like you and me. One might accordingly be tempted to think that
the account properly applies only to individual agents, and that it becomes implausible if the
“S1” and “S2” variables are allowed to range over corporate agents like Interpol or Google Inc.
But I see no good reason to restrict the scope of the account in this way. Indeed, I think there
is good reason not to do this. For not only does it seem that there are corporate agents (about
which there can be sensitive information), it seems that they can have moral obligations, know
things, be ignorant of things, perform such epistemic activities as learning and forgetting, and
have the ability to look after themselves and others in morally appropriate ways.3 Moreover, it
seems that corporate-agent analogues of the situations to which I have appealed in support of
my account are situations in which the plausibility of attributing the relevant moral obligation
to the relevant corporate agent matches the plausibility of attributing this obligation to you in
the situations to which I have appealed.
To illustrate this last point, consider one final situation:
Online invasion
You know that an online search engine company has learned some sensitive infor-
mation about you, because you discovered this information by using the very search
engine the company operates. You have a moral right to privacy with respect to
sensitive information about yourself of this sort, and you are incensed because you
also know that the only way the company could have learned the information is
by violating this right. With appropriate retrieval-inhibiting effort (e.g., actively
removing hyperlinks to the information from its search engine), the company can
over time successfully forget the information it illicitly learned. And, plainly, the
effort would not unduly diminish the company’s ability to care for itself or others in
morally appropriate ways.
Assuming (as I do) that it makes sense to talk of an online search engine company performing
the epistemic activity of forgetting sensitive information about you, it seems just as plausible
that the company has a moral obligation to perform this activity in Online invasion as it is that
you have an obligation to perform the epistemic activity of forgetting the sensitive information
about your colleague in Invasion.
4. On criticisms of a recently affirmed “right to be forgotten”

With reference to the European Union’s 1995 Data Protection Directive 95/46/EC (European
Commission [EC] 1995), the Court of Justice of the European Union has recently affirmed a
“right to be forgotten” online that imposes an obligation on agents—particularly corporate ones
like Google Inc.—to forget sensitive information about other agents (Court of Justice of the
European Union [CJEU] 2014: §§89–99 and Ruling 4; European Commission [EC] 2014). In
this closing section, I want briefly to consider how my reflections on the moral obligation to
forget may help us see the problematic nature of two broad criticisms of the Court’s affirmation.
The first broad criticism—call it “the nonsense criticism”—is that the Court’s affirmation
of the right to be forgotten is wrongheaded because the obligation the right imposes makes no
sense. Whereas obligations to do certain things make sense, obligations to be in certain mental
states (e.g., epistemic ones like knowledge or ignorance) make no sense, for mental states are by
their very nature free from obligations. But, the nonsense criticism goes, the obligation to forget
369
David Matheson
sensitive information about other agents imposed by the right to be forgotten is an obligation
of the latter sort—an obligation to be in a certain mental state; hence the obligation imposed
by the right makes no sense. Thus David Rodin writes that once we have “clearly analyzed” it,
we can see that
there is no right to be forgotten. There is not even a right to be remembered fairly.

This is for the simple reason that thought is free. We reject the notion that what we
remember, and what we forget, can ever be subject to moral or legal constraint.
(Rodin 2015)
The flaw in the nonsense criticism is its assumption that the obligation imposed by the right
to be forgotten must be an obligation to be in a certain mental state: the assumption fails to
appreciate that this obligation may best be understood as an obligation to do something. The
nonsense criticism’s other assumptions—that if the obligation imposed by the right to be forgot-
ten makes no sense then the Court’s affirmation of the right is wrongheaded, that the sensibility
of the imposed obligation requires that it not be an obligation to be in a certain mental state—
are eminently plausible; but the assumption that the imposed obligation must be an obligation to
be in a certain mental state is not plausible. And we can see this by understanding the imposed
obligation along the lines of the moral obligation to forget as I have construed it: as an obligation
to perform a certain epistemic activity.
The second broad criticism of the Court’s affirmation of the right to be forgotten, which I’ll
call “the intolerable conflict criticism,” is that the Court’s affirmation of the right is misguided
because the obligation the right imposes intolerably conflicts with other (equally if not more)
important rights, such as rights to freedom of expression and to freedom of the press. Stanley
Rosen anticipates this criticism when he writes that the affirmation of a right to be forgotten
online would represent “the biggest threat to free speech on the Internet in the coming decade”
(2012: 88). Emma Llansó expresses the criticism when she says she sees no way “to square” the
affirmation of such a right “with a fundamental right to access to information” (Manjoo 2015).
In a similar vein, Jeffrey Toobin claims that, given the nature of the right that it affirms and
the “American regard for freedom of speech, reflected in the First Amendment,” the Court’s
affirmation “would never pass muster under U.S. law” (2014: 26).
But does the obligation imposed by the affirmed right to be forgotten intolerably conflict with
other important rights? I think there is good reason to believe that it does not, and that my account
of the moral obligation to forget sensitive information about other agents can help us see this.
First, consider the plausibility of holding that the obligation imposed by the affirmed right
to be forgotten (hereafter, simply “the Court-imposed obligation”) has the moral obligation
of which I have provided an account as its moral ground. If the Court-imposed obligation
has any moral ground at all (and presumably the Court, at least, would want to insist that it
does), the moral obligation of which I have provided an account seems to be a particularly apt
candidate. Both the Court-imposed obligation and this moral obligation pertain to sensitive
information about agents;4 both apply to corporate as well as individual agents; both seem best
understood as requiring the agents to which they apply to perform an epistemic activity under
certain conditions, and both plausibly implicate a general right to privacy.5
If it is thus plausible that the Court-imposed obligation has the moral obligation to forget
sensitive information about other agents as its moral basis, and if my account of that moral
obligation is (at least along its central lines) plausible, then it is also plausible that the Court-
imposed obligation is constrained by conditions like (1)–(3) of my account. And if, moreover,
370
the Court-imposed obligation is plausibly constrained by such conditions, it would seem to

be keyed to important rights other than the affirmed right to be forgotten from which it
derives. To the extent that it is constrained by something like condition (2) of my account,
the Court-imposed obligation requires that the bearer of the corresponding right to be for-
gotten have a privacy right with respect to the relevant sensitive information about her; and
there is little doubt in the privacy literature that such a privacy right must be keyed to other
important rights.6 Hence to the extent that it is constrained by something like condition (2) of
my account, the Court-imposed obligation would seem to be keyed to these other important
rights. To the extent that it is constrained by something like condition (3) of my account,
moreover, the Court-imposed obligation requires that the agent on whom the obligation is
imposed can successfully perform the obligated epistemic activity without unduly diminishing
her ability to care for herself or others in morally appropriate ways; and whether she can so
perform that activity is undoubtedly determined in part by whether she can do so in a way that
respects other important rights that she and other agents possess. Hence to the extent that it
is constrained by something like condition (3) of my account, the Court-imposed obligation
would also seem to be keyed to important rights other than the affirmed right to be forgotten
from which it derives.
But if the Court-imposed obligation is so keyed to important rights other than the affirmed
right to be forgotten, there is good reason to believe that it does not intolerably conflict with
other (equally if not more) important rights. Thus my account of the moral obligation to forget
sensitive information about other agents can help us see the problematic nature of the intoler-
able conflict criticism: with this account, we may well have good reason to believe that the
Court-imposed obligation does not intolerably conflict with important rights other than the
affirmed right from which it derives, because with this account we can see that the Court-
imposed obligation is governed by conditions that key it to such rights.
Notes
1 Remembering of the preservative sort, that is. On whether there is an epistemically generative sort of
remembering that characteristically results in new knowledge, see Lackey (2005) and Senor (2007 and
Chapter 25 of this volume).
2 I take the moral obligation to forget sensitive information about other agents to be a species of the
moral obligation to forget because it seems to me that there can be situations in which agents are mor-
ally obligated to forget information of other sorts. Consider, for example, a field intelligence officer with
knowledge of highly classified design schematics or the access codes to top-secret databases. The infor-
mation she knows is not itself sensitive information about other agents, but it’s not difficult to imagine
situations in which she is morally obligated to forget it.
3 Commitment to the truth of such claims does not require commitment to ontologically extravagant
views according to which corporate agents, their moral obligations, etc. float free of individual agents,
their moral obligations, etc.; it requires only the rejection of eliminativist views according to which there
simply are no corporate agents, corporate moral obligations, etc.The philosophical literature of relevance
to this topic is large, but for one prominent recent treatment see List and Pettit (2011).
4 Thus, the Court-imposed obligation to forget pertains to such information as that at a certain time in the
past a specific individual’s real estate was auctioned for recovery of that individual’s social security debts
(CJEU 2014: §14; EC 2014: 1).
5 Cf. condition (2) of my account, and the fact that the affirmed right to be forgotten from which the
Court-imposed obligation derives is universally described in both the legal and the popular literature
(see, e.g., EC 2014: 4; Rosen 2012: 88; Manjoo 2015: passim) as a putative privacy right.
6 On the “distributive reductionist” account of the right to privacy I provide in (2008), this follows
from the very nature of the right, which itself implicates more fundamental rights to person, property,
and liberty.
371
David Matheson
References
Bach, K. (1981) “An Analysis of Self-Deception,” Philosophy and Phenomenological Research 41: 351–70.
Bach, K. (1997) “Thinking and Believing in Self-deception,” Behavioral and Brain Sciences 20: 105.
Barnier, A., M. Conway, L. Mayoh, J. Speyer, O. Avizmil, and C. Harris. (2007) “Directed Forgetting of
Recently Recalled Autobiographical Memories,” Journal of Experimental Psychology 136: 301–22.
Court of Justice of the European Union [CJEU]. (2014) “Google Spain and Inc v. Agencia Española de
Protección de Datos (AEPD) and Mario Costeja González, C-131/12, 13 May.” Retrieved from http://
curia.europa.eu/juris/document/document_print. jsf?doclang=EN&docid=152065.
European Commission [EC]. (1995) “Directive 95/46/EC of the European Parliament and of the Council
of 24 October 1995 on the Protection of Individuals with Regard to the Processing of Personal Data
and on the Free Movement of Such Data.” Retrieved from http://eur-lex.europa.eu/LexUriServ/
LexUriServ.do?uri=CELEX:31995L0046:en:HTML.
European Commission [EC]. (2014) “Factsheet on the ‘Right To Be Forgotten’ Ruling (C-131/12).”
Retrieved from http://ec.europa.eu/justice/data-protection/files/factsheets/factsheet_data_protection_
en.pdf.
Joslyn, S. and M. Oakes. (2005) “Directed Forgetting of Autobiographical Events,” Memory and Cognition
33: 577–87.
Lackey, J. (2005) “Memory as a Generative Epistemic Source,” Philosophy and Phenomenological Research
70: 636–58.
List, C. and P. Pettit. (2011) Group Agents: The Possibility, Design, and Status of Corporate Agents. Oxford:
MacLeod, C. (1998) “Directed Forgetting,” in J. Golding and C. MacLeod (eds.), Intentional Forgetting:
Interdisciplinary Approaches. Hillsdale, NJ: Erlbaum, pp. 1–57.
MacLeod, C. (2012) “Directed Forgetting,” in N. Seel (ed.), Encyclopedia of the Sciences of Learning.
Dordrecht: Springer, pp. 993–5.
Manjoo, F. (2015) “‘Right To Be Forgotten’ Online Could Spread,” The New York Times, August 5,
p. B1.
Matheson, D. (2008) “A Distributive Reductionism about the Right to Privacy,” The Monist 91: 108–29.
Matheson, D. (2013) “A Duty of Ignorance,” Episteme 10: 193–205.
Michaelian, K. (2011) “The Epistemology of Forgetting,” Erkenntnis 74: 399–424.
Rodin, D. (2015) “There Is No ‘Right To Be Forgotten,’” The Huffington Post Blog, February 10. Retrieved
from www.huffingtonpost.com/david-rodin/there-is-no-right-to-be-f_b_6645776.html.
Rosen, J. (2012) “The Right To Be Forgotten,” Stanford Law Review Online 64: 88–92. Retrieved from
www.stanfordlawreview.org/sites/default/files/online/topics/64-SLRO-88.pdf.
Sahdra, B. and P. Thagard. (2003) “Self-Deception and Emotional Coherence,” Minds and Machines 13:
213–31.
Senor, T. (2007) “Preserving Preservationism: A Reply to Lackey,” Philosophy and Phenomenological Research
74: 199–208.
Toobin, J. (2014) “The Solace of Oblivion: In Europe, the Right To Be Forgotten Trumps the Internet,”
The New Yorker 90: 26.
von Hippel, R. and R. Trivers (2011) “The Evolution and Psychology of Self-Deception,” Behavioral and
Brain Sciences 34: 1–16.
372
29
THE ETHICS OF MEMORY
MODIFICATION
S. Matthew Liao
1. Introduction
Many people would like to have a better memory. It could help them to do better at school,
at work, and remember where they put their car keys. Some people hope to stop or slow the
loss of their memory. An estimated 5.3 million Americans in 2015 have Alzheimer’s disease,
a progressive brain disorder that slowly destroys memory and thinking skills, and eventually
the ability to carry out the simplest tasks.1 And some people wish that they could get rid of
certain, traumatic memories. A 10-year-old girl, whose rapist was sentenced to 15–18 years
in Florida in 2014, told the court: “I wish I could forget and have a new brain, so I don’t
remember.”2 A woman who survived a violent sexual assault said: “Days after, my scars were
healing on the outside, but the memories hurt just as bad because no matter what I tried, I
couldn’t get them out of my mind . . . I will always have a constant reminder for the rest of
my life of the one night I wish I could forget.”3 Likewise, the terrible burden of memory
weighs heavily on returning soldiers, many of whom suffer flashbacks to those memories as
a symptom of Post Traumatic Stress Disorder (PTSD). For instance, Arthur Ricker Sr., a
94-year-old veteran of the D-Day invasion said: “It’s still in my memory. Sometimes I wish
I could forget but I can’t.”4
There are “low tech” means that can improve memory or reduce traumatic memories to
some extent, including better sleep, more exercise, and psychotherapy. But now scientists are
uncovering new, “high tech” means that promise to manipulate memories in more targeted
ways. It is possible that within the next decade memory modifications technologies (MMTs)
will be available for general use.
MMTs raise a myriad of ethical issues.5 I shall explore some of these issues here. I begin, in
the Section 2 of this chapter, with a brief overview of the memory system and some of the latest
science on how memories can be manipulated. In Section 3, I consider how MMTs could be
used to harm others and how they could exacerbate existing inequalities. In Section 4, I point out
that MMTs might also result in harm to self in subtle ways. I then argue in Section 5 that despite
these concerns, it can nevertheless be morally permissible to use MMTs in certain circumstances.
373
S. Matthew Liao
2. The science of memory modification technologies

Memory is a system for encoding, storing, and recalling experiences.6 When an individual
experiences a certain event, certain networks of neurons are activated in the individual’s brain.
Initially, the memory representations of the event are translated and held in the short-term memory
in an active, labile state. At some point, through consolidation, in particular, a process called
long-term potentiation (LTP), the neurons holding the memory representations become more
strongly connected, and the memory representations become encoded into long-term memory,
which is less active and more stable.7 The neurons associated with the memory representations
can then be reactivated again when presented with stimuli that are similar to the original stimuli,
enabling the recreation of past active states and associations.8 Here it is helpful to note that while
it is common to speak of “storing” memories, memories are not spatially localized; they are
instead spread across different structures, likely as distributed networks of potentiated synapses.
The latest scientific research suggests that there are numerous ways by which memories
can be enhanced, deleted, and even injected, at the biological level.9 For instance, there is
a wide range of nutrients, hormones, stimulants, and neuromodulators that can be used to
enhance memory consolidation.10 Ritalin, a stimulant methylphenidate that increases central cat-
echolamine levels, appears to improve working memory and verbal learning and memory,11
while modafinil, a wake-promoting stimulant with lower abuse potential than methylphenidate,
appears to enhance learning and memory.12 Genetic modifications of receptors, messengers,
and growth factors13 also seem to improve memory functions in animals, and there are similar
genetic targets for memory functions in humans.14
Furthermore, transcranial stimulation and deep brain stimulation also have the potential to
strengthen memories. For instance, non-REM sleep specific brain rhythms such as slow wave
oscillations and spindles are important for memory consolidation. It has been found that subjects
who received transcranial stimulation that boosted both slow wave oscillations and spindles
while they were sleeping have better task performance compared to control subjects who
received fake stimulation.15 Using deep brain stimulation, a method that involves placing elec-
trodes in deep neural structures, it has been found that stimulating the fornix/hypothalamus,16
or entorhinal cortex17 can also provide memory enhancement. In a program called Restoring
Active Memory (RAM), researchers at the U.S. Defense Advanced Research Projects Agency
(DARPA) have recently implanted small electrode arrays in brain regions involved in the for-
mation of declarative memory, spatial memory, and navigation.18 The researchers were able to
“read” the neural processes involved in memory formation and retrieval, as well as send signals
back to specific groups of neurons and influence the accuracy of recall.
At the same time, drugs such as scopolamine,19 benzodiazepines,20 and kinase inhibitors21
are known to block memory consolidation. Likewise, using the beta-blocker propranolol
shortly after a traumatic event appears to reduce the intensity of the memory and the risk of
PTSD,22 and targeting glucocorticoid, glutamatergic, GABAergic, and other receptors seems
to disrupt fear-related memories.23 Moreover, on the assumption that LTP is important for
the continued maintenance of long-term memory, there is growing evidence from studies of
rodents that certain proteins such as protein kinase M-zeta (PKMzeta) are instrumental for
maintaining LTP and strengthening the connections between brain cells.24 Strikingly, it has
been found that injecting protein inhibitors such as zeta inhibitory peptide (ZIP) appears to
cause specific memories to be deleted.25
Lastly, there is ample evidence that new and false memories can be induced. The memory
retrieval process is to a large degree reconstructive rather than a faithful representation of the
original experience. Indeed, each time we recall a particular event, a piece of memory may
374
The ethics of memory modification
be somewhat different and the memory can be affected by information available at the time
of recall.26 An example of inserting false memories is the “lost in the mall” study, which
involved giving subjects a list of childhood events supposedly described by family members
and then asking the subjects to recall details of the events.27 Mixed in with real events was a
false event such as being lost in the mall. The recall process caused the subject to elaborate false
details about being lost in the mall even though the event never took place.28 Recently, using
optogenetics—a technique that uses light to manipulate and study nerve cells that have been
sensitized to light—a group of researchers at MIT were able to identify, label, and reactivate a
small cluster of cells encoding a rodent’s fear memory, in this case a memory of an environment
where the mouse had received a foot shock.29 The researchers were then able to inject a false
memory into a rodent by causing the rodent to “remember” being shocked in one location,
when in fact the electric shock was delivered in a different location.30 In a follow-up study,
the researchers found that unpleasant memories in rodents can be neutralized and/or even
re-associated with more positive emotions.31 Together, all these findings suggest that soon we
may be able to use MMTs to modify people’s memories in targeted ways.
3. Memory modification technologies, harm to

others, and inequality
So what are some of the ethical issues surrounding the development and use of MMTs? In
the short term, the biggest challenges to the development of MMTs will be to make sure
that MMTs can function in a safe and targeted manner. As we have seen, memories are not
discrete objects, but overlap and interconnect. Any MMT that would inadvertently remove
memories that an individual may want to keep would obviously be undesirable. On the flip
side, MMTs that would enable an individual to remember everything in great detail are also
likely to be undesirable.32 One reason is that being able to remember things in great detail may
cause constant recollections of past experiences, which may undermine one’s ability to focus
on the here-and-now. Another reason is that when one remembers every detail of something,
it may become difficult for one to abstract and recognize larger patterns. Indeed, Solomon
Shereshevsky, a patient with exceptional memory, who was studied by the neuropsychologist
Alexander Luria, had trouble recognizing people by their faces, because he found people’s faces
to be very changeable from day to day.33
Suppose that these technical issues in the development of MMTs can be overcome. An obvi-
ous issue is that MMTs may be used to harm other people. Here are a few of the many ways in
which MMTs could be used for surreptitious purposes.
• A criminal commits a crime against his victim and then gives the victim a drug so that the
victim would not be able to recall the crime perpetrated against her.
• A sadist provides an individual with memory-enhancing drugs so that the individual would
remember more vividly every detail of the torture the sadist is about to inflict on the individual.
• An unscrupulous government attempts to inject false memories into its citizens so that the
citizens are more accepting of the unjust political regime.
A full discussion of how our society should prevent MMTs from being used in nefarious ways
is beyond the scope of this chapter. At the minimum, however, it seems that citizens should be
informed and consulted regarding such possibilities and there should be public dialogues about
how access to MMTs would be regulated.
375
S. Matthew Liao
As with other kinds of novel technologies, MMTs could also exacerbate the inequality
between the haves and the have-nots. When novel technologies first appear on the market,
they tend to be expensive and tend to be available only to the few who can afford them. If
MMTs were available only to the rich, the disparity between the quality of life enjoyed by the
rich and the poor could become even greater as the rich become more enhanced in comparison
to the poor, thus adding to the existing advantages that the rich enjoy. Any society that cares
about equality should regard it as highly important to minimize this effect. One possibility may
be to tax those who enhance and then use the proceeds to subsidize enhancement for the dis-
advantaged. Such an approach presupposes though an infrastructure that recognizes the benefits
of using drugs for enhancement as well as for therapy. This structure is currently lacking, since
drugs are currently funded, developed, and prescribed for therapeutic use, but not for enhance-
ment use. To ensure that MMTs can benefit everyone, it may therefore be necessary to rethink
the way our society currently prescribes drugs.
However, even if MMTs are not used to harm others or do not exacerbate existing inequali-
ties, MMTs may also result in harm to self in subtle ways.
4. Memory modification technologies and harm to the self

There are at least seven ways by which MMTs can result in harm to the self. First, memories are
evidence for events that have taken place and for one’s roles in these events. Modifying one’s
memories could therefore alter what one believes to be true about these events and cause one to
live in falsehood. This is most obvious in the case of inserting false memories, e.g., by deliber-
ately undergoing therapy that promotes the construction of them. This could also occur in the
case of reducing the emotional strengths of one’s memories. For example, in using propranolol
to weaken his traumatic memories, a soldier may come to remember and believe that he did not
really want to kill the enemy, when in fact he lusted after the killing.
Second, memories enable us to form a certain narrative identity, which is crucial to our hav-
ing a sense of what we believe to be true about ourselves. From our past experiences and our
memory of them, we come to have certain views about whether we are brave or cowardly,
altruistic or selfish, and generous or stingy. If we modify our memories in a certain way, this
may change what we believe to be true about ourselves and cause us to lose our true self. For
example, suppose that Beth’s memories of her father’s frequent disparaging remarks against
women were what prompted her to become and to self-identify as a feminist. Suppose that
being a feminist was an integral part of Beth’s true self. Suppose further that Beth found these
memories of her father to be hurtful. As a result, Beth decided to have these memories removed,
e.g., by triggering and reactivating these memories under the supervision of a therapist, who
then prevented their reconsolidation using pharmacological means, possibly combined with a
beta-blocker treatment to reduce the emotional impact. If being a feminist was an integral part
of Beth’s true self, Beth’s modifying her memories might result in her losing her true self.
Third, modifying our memories may affect our ability to respond as moral agents. When
an event occurs, as moral agents, there are more and less appropriate ways of responding to
these events. For example, if a friend betrays you, an appropriate moral reaction is to feel
indignation. Then, suppose that the friend apologizes. It seems that at some point, you should
forgive your friend. However, if you used MMTs to weaken your memory of your friend’s
betrayal, isn’t there a risk that you would forgive your friend too easily? Forgiveness is an
important moral reaction that requires us to overcome our indignation or resentment for
the sake of our moral values or personal commitments. A failure to feel such emotions may
preclude genuine forgiveness.
376
Fourth, memories are constitutive of learning. Without them, we would repeatedly ask the
same questions and make the same mistakes. Moreover, memories of past events can also help
us infer how we might act when confronted with similar situations in the future. If one modifies
these memories, one may be depriving oneself of the opportunity to learn about how one might
act when confronted with similar situations in the future.
Fifth, retaining one’s memories of an event gives one the opportunity to think through this
event for oneself and to address it if necessary. Doing so is exercising appropriate agency and
respecting oneself as an agent. When one modifies a particular memory, e.g., by removing or
changing it, one may be depriving oneself of the opportunity to think through this event for
oneself, thereby not respecting oneself as a normative agent. Consider an analogy. Suppose you
are doing some math problems, the answers of which are in the back of the book. You could
try to work through the problems or you could just look up the answer in the back of the book.
If every time you are stuck you would just look up the answer without thinking through the
problem, it seems that you would not be exercising appropriate agency, and you would not be
respecting yourself as an agent.
Sixth, we may have moral and legal duties to remember certain events. Memories serve as
evidence not just for oneself but often also for others. For example, Neil Armstrong’s memory
of landing on the moon, or a Holocaust victim’s memory, or a statesman’s memory of what he
did in office, may not just be evidence for them, but also for others. Some of these memories
might be so important to others that there is a duty to remember them, a duty that could not
be met if they were altered by MMTs. In these cases, using MMTs to remove certain memories
may prevent us from meeting our obligations to ourselves and to others.
Finally, even if certain memories are very important such that one should try to remember
them, at present, without MMTs, one may not be held responsible for forgetting. However,
once MMTs are available, one may be obligated to take MMTs to maintain these memories,
and one may be held responsible for forgetting. Compare: before the Internet, it was deemed
permissible not to respond to correspondences immediately. After the Internet, some people
believe that one is obligated to respond to emails as soon as one can.
In sum, the use of MMTs can result in subtle harm to self by denying access to important truths,
causing us to lose our true selves, affecting our moral reactions, reducing our self-knowledge, pre-
venting us from satisfying our obligations to ourselves and to others, and adding moral obligations
that did not exist before. However, while these concerns are important to keep in mind, they do
not entail that it is impermissible to use MMTs.
5. The permissibility of using memory modification technologies

To see why it can generally be permissible to use MMTs, let us consider the issues discussed in
the previous section in more detail.
First, let us grant that MMTs can sometimes alter what one believes to be true about certain
events and cause one to live in falsehood. Even so, it is helpful to point out that not all MMTs
would have such an effect. Some MMTs would involve restoring forgotten memories, while
other MMTs would involve removing trivial memories. These uses of MMTs should not affect
what one believes to be true and cause one to live in falsehood. Also, a bit of falsehood might
be welcomed in certain cases. Suppose that it were possible to inject false memories of one’s
having a nice holiday so that one can feel more relaxed. Doing so seems unproblematic as long
as having such false beliefs has no further untoward, downstream consequences. More impor-
tantly, there are circumstances where it seems beneficial to, and necessary for, one’s well-being,
that one removes certain memories. For instance, consider Sophie in Sophie’s Choice. Sophie was
377
S. Matthew Liao
tormented by her memory that in the concentration camp, she decided to save her son rather
than her daughter. In the end, partly as a result of her feelings of guilt regarding this decision,
she committed suicide. In Sophie’s case, suppose that it was not possible for Sophie to recognize
that she did not really have a choice at the concentration camp. Given that Sophie was suffering
unbearable pain as a result of her memory of what she had done, were MMTs available such
that she could use them to forget about her role in this tragic event, it seems that it could be
permissible for her to remove these memories using MMTs.
Second, regarding the issue that altering our memories may alter one’s true self, if a piece of
memory is integral to one’s true self, it seems likely that this memory would be nested in many
other beliefs and memories that are mutually reinforcing. If so, it seems unlikely that one would
be able to lose one’s true self just by removing this particular piece of memory, since one’s belief
about one’s true self is likely to be supported by other beliefs and memories as well. Also, sup-
pose that a piece of memory really were crucial to one’s true self. In most cases, it seems unlikely
that one would seek to modify it.
Furthermore, some people may seek to be a different person altogether. As long as the indi-
vidual is aware of the consequences, this does not seem problematic from a personal point of
view. Consider an analogy. Some people seek out spiritual conversions, fully aware that they
would become different people altogether. Doing so does not seem problematic from a per-
sonal point of view. In fact, MMTs may sometimes enable one to have access to one’s true self.
Following Harry Frankfurt’s distinction between first-order and second-order desires, where
one’s second-order desires represent one’s true self, one may have a second-order desire to be
a certain kind of person, but one may be prevented from becoming such a person because of
one’s past memories.34 Using MMTs to remove these memories may enable one to fulfill one’s
second-order desire and finally become the person whom one regards as one’s true self.
What about the concerns that modifying one’s memories may affect various aspects of one’s
agency such as being able to respond appropriately to certain events and being able to learn from
one’s mistakes? Again, it is helpful to remember that in some cases, an event might be so traumatic
that one would want to use MMTs to help one forget the event altogether. In such cases, it might
not matter to one whether one has had the opportunity to react in an appropriate way. Also, it
is important to note that MMTs may enable an individual to acquire self-knowledge in certain
circumstances. As an example, a person who has acted wrongly and caused a major harm may be
prevented from learning from this experience owing to massive guilt, making rational reflection and
assessment of his act impossible. Reducing the emotional intensity of the memory (possibly only
temporarily) may enable rational reflection and assessment of his role in this act.35 It is true that when
confronted with a traumatic event, it may be undesirable immediately to resort to using MMTs.
But suppose that one knows that one will not be able to handle a traumatic event and retain one’s
agency. In such a case, using MMTs could be permissible, because one would be putting oneself
in a position where one would still be a functioning agent after the event. For example, suppose
that some people with PTSD seek to use MMTs so that they can remain agents. It seems that they
would still be respecting themselves as agents given that their aim was to preserve their agency.
On the issue of a duty to remember, it may be helpful to distinguish between episodic
memory (memory of the experiences), and semantic memory (memory of the facts regarding
the event itself). If what is valuable are the facts regarding the event, it may be enough to have
a duty to retain semantic memory, where one may be able to satisfy this duty by storing one’s
memory elsewhere, e.g., in a diary. Indeed, lesion studies suggest that semantic and episodic
memory are dissociable and have overlapping yet also distinct neural basis.36
It might be thought that in addition to preserving semantic memory, there is also value in
preserving episodic or emotional memory. For example, it might be thought that if possible,
378
preserving Sophie’s full memory of her experience at the concentration camp would be valu-
able. This might be so, but if preserving Sophie’s episodic memory was too painful for her, it
might be too onerous to demand that Sophie retained her memory of the event. In any plau-
sible moral theory, moral obligations should typically not be so demanding such that one must
make enormous sacrifices in order to fulfill them. As Judith Jarvis Thomson observes, “nobody
is morally required to make large sacrifices, of health, of all other interests and concerns, of all
other duties and commitments . . . in order to keep another person alive.”37 It is difficult to say
exactly where the limit of moral demandingness lies. But, arguably, if Sophie would commit
suicide if she retained her episodic memories, then maintaining her memories would seem to be
too demanding for her, which implies that she was not obligated to do so.
We have granted that there can be a duty to retain some semantic memories. Is there a duty
to retain all semantic memories? It does not seem so, because if there were, one would be obli-
gated to retain even very trivial memories such as my memory that I had cereal this morning.
Regarding the concern that once MMTs become available, one would be obligated to use
them, this is a problem that applies to many, if not all, new technologies. It does not entail that
someone who has difficulty remembering must immediately resort to using MMTs. There may
be simpler means such as writing in a diary, asking a friend to remind oneself, and so on, which
may have less normative problems. Still, perhaps taking MMTs could sometimes be required if
no other means are effective. For example, a doctor in a remote region of the world may have
a duty to use MMTs to help her remember certain medical facts if she has no other means of
recalling these facts.
At this point, it is worth recognizing that our identities are amazingly fluid and inconsistent.
Psychological studies have shown that a surprising number of our own memories—even about
key autobiographical facts—are actually incorrect in some details or even outright false.38 The
open use of MMTs may therefore have the beneficial effect of making people more skeptical
about the veracity of their memories in general. Also, if facing up to reality is important, then
perhaps we should do more to prevent natural memory erasure—to improve memory in those
who forget too easily. This would be a reason in favor of developing at least certain kinds of
MMTs. Lastly, we should remember that many MMTs are currently being developed in order
to alleviate certain kinds of suffering such as PTSD, where the pain is unbearable. Indeed, it
seems permissible for Sophie to use MMTs to help her forget, even though doing so might
cause her to live in falsehood, to lose part of her core identity, and to not have a chance to come
to terms with the tragic event. If all of this is right, there are circumstances in which it can be
permissible to use MMTs, which means that our assessment of the permissibility of using MMTs
should be done on a case-by-case basis.
6. Conclusion
It may soon be possible to use MMTs to manipulate our memories. Similar to other novel tech-
nologies, MMTs raise ethical issues such as safety for their users, whether they might be used to
harm others, and whether they might exacerbate existing inequalities. In addition, MMTs also
raise distinct and interesting issues of self harm such as preventing one from accessing important
truths, affecting the way one reacts to moral situations, and precluding one from meeting one’s
legal and moral obligations. That MMTs can result in these subtle harms to self does not mean
that it is morally impermissible to use them. There are circumstances where the use of MMTs
can help us craft and shape the kind of identity that we want or ought to have. Hence, while
we should proceed cautiously when developing and using MMTs, we should also embrace the
potential that these technologies can have in enhancing and transforming our lives.
379
S. Matthew Liao
Notes
1 www.alz.org/facts/overview.asp#quickFacts.
2 www.gazettenet.com/home/11398328-95/richard-meyer-of-cummington-sentenced-to-15-18-years-
in-state-prison-for-sexually-assaulting.
3 http://nypost.com/2015/06/18/rape-victim-details-chilling-ordeal-at-attackers-sentencing/.
4 www.knoxnews.com/news/state/pair-of-greene-countians-among-dwindling-number-of-d-day-
veterans.
5 Elsewhere, I have examined a number of these issues. See, e.g., Liao and Sandberg (2008). This chapter
draws on that discussion and, among other things, presents the latest scientific findings on this topic.
6 For a view of how the memory system works, see, e.g., Atkinson and Shiffrin (1968).
7 Waugh and Norman (1965); Baddeley (1966a, 1966b).
8 See Kandel (2001); Hebb (1949); Squire and Kandel (2000).While long-term memory depends on LTP,
short-term memory appears to be independent of it. According to most biological theories of working
memory, short-term memory consists of self-sustaining neural activity patterns rather than synaptic
change (Fuster 1995).
9 For a more detailed discussion of some of the latest science on manipulating memories, see Spiers and
Bendor (2014).
10 Sandberg and Bostrom (2006); Lynch (2002).
11 Linssen et al. (2014).
12 Battleday and Brem (2015).
13 Tang et al. (1999); Routtenberg et al. (2000); Wang et al. (2004).
14 de Quervain and Papassotiropoulos (2006).
15 Marshall et al. (2006).
16 Laxton et al. (2010).
17 Suthana et al. (2012).
18 www.darpa.mil/program/restoring-active-memory.
19 Caine et al. (1981).
20 King (1992).
21 Pastalkova et al. (2006); Shema et al. (2007).
22 Pitman et al. (2002);Vaiva et al. (2003).
23 Kaplan and Moore (2011).
24 Ling et al. (2002).
25 Serrano et al. (2005). It was initially hypothesized that ZIP functions by blocking PKMzeta. However,
recent studies suggest that the specificity of ZIP remains unclear. See, e.g., Volk et al. (2013).
26 Schacter (2001).
27 Loftus (1997, 2003); Thomas and Loftus (2002); Hyman and Loftus (1998); Gonsalves and Paller (2002).
28 Lindsay et al. (2004).
29 Liu et al. (2012).
30 Ramirez et al. (2013).
31 Redondo et al. (2014).
32 Michaelian (2011).
33 Luria (1987).
34 See, e.g., Frankfurt (1988).
35 Greene and Haidt (2002).Whether lowering emotional arousal could contribute to more adaptive moral
cognition beside changed moral cognition is an interesting question. See Liao (2016), for a discussion of
these issues.
36 Nyberg (1996); Kensinger and Giovanello (2005).
37 Thomson (1997: 77).
38 Patihis et al. (2013).
References
Atkinson, R. C. and Shiffrin, R. M. 1968. Human memory: A proposed system and its control processes.
In: Spence, K. W. and Spence, J. T. (eds.), The Psychology of Learning and Motivation. New York:
Academic Press.
Baddeley, A. D. 1966a. Influence of acoustic and semantic similarity on long-term memory for word
sequences. Quarterly Journal of Experimental Psychology, 18(4): 302–9.
380
Baddeley, A. D. 1966b. Short-term memory for word sequences as a function of acoustic semantic and
formal similarity. Quarterly Journal of Experimental Psychology, 18(4): 362–5.
Battleday, R. M. and Brem, A. K. 2015. Modafinil for cognitive neuroenhancement in healthy non-
sleep-deprived subjects: A systematic review. European Neuropsychopharmacology, 25: 1865–81.
Caine, E. D., Weingartner, H., Ludlow, C. L., Cudahy, E. A. and Wehry, S. 1981. Qualitative analysis of
scopolamine-induced amnesia. Psychopharmacology (Berl.), 74: 74–80.
De Quervain, D. J. and Papassotiropoulos, A. 2006. Identification of a genetic cluster influencing memory
performance and hippocampal activity in humans. Proc. Natl. Acad. Sci. U.S.A., 103: 4270–74.
Frankfurt, H. 1988. The Importance of What We Care about: Philosophical Essays, Cambridge: Cambridge
University Press.
Fuster, J. M. 1995. Memory in the Cerebral Cortex, Cambridge, MA: MIT Press.
Gonsalves, B. and Paller, K. A. 2002. Mistaken memories: Remembering events that never happened.
Neuroscientist, 8: 391–5.
Greene, J. D. and Haidt, J. 2002. How (and where) does moral judgment work? Trends in Cognitive Sciences,
6: 517–23.
Hebb, D. O. 1949. The Organization of Behavior, New York: Wiley-Interscience.
Hyman, I. E. and Loftus, E. F. 1998. Errors in autobiographical memory. Clinical Psychology Review, 18:
933–47.
Kandel, E. R. 2001. Neuroscience: The molecular biology of memory storage: A dialogue between genes
and synapses. Science, 294: 1030–38.
Kaplan, G. B. and Moore, K. A. 2011. The use of cognitive enhancers in animal models of fear extinction.
Pharmacology Biochemistry and Behavior, 99: 217–28.
Kensinger, E. A. and Giovanello, K. S. 2005. The status of semantic and episodic memory in amnesia. Progress
in Neuropsychology Research: Brain Mapping and Language, Hauppauge, NY: Nova Science Publishers, Inc.
King, D. J. 1992. Benzodiazepines, amnesia and sedation: Theoretical and clinical issues and controversies.
Human Psychopharmacology, 1: 79–87.
Laxton, A. W., Tang-Wai, D. F., McAndrews, M. P., Zumsteg, D., Wennberg, R., Keren, R., Wherrett,
J., Naglie, G., Hamani, C., Smith, G. S., and Lozano, A. M. 2010. A phase I trial of deep brain stimula-
tion of memory circuits in Alzheimer’s disease. Annals of Neurology, 68: 521–34.
Liao, S. M. (ed.) 2016. Moral Brains: The Neuroscience of Morality, New York: Oxford University Press.
Liao, S. M. and Sandberg, A. 2008. The normativity of memory modification. Neuroethics, 1: 85–99.
Lindsay, D. S., Hagen, L., Read, J. D., Wade, K. A., and Garry, M. 2004. True photographs and false
memories. Psychological Science, 15: 149–54.
Ling, D. S. F., Benardo, L. S., Serrano, P. A., Blace, N., Kelly, M. T., Crary, J. F., and Sacktor, T. C. 2002.
Protein kinase M[zeta] is necessary and sufficient for LTP maintenance. Nature Reviews Neuroscience, 5:
295–6.
Linssen, A. M. W., Sambeth, A., Vuurman, E. F. P. M., and Riedel, W. J. 2014. Cognitive effects
of methylphenidate in healthy volunteers: A review of single dose studies. International Journal of
Neuropsychopharmacology, 17: 961–77.
Liu, X., Ramirez, S., Pang, P. T., Puryear, C. B., Govindarajan, A., Deisseroth, K., and Tonegawa, S. 2012.
Optogenetic stimulation of a hippocampal engram activates fear memory recall. Nature, 484: 381–5.
Loftus, E. F. 2003. Science and society: Our changeable memories: Legal and practical implications. Nature
Reviews Neuroscience, 4: 231–4.
Loftus, E. F. 1997. Creating false memories. Scientific American, 277: 70–75.
Luria, A. 1987. The Mind of a Mnemonist: A Little Book about a Vast Memory, Cambridge, MA: Harvard
University Press.
Lynch, G. 2002. Memory enhancement: The search for mechanism-based drugs. Nature Reviews
Neuroscience, 5(Suppl.): 1035–8.
Marshall, L., Helgadottir, H., Molle, M., and Born, J. 2006. Boosting slow oscillations during sleep
potentiates memory. Nature, 444: 610–13.
Michaelian, K. 2011. The Epistemology of Forgetting. Erkenntnis, 74: 399–424.
Nyberg, L. 1996. Classifying human long-term memory: Evidence from converging dissociations. European
Journal of Cognitive Psychology, 8: 163–84.
Pastalkova, E., Serrano, P., Pinkhasova, D., Wallace, E., Fenton, A. A., and Sacktor, T. C. 2006. Storage
of spatial information by the maintenance mechanism of LTP. Science, 313: 1141–4.
Patihis, L., Frenda, S. J., Leport, A. K. R., Petersen, N., Nichols, R. M., Stark, C. E. L., Mcgaugh, J. L.,
and Loftus, E. F. 2013. False memories in highly superior autobiographical memory individuals.
Proceedings of the National Academy of Sciences, 110: 20947–52.
381
S. Matthew Liao
Pitman, R. K., Sanders, K. M., Zusman, R. M., Healy, A. R., Cheema, F., Lasko, N. B., Cahill, L., and
Orr, S. P. 2002. Pilot study of secondary prevention of posttraumatic stress disorder with propranolol.
Biological Psychiatry, 51: 189–92.
Ramirez, S., Liu, X., Lin, P.-A., Suh, J., Pignatelli, M., Redondo, R. L., Ryan, T. J., and Tonegawa, S.
2013. Creating a false memory in the hippocampus. Science, 341: 387–91.
Redondo, R. L., Kim, J., Arons, A. L., Ramirez, S., Liu, X., and Tonegawa, S. 2014. Bidirectional switch
of the valence associated with a hippocampal contextual memory engram. Nature, 513: 426–30.
Routtenberg, A., Cantallops, I., Zaffuto, S., Serrano, P., and Namgung, U. 2000. Enhanced learning after
genetic overexpression of a brain growth protein. Proc. Natl. Acad. Sci. U.S.A., 97: 7657–62.
Sandberg, A. and Bostrom, N. 2006. Converging cognitive enhancements. Ann. N.Y. Acad. Sci., 1093:
201–27.
Schacter, D. L. 2001. The Seven Sins of Memory: How the Mind Forgets and Remembers, Boston, MA and New
York: Houghton Mifflin.
Serrano, P., Yao, Y., and Sacktor, T. C. 2005. Persistent phosphorylation by protein kinase m maintains
late-phase long-term potentiation. Journal of Neuroscience, 25: 1979–84.
Shema, R., Sacktor, T. C., and Dudai, Y. 2007. Rapid erasure of long-term memory associations in the
cortex by an inhibitor of PKM zeta. Science, 317: 951–3.
Spiers, H. J. and Bendor, D. 2014. Enhance, delete, incept: Manipulating hippocampus-dependent memo-
ries. Brain Research Bulletin, 105: 2–7.
Squire, L. R. and Kandel, E. R. 2000. Memory: From Mind to Molecules, New York: Owl Books.
Suthana, N., Haneef, Z., Stern, J., Mukamel, R., Behnke, E., Knowlton, B., and Fried, I. 2012. Memory
enhancement and deep-brain stimulation of the entorhinal area. New England Journal of Medicine, 366:
502–10.
Tang, Y. P., Shimizu, E., Dube, G. R., Rampon, C., Kerchner, G. A., Zhuo, M., Liu, G., and Tsien, J. Z.
1999. Genetic enhancement of learning and memory in mice. Nature, 401: 63–9.
Thomas, A. K. and Loftus, E. F. 2002. Creating bizarre false memories through imagination. Memory &
Cognition, 30: 423–31.
Thomson, J. J. 1997. A defense of abortion. In: Lafollette, H. (ed.), Ethics in Practice, Oxford: Blackwell.
Vaiva, G., Ducrocq, F., Jezequel, K., Averland, B., Lestavel, P., Brunet, A., and Marmar, C. R. 2003.
Immediate treatment with propranolol decreases posttraumatic stress disorder two months after trauma.
Biological Psychiatry, 54: 947–9.
Volk, L. J., Bachman, J. L., Johnson, R., Yu, Y., and Huganir, R. L. 2013. PKM-[zgr] is not required for
hippocampal synaptic plasticity, learning and memory. Nature, 493: 420–23.
Wang, H., Ferguson, G. D., Pineda, V. V., Cundiff, P. E., and Storm, D. R. 2004. Overexpression of
type-1 adenylyl cyclase in mouse forebrain enhances recognition memory and LTP. Nat Neurosci, 7:
635–42.
Waugh, N. C. and Norman, D. A. 1965. Primary memory. Psychological Review, 72: 89–104.
382
PART IX
History of the philosophy

of memory
30
PLATO
The first person in the Western canon to do serious philosophy about memory is Plato
(427–347 bc). In his works, Plato develops three separate main themes about memory.1 The
first is memory as knowledge-over-time. The second theme starts from the famous discussions
of recollection in the Meno and the Phaedo. For the third theme, the locus classicus is Plato’s
images of the Wax Block and the Aviary in the Theaetetus.
1. Memory as knowledge-over-time
The first theme, deducible from numerous remarks scattered through Plato’s works, boils
down to a relatively obvious thought. It is that memory is the temporal dimension of know-
ledge, and remembering the temporal analogue of understanding;2 after all, in Greek truth
itself, according to a long-familiar and probably correct etymology, is un-forgetful-ness,
a-lêth-eia. Knowledge, then, does not “flicker” like a bad Internet connection. Nothing can
be called knowledge unless it involves a stable, secure, and enduring disposition, e.g.,3 to
affirm a proposition’s truth. When we look at t3 at this disposition as having held over t1–t2,
we are treating knowledge at t3 as (at least involving) memory during t1–t2.
Likewise, no one can be said to understand at any t unless her understanding both lasts for
longer than just t, and is networked to a reasonably large number of other items of understand-
ing that she also affirmed in the past, and/or will affirm in the future. Hence, memory has high
epistemic value. Having a good memory is a key sign of intelligence and the ability to learn;
being able to keep one’s grip on, and to marshal and deploy, items of knowledge and/or under-
standing that one has acquired over some past time-span is an essential part of what it is to have
understanding and wisdom now.
For Plato, this ability is not always simply a discursively intellectual one. It is also at least a
necessary, perhaps even the sufficient, condition of virtue. So in the Republic (429b–c), Plato
speaks of courage and (by analogy) self-control as the preserving of an opinion about which dangers
a person should face or flee from, and which pleasures she should yield to or resist; and in the
Phaedrus (248c) the two main things that can go wrong with a soul are “badness” (kakia) and
“forgetfulness” (lêthê).
So far, so near-truistic. It is useful work in conceptual geography—and in its time, fabu-
lously innovative—for Plato to spell out these basic connections. But there is not much here
385
that anyone not in the grip of a theory is likely to dispute. Certainly Plato’s inclination to state
these obvious truths is invariant with respect to his other views about knowledge and memory.
Things get more controversial with Plato’s second theme about memory: recollection.
2. Recollection in the Meno

For an ancient Greek whose ear is as sensitive as Plato’s was to the nuances and history of words,
there must have been manifold etymological temptation to find something deep and essential
about connections between memory, understanding, and knowledge. The same Indo-European
root *m(o/e)ñt- is apparently there in plenty of languages’ words for mentality, mind, mean-
ing, and memory (Latin mens, memor, memini, Greek mnêmê, manthanw, mathêmatika, Sanskrit
mantra, manas, English moot and the four words at the head of this list, German meinen); in words
for remaining and abiding (Latin manere, mansio, Greek menein); and in words for measuring
(Latin mensere, momentum, Greek mêtron, English meet and mete, German messen, messer, -mäßig).
It may even be that, historically speaking, man is a measure(r) after all; maybe the moon (month,
mênê, mensum) too is another kind of measure(r).
Whether or not Plato is consciously appealing to anything like this busy philological back-
drop, one of his aims in the Meno (yes, that root again) seems to be the quite audacious project
of reducing learning and (intellectual) inquiry to recollection. The argument begins from a
puzzle that Meno presents to Socrates, and which Socrates himself then restates as this dilemma:
No one can inquire either into what he knows or into what he doesn’t know. He can’t
inquire into what he knows—for he knows it, and there is no need for this inquiry for
anyone who knows. Nor can he inquire into what he doesn’t know—for he doesn’t
know what it is that he should inquire into.4
(Meno 80e)
Inquiry is search (literally so, in the Greek: zêtêsis), and there is no searching unless you know
already what you are looking for. But if what you are searching for is not, say, a lost sock, but
an idea—something mental or intellectual in nature—then how can there be any gap between
quest and quarry? The quest is: looking for it, knowing what you are looking for; the quarry is: know-
ing it—which is already part of the quest. Here it is not so much “Seek and ye shall find” as “To
seek is already to have found.” But this is deeply puzzling. It certainly seems that we can inquire
or seek not only for missing socks, but also for missing ideas.5 If that sort of inquiry is in fact
impossible, then everyone who engages in it is, as Socrates says (Meno 81d), wasting her time.
She had better give up on hunting for ideas,6 and go back to sock-hunting instead.
As Plato presents Socrates’ response to Meno’s dilemma it is, on the one hand, a provokingly
logically untidy piece of rambling rhetorical exposition, barely deserving the name of argument,
leaning heavily on the shady authority of “priests, priestesses, Pindar, and godlike poets” (Meno
81a–b), and involving Socrates in some heavyweight metaphysical assumptions that most mod-
ern philosophers will run a mile from; not to mention that it involves at least one pretty clearly
fallacious inference. On the other hand, Socrates’ response is also a world-changingly brilliant
moment of philosophical creativity.
The essence of Socrates’ reply to Meno’s dilemma is that, in intellectual searching and where
lost socks are not in question, it is just true that to seek is already to have found. The stereotype
of all inquiry is indeed “looking for it, knowing what you are looking for”, and “knowing it” is
indeed already a part of this stereotype. In the intellectual case—where there can be no physical
dislocation between you and your quarry, as there can be between you and a lost sock—this
386
Plato
really does mean that, even at the beginning of the inquiry, you already have hold of what you
are seeking. But crucially, you can know what you are seeking without realizing that you know it.
What you need to do when you are in this condition is find a way to access, summon up, or
realize that you already know the answer to your question. This kind of bringing-to-consciousness
what you already unconsciously know is what Plato means by anamnêsis, recollection:
Since the soul is immortal and has come into being many times, and since it has seen
this world and the World of the Dead and everything there is, there is nothing that it
has not always already learned. Therefore it is no wonder that the soul should be able
to remember everything it ever knew, about everything including virtue; for the soul
already knew it all before. Everything that exists is akin to everything else, and the
soul always already knew everything. So once the soul has recollected just one thing
in this life—this recollection is what people call learning—there is nothing to prevent
the soul from drawing out of it everything else there is to know: if we are only coura-
geous in our quest, and do not grow weary. And so, inquiry and learning is entirely
recollection.
(Meno 81c–d)
The possibility of intellectual inquiry depends not on a gap between quest and quarry, or
between not-knowing and knowing: the advocates of Meno’s dilemma are right that in the
intellectual case there is no such gap. (The mental is, as one later Platonist—Descartes—will
famously put it, unextended.) Instead, it depends on a gap between unconscious and conscious
knowing. This is the gap that is crossed by recollection. This may be the first time in its history
that someone presents a distinction between the conscious and the unconscious.
However, an obvious objection looms: one that Charmides 166e–169c’s interesting (and osten-
sibly skeptical) discussion of self-knowledge suggests Plato was aware of, either already when he
wrote the Meno, or not long after. Socrates’ proposal offers us locutions of the form “I know X
but I have not recollected X,” or “I know X but I am unconscious that I know X.” Surely, runs
the objection, we can rephrase either of these as a claim about second-order knowledge: “I know
X but I don’t know that I know X.” But if this rephrasing is permitted, then we can treat the
gap between unconscious and conscious knowing as a gap between not-knowing and knowing
after all—only second-order not-knowing and knowing. And then Meno’s dilemma re-emerges at
this second order. Before we can cross the first-order gap between unconscious knowledge and
conscious knowledge, we must already have crossed the second-order gap between not knowing
(that we know) and knowing (that we know). But, the proponent of the dilemma will insist, this
move up an order makes no difference to the basic impossibility of getting from not-knowing to
knowing by way of inquiry (as opposed to, say, by luck or by divine dispensation: cf. Meno 99e).
“You cannot search for anything if you do not know what it is” is a ban that applies just as much
to second-order knowledge as to first-order. And—to forestall any further, regressive flight up the
orders—it will apply at every higher order of knowledge too.
We can’t tell, just from the circumstantial evidence that the roughly contemporaneous
Charmides discusses (and raises serious doubts about) second-order knowledge, whether Plato
has this objection in mind in the Meno. However, Plato’s best response to this putative objection
would be to stop the regress before it can get started. And this does seem to be what he does
in the Meno. He insists that his talk of learning as recollection is not, at any order, rephrasable as
talk of learning as coming to know: “inquiry and learning is entirely recollection” (Meno 81d3).
This last ringing pronouncement may have another purpose too, namely to put the Lost-
Sock Objection finally and decisively out of business. Plato is offering us an account of search or
387
inquiry that treats learning as recollecting things that I already know. The Lost-Sock Objection
threatens this by pointing out that, in any case of physical search, I don’t already know what I
want to know, namely the location of that darned sock; finding the sock is therefore an instance
of learning that is not recollection. Plato can respond by simply turning this on its head. (Can
respond: again, I don’t mean that he does so explicitly in the Meno, though some such reply may
already have been in his mind. This was his explicit response by the time he wrote the Republic.)
He can reply that if lost-sock cases do not satisfy his conditions for learning by recollection,
then what that casts doubt on is not his conditions for learning; it is the status of lost-sock cases
as genuine cases of learning. Such cases, he may suggest, do not involve recollection precisely
because they are not learning properly speaking, and do not issue in knowledge properly speaking,
but in some lesser epistemic state: something perhaps barely worth calling an epistemic state at all,
such as a merely physical reunion with a sock.
The scope that Plato gives “real” learning and “real” knowledge in the Meno is what we
now call the a priori, and it is right here, in this very passage (Meno 81a–86c), that the a priori/a
posteriori distinction is drawn for the first time ever. We are relatively accustomed these days to
the rather surprising view (for different versions of which see e.g., J.S. Mill, System of Logic, or
W.V.O. Quine passim) that the only real knowledge is a posteriori (empirical) knowledge. We
may be even more surprised to find that—at least on one possible reading—Plato’s view in
the Meno is the even less customary converse: the only real knowledge is a priori (conceptual)
knowledge. (So although on the face of it Meno 97a–b appears to contrast knowing the way to
Larisa with having true belief about the way to Larisa, it may well be that, strictly speaking, Plato
in the Meno takes it to be impossible to know the way to Larisa; since the way to Larisa is not an
a priori matter, neither can it be a matter of knowledge. Possibly parallel perplexities arise about
the famous example of the Jurymen, in Theaetetus 200d–201c.)
What we can know or learn—or even, properly speaking, inquire into—is only what we
can recollect; which is what we have always known. Certain kinds of truths, such as those
of mathematics, are unchangingly fixed in the nature of things; and we too are part of the
nature of things. What is eternal in us is, as Plato puts it, “akin” (syggenes)—as we might put
it, networked—with what is eternal in nature. This has the corollary that there is something
eternal in us, i.e., that our souls are immortal.
At this point, the reader might still be wondering what positive reason there is to believe
Plato’s account of learning as recollection. The most positive reason we have so far is the
escape that the recollection theory offers from Meno’s dilemma; but that move involves
the costs (1) of agreeing that Meno’s dilemma is not a mere sophistry, and (2) of denying
that there is any a posteriori knowledge. Has Plato any more positive reason why we should
embrace his theory?
Plato’s main positive motivation for his account of learning as recollection seems to be phe-
nomenological. The point, or one point, of the long lesson that Socrates gives the slave-boy
in geometry (Meno 81e–86c) is to describe and explore what it’s like to inquire and to learn.
Plato’s portrait of the activity of learning—even the learning of a maximally ignorant pupil
who needs a maximum amount of coaching from his interlocutor—suggests, among much else,
that a priori learning involves what feel like recognitions: moments of insight when we think “Of
course—obviously—that has to be true”; as a mathematician might come to see a proof, and
see it as a proof, while she was asleep. Plato thinks that these seeming-recognitions are literally
re-cognitions—encounters with things we’ve cognized before. He also thinks that, if he gives
the well-educated reader a vivid enough description of what it feels like to experience such
recollections—as he tries to in the exchange between Socrates and the slave-boy—then he can
get the reader to recognize the feeling that he means from her own experience.
388
Plato
Obvious skeptical questions arise at this point. Plato’s idea seems to be that the process of a
priori learning is (so to speak) trademarked by a particular feel, or group of feels, that go with the
experience “recollecting.” Even if that is right, it looks, logically speaking, like an irrelevance.
Why should having that feel—or any feel—be an incorrigible sign that one is undergoing
genuine a priori learning, rather than some impostor process that just feels like a priori learning?
And turning to the interlocutor, he and his leading questions look like an irrelevance too. Why
should anyone think that, in itself, the fact that a pupil has been led to adopt some a priori belief
by a series of suitably apt questions gives him, or us, any reason to think that that belief is true?
Why should a skillful series of questions be any trademark of truth either?
Plato does not address these skeptical questions directly in the Meno. It seems likely that, if he
did, he might give them a strikingly Cartesian answer (see Descartes, Meditations III–IV). The
seeing of real connections between the objects of recollection, he might say, is a process that has
infallible results, whenever the real connections are seen clearly and distinctly; whenever that happens,
it does have a distinctive phenomenology. Of course we can be mistaken about whether we
are seeing things clearly and distinctly, and about when we are experiencing the accompanying
phenomenology. Since these mistakes are certainly possible, no doubt the skeptic is right that
we cannot infallibly know whether we are undergoing the epistemic process whereby we can
infallibly know. Still, there is an epistemic process whereby we can infallibly know, even if we
can’t ever be infallibly sure we are in it.
Even if the Meno does not give this—or any other—direct response to these skeptical wor-
ries, we can hardly accuse Plato of failing to notice them. After the Meno, he is pretty well
constantly preoccupied with the question how to tell a genuine philosophical discovery from a
fake or merely apparent one: just that is exactly what a philosophical midwife does (Theaetetus
148e–151d). His interest never fades, either, in the accompanying question: how we tell a con-
structively leading series of philosophical questions, from a destructively misleading series. This
is one of the main reasons why he is always so keen—see Euthydemus, Gorgias, Sophist passim—to
know what differentiates philosophers from sophists.
What does fade out is the Meno’s insistence that the key to making these discriminations, and
more broadly the key to understanding what we pre-philosophically call learning, is in any very
definite sense a matter of recollection or remembering or re-cognizing anything. Is the reason perhaps
that, in the end, Plato abandons the Meno’s notion of recollection?
He never explicitly says he has done so. But it is not hard to see a route from the Meno’s
preoccupation with recollection, to the Republic’s related but rather different concerns, that
shows us how his theory of recollection might have faded into the background for Plato: not
faded out altogether, but ceased, at any rate, to be of live and central importance to his phi-
losophy. The route I have in mind takes us straight through one of the main arguments of what
may well have been the next dialogue that Plato wrote after the Meno: namely the Phaedo.
3. Recollection in the Phaedo

In the Phaedo Socrates is awaiting execution, and his friends’ and followers’ confidence in their
customary views is shaken. So they are seeking to find, indeed to recollect, arguments for the
immortality of the soul. Among the other arguments discussed, Cebes offers this:
Also, if it is true, in line with that argument that you used to make so frequently,
Socrates, that for us learning is really nothing other than recollection, then by this
argument it is presumably (pou) necessary that there must have been some earlier time
in which we learned what we now recollect. But presumably (pou) this is impossible,
389
unless our soul existed before it came to be in this human form . . . The best proof
of this is that when men are asked questions—provided someone asks them the right
way—then they state everything just as it is; even though, if they did not actually have
knowledge and right reason present in them, they would not be able to do this. And
then when someone confronts them with diagrams or something like that—that is
when it is more clearly demonstrated that this is how things are.
(Phaedo 72e–73b)
This is a clear back-reference to Meno 81c–86e, and Phaedo 72e–77a certainly takes over the
Meno’s theme of learning as recollection; it also adds some interesting further descriptions of the
phenomenology of recollecting (73d–e). However, the Phaedo transforms what it takes over in
at least two ways. First, the Phaedo like the Meno wants to offer an argument from recollection
to immortality. But it gives a different and in fact incompatible argument, and the Phaedo is
more specific about what that argument is, and more insistent on making it. Second, where the
Meno is all-inclusive about what is recollected—saying sweepingly that recollection can summon
up for us everything there is to know (hapanta, 81c–d)—the Phaedo is, again, more specific; and
this in a way that effectively moves Plato away from the theory of recollection, and towards the
theory of Forms.
We need a closer look at both these transformations. Meno 81c–d suggests this argument
from recollection to immortality:
Meno: Learning (strictly speaking) is impossible: there is only knowing, and recollect-
ing what we already know. But we do know some truths, and can recollect more.
Therefore we have never learned, and we have always known, the truths that we
recollect. Therefore we have always existed.
(Meno 81c–d)
Despite Cebes’ clearly signaled intention to allude to the Meno, the Phaedo not only does not
repeat this argument from recollection to immortality; it actually contradicts it. The argument
that the Phaedo twice offers us—through Cebes at 72e, through Socrates at 75c—is this:
Phaedo: At any point in this life, there are some truths that we can be made to recol-
lect by exercises like the slave-boy’s geometry lesson in the Meno. Therefore, there
are always some truths that we already know. Therefore, we cannot have learned
these truths in this life. But anything we know, we must have learned at some time.
Therefore, we must have learned these truths before we began this life. Therefore, we
existed before we began this life. Therefore, we have always existed.
The Meno argument says that learning is impossible; the Phaedo argument entails that learning
is possible. To say both that “Learning is really impossible, and what we call learning is actu-
ally recollection of pre-existing knowledge,” and also that “Anything we know, we must have
learned some time,” is flatly inconsistent. Nor can we fix this by adding, truly enough, that the
two arguments agree that learning is impossible in this life. The argument given in the Meno for
the impossibility of learning is Meno’s dilemma. If that works at all, it works in any life you like.
Maybe the Plato of the Phaedo no longer thinks that Meno’s dilemma does work. (Maybe
Cebes’ rather quick and loose presentation of the argument—with its repetition of the hesitant
particle pou, “I suppose” or “presumably”—is meant to signal this.) Certainly the Phaedo nar-
rows down the scope of what we recollect; it is no longer everything, but only specific kinds of
390
Plato
objects of knowledge. Alongside this new specificity about the objects of recollection comes a
new explicitness about the relation of thought and perception in the process of coming to know.
The key passage here is Phaedo 74a–75b: the objects of perception serve as reminders of
the absolute standards. Plato’s example of an absolute standard is equality, which may be an
unfortunate choice because, being a relation rather than a simple predicate, it raises difficul-
ties irrelevant to Plato’s main point.7 Here I will put Fness as a general term for any absolute
standard, irrespective of the adicity of the corresponding predicate.
F things conform to the standard set by Fness to some extent, but never perfectly. We need
perception to get our idea of Fness (74b); but what we perceive is always “F in one respect,
not-F in another” (74b–c). By contrast, the standard of Fness whereby we make these Fness-
judgments is itself always perfectly F (74d–e). Perception brings us knowledge of this standard
by reminding us of it (75a–b). But given their perfection and the imperfection of all objects of
this-worldly perception, standards like Fness are not themselves possible objects of (ordinary)
perception at all. So we already know Fness itself before we do any (ordinary) perceiving at all
of particular F things. But ordinary perceiving begins at the very beginning of this life (75b–c).
So before this life began we must already have known, by some special sort of perception, Fness
and all the other standards of which this-worldly perception now reminds us (76d).
Here the Phaedo gives an important positive role to perception in the acquisition of knowl-
edge. Plato is often accused of denigrating or disparaging the epistemic value of perception,
but the Phaedo does not do that. On the contrary, it says that we cannot attain knowledge
without perception. Certainly it argues that perception is not itself knowledge, but only a nec-
essary condition of knowledge: it is perception that reminds us of the absolute standards. This
means that, for the Phaedo, perceiving and knowing work together; it also means that they
work in the same world. There may be another world, the pre-birth one, where we perceive
or somehow cognize the Forms directly. But that is not how things are for us here and now;
here and now knowledge of the Forms and perception of the particulars that remind us of the
Forms go hand in hand as two aspects of the same knowledge-gaining process. If Plato is ever
guilty of the familiar charge of “two-worlds Platonism,” it is not in the Phaedo that he is guilty
of it. (The charge may be more fairly made on the evidence of the Republic and of Phaedrus
247c–249d; but there too, I suggest, contemplation of the Forms “themselves by themselves”
is an exceptional experience. The norm, for which Republic 520c2–3 says this contemplation
is a preparation, is encountering and negotiating a world that is simultaneously both perceptu-
ally and formally structured.)
The Phaedo argument is a turning point, because it begins to turn Plato’s focus away from
the Meno’s insistence that what matters for knowledge is a special faculty of recollection. As
we’ve just seen, recollection is still invoked in the Phaedo. However, the key thing there is not
recollection alone, but the combination of a special cognitive faculty with some special cognitive
objects—namely the absolute standards that the Phaedo begins to talk about, and which quickly
prove to be the Platonic Forms. This is the key combination in Plato’s Republic too, with the
further twist that, by then, it is no longer so clear that the crucial thing about the special faculty
whereby we cognize the Forms is that it should be any kind of memory particularly. Awareness
of the Forms, in the Republic, is the key focus, and this is taken to be a matter of knowledge,
understanding, or wisdom. Plato never denies in the Republic that knowledge of the Forms
involves recollecting. But it no longer seems as important as it did even in the Phaedo for him
to insist on that emphasis, perhaps partly because he no longer wishes to run the Phaedo’s argu-
ment from recollection for the immortality of the soul. The Plato of the Republic believes in
the immortality of the soul anyway; but not necessarily on the basis of the arguments adduced
in the Meno or the Phaedo.
391
And so fades out Plato’s early emphasis on memory. He returns to the topic of memory
with new arguments and new thoughts in the Theaetetus, where—as I said—his approach seems
quite different from, on some interpretations even inconsistent with, the Meno’s and Phaedo’s
arguments. I turn to the Theaetetus in Section 3 of this chapter.
4. Memory in the Theaetetus

Meno’s dilemma asks how someone can search for something unless she already knows what she
is searching for; but if she already knows what she is searching for, it is hard to see why she needs
to search for it. One possible response to the Meno’s attempted resolution of this dilemma is that
it is odd of Plato to appeal to memory to resolve it, since it is precisely in the case of memory that
that difficulty looks most acute. Compare what you do when you try to recover (a) a mislaid
sock, and (b) a mislaid name. When you look for a sock, what you do is try and grasp something
that you do not currently have in your grasp, because it is not currently in your physical locality;
this kind of physical search seems entirely unpuzzling. But what happens when you try to recall
a name that you have forgotten? Well, you aim to retrieve a name from your memory. But:
how? There is no point in trying to do this unless that name is already there in your memory,
right where you are and without any physical separation between you and it; so, it needs to be
something that you already know. How can it be possible for you to know the name, and yet
not be able to come out with it when questioned? Do you know it, or don’t you?
What exactly is the search-technique whereby you prompt yourself to remember the name?
Apparently the task prescribed me is to prompt into my conscious awareness some item that I have
forgotten. But under what description am I to get hold of the item when, given that I’ve forgot-
ten it, I have no description of it available—or at least, no description that I know applies to it?
My echo here of Meno’s own wording of his dilemma (Meno 80d) is deliberate. Difficult
and paradoxical questions really do lurk here about how to conceptualize what lies on, or just
beyond, the boundaries of our conscious experience, and about how to understand the pro-
cesses whereby we can, usually, deliberately prompt some particular item or items to cross those
boundaries. It seems fairly clear that the Meno is the first text in Western philosophy to raise
these questions, and that the Theaetetus is the first to offer an extended examination of some
attempts to answer them.
True, the Theaetetus’ discussion in 187e–200e is billed as a discussion of false belief, not of
mistakes about memory; and true, at an early point in the discussion (188a), Socrates is made
to restrict the discussion to knowledge and ignorance only, and say “For the moment I am not
going to consider learning and forgetting, which are so to speak intermediate states between
knowledge and ignorance”—as if the topic of learning and forgetting were cleanly separable
from the topic of true and false belief. However, as the discussion of false belief proceeds, it
becomes clearer and clearer to Socrates and Theaetetus that they can’t make progress without
looking at learning and forgetting, and so at what lies between those two processes, namely
remembering. Their two main attempts to do so are the famous episodes of the Wax Tablet
(Theaetetus 190e–196c) and the Aviary (Theaetetus 196c–200d).
The image of memory as a Wax Tablet is another of Plato’s most fertile tropes, an image that
is with us to this day, and has been enduringly important to empiricist philosophers like Locke,
Hume, and indeed Aristotle. It is Plato, in this very passage, who coins the empiricists’ famous
terminology of ideas (ennoiai) and impressions (via the verb anatypousthai). The image is not, in
fact, original to Plato; the idea of the mind as a place where one writes down memories occurs
at least once in tragedy (Aeschylus, Choephoroi 450). What is original to Plato’s use of the image
is the vividness and ingenuity with which he develops it:
392
Plato
For the sake of the argument, please suppose that there is in each of our souls a block
of writing-wax. The block is bigger in some people and smaller in others; it is made
of purer wax in some people, of less pure in others; it is harder in some and softer
in others; but there are those in whom its consistency is just right . . . whenever we
want to keep in our memory some one of the things that we see or hear or conceive
of for ourselves, we hold the wax tablet under these perceptions or conceptions, and
make an impression of them on it, as if we were using signet-rings to make wax seals.
Whatever we so imprint, we remember and know as long as its image remains there
on the block. But whatever gets rubbed out, or cannot be imprinted in the first place,
we forget and do not know.
(Theaetetus 191c–d)
Having a good memory, on this model, means having a wax tablet which is neither too
hard (so that nothing makes an impression) nor too soft (so that everything does, and all
the resulting impressions are overlaid and blurred together), and which is made of pure wax
(so that impurities do not prevent it from reflecting the shapes imposed on it). Similarly,
being intelligent means having a large enough wax tablet to store lots of such impressions
(194c–d). Truly a seductively simple and straightforward picture of the mind, and of the part
that memory plays in belief, understanding, and knowledge; its longevity after Plato is not at
all hard to understand.
Yet Plato himself no sooner throws out this striking image than he, well, throws it out.
(Contrast Aristotle, who gratefully takes over the wax-tablet image for his own purposes.)
Plato rejects the Wax Tablet immediately, because (“Socrates” argues, Tht. 195a–196a) the
Wax Tablet can only account for false belief in cases where a memory—an impression on the
tablet which is taken from a past perception—is mismatched with a present perception or with
a thought. But there are false beliefs which are just about thoughts, and have never involved per-
ception at any point; conceptual and mathematical false beliefs, for instance. So the Wax Tablet
cannot give a general account of how we come to have false beliefs.
Does that mean that the Wax Tablet achieves nothing? It might fail to account for concep-
tual false beliefs, yet still stand as an account of some of our false beliefs, and more generally of
parts of our mental life, including memory. How much it achieves is a vexed interpretive ques-
tion; the answer depends partly on what you think Plato is doing in the Theaetetus as a whole.
Some interpreters see the Theaetetus overall as a series of sincere but unsuccessful attempts on
Plato’s own part to give a positive account of knowledge, one that will, ideally, culminate in an
actual definition of knowledge. Others think that the Theaetetus is not this sort of failure at all,
but rather that it is a successful attempt to show, by implication, that there can be no defining
or positive account of knowledge for those who make certain assumptions that Plato himself
rejects, and is using the Theaetetus to attack. On the latter view—which I have defended in
Chappell 2005—many think it much less likely that Plato himself can accept the picture of our
mental life, and of memory, that is suggested by the Wax Tablet. But in fact the former view
too might be obliged to write off the Wax Tablet as just another failure in the list of blunders
that it sees in the Theaetetus. Moreover, one can take the latter view and still agree that the
Wax Tablet picture of knowledge and memory is of the greatest philosophical interest, and
also that the detail of the view is pretty definitely Plato’s own invention—whether or not Plato
himself endorses it.
What is certain is that the Theaetetus makes no obvious attempt to salvage anything from
the refutation of the Wax Tablet that Socrates completes by 196c. Instead Socrates and
Theaetetus move on to a new and—to all obvious appearances—wholly independent attempt
393
to account for false belief. This is the Aviary, which also implies some extremely interesting
claims about memory:8
[It is] possible for someone to possess knowledge but not hold it . . . As if some hunter
should catch some wild birds . . . and build them an aviary and keep them as pets at
home. In one sense . . . he holds them all the time, because he possesses them . . . In
another sense he doesn’t hold any of the birds. But he has got them under his control
in a cage that belongs to him . . . So he has acquired a power over the birds to take
hold of them and grasp them whenever he wants, by hunting down whichever bird he
likes, and then letting it go again. And he is able to do this as many times as he pleases.
(Theaetetus 197c–d)
The Aviary has the generality that the Wax Tablet lacks, because the objects of knowledge or
memory that it treats as birds in the Aviary can, for all it says, be of any kind at all—they don’t
have to stand in any particular relation to perception. What is crucial about the “birds” is simply
the distinction between “possessing” and “holding” them, between having them in your avi-
ary and having them in your hand. Like the Wax Tablet’s distinctions between the different
routes whereby something can enter our minds, the possessing/using distinction seems of the
first importance for the philosophy of memory. It seems closely related, too, to the distinction
between known-but-not-recollected and known-and-recollected that dominated the Meno and
the Phaedo; also to the modern dispositional/occurrent distinction.
Yet here too Plato no sooner marks this key distinction than—apparently—he abandons the
whole idea of the Aviary, and for the same reason as before: because his objective in this part of
the Theaetetus is an account of false belief, and the image of the Aviary cannot give him that. As
200a–b explains, the Aviary fails to account for false belief because it involves us in supposing
that someone can form a false belief by voluntarily picking the wrong “bird,” e.g., by picking the
11-bird as the answer to 5 + 7 = ?. But to do that is to mistake the 11-bird for the 12-bird. And
that sort of mistake cannot be the answer to Plato’s puzzle about false belief; it is Plato’s puzzle
about false belief.
We are left, then, with the same uncertainty as we were left with by the Wax Tablet, about
how much of the philosophical material that the Aviary gives us Plato wants to keep, and how
much he abandons. Certainly there are materials here that might have considerable philosophi-
cal value, in the right context. It is not obvious that Plato ever puts them to any constructive
use anywhere outside these passages, or even wants to. That, of course, need not restrain anyone
else from appropriating these materials for the purpose of articulating their own philosophy of
memory. It certainly didn’t restrain Aristotle.9
Notes
1 There are familiar and often uncrossable gaps between Plato’s texts and Plato’s intentions, and between
Plato’s Socrates and the historical one. But for present purposes and for ease of exposition, I permit
myself here to speak loosely of “Plato’s aims” and the like.
2 Likewise, as Plato argues in Philebus 34b ff., memory is the temporal dimension of experience, and in
particular of our experience of pleasure and pain.
3 “E.g.,” because Plato is not obviously an adherent of the broad modern consensus in Anglophone ana-
lytic philosophy that knowledge is primarily, perhaps exclusively, propositional. Alongside propositional
knowledge, there is room in Plato’s epistemology for basic and distinct notions of objectual knowledge,
knowledge how, and acquaintance-knowledge. These notions may even be indispensable to Plato’s epis-
temology. See Chappell (2014: Ch.11).
394
Plato
4 All translations used here are my own.

5 Caveat: this distinction between physical and intellectual searching is my interpolation. The Meno does
not make this distinction—not explicitly. But it seems essential to interpolate it, if only on grounds of
charity.Without the distinction, Socrates has no answer to what we may call the Lost Sock Objection—
as indeed many puzzled commentators have assumed he hasn’t. Where possible I prefer not to impute
uninterestingly lame arguments to indisputably great philosophers. And not merely lame; if Plato’s argu-
ment can be stopped in its tracks by a lost sock, it is holey and bootless too.
6 Williams (2005: 74, fn.47) reminds us of Thomas Gilby OP’s striking remark that “Man is the hunter of
forms.” I don’t know if Gilby said this with Plato’s Sophist in mind; but taken that way, it could be seen as
echoing one reason why Plato starts the Sophist by talking about hunters. Cf. Theaetetus 197c1.
7 For these difficulties see Bostock (1986: Ch. 4). My own suspicion is that Plato’s choice of “equality” as
his example of an absolute standard is prompted by an unexpressed background view, held at the time of
the Phaedo but already eclipsed by other ideas by the time of the Republic, that justice is a kind of equality, or
at least has something crucial to do with equality (probably equality between desert and receipt; certainly
not equality between citizens).
8 Another interesting point about the Aviary is that it explicitly rejects the idea of innate knowledge:
“When we are children, this aviary is empty” (197c).This may be a signal that the Aviary is not intended
to be Plato’s own model of knowledge or memory. Although, as above, there is every reason to think
that Plato’s central interest in recollection has faded out by the time he writes the Theaetetus, there is
no evidence that he ever actually rejects the notion of innate knowledge. Unless of course passages like
this are taken as evidence that he does so; but they needn’t be, and on my reading of the Theaetetus,
shouldn’t be.
9 Thanks for comments to Marcel van Ackeren, Carol Atack, Richard King, Kourken Michaelian, David
Robjant, and Rui Zhu.
There is one further significant argument about memory in Plato, at Philebus 34b ff. It turns out to be
more convenient to discuss this in the opening pages of my chapter on Aristotle on memory: qv.
Bibliography
Ackrill, J. L. (1997) “Anamnesis in the Phaedo, Remarks on 73c–75c,” in his Essays on Plato and Aristotle.
Oxford University Press (New York, 1997), pp. 13–32.
Bluck, R. S. (ed.) (1961) Plato’s Meno. Cambridge: Cambridge University Press.
Bostock, D. (1986) Plato’s Phaedo. Oxford: Oxford University Press.
Chappell, T. (2005) Reading Plato’s Theaetetus. Indianapolis, IN: Hackett Publishing.
Chappell, T. (2014) Knowing What to Do. Oxford: Oxford University Press.
Fine, G. (2016) The Possibility of Inquiry. Oxford: Oxford University Press.
Frentz, T. (2006) “Memory, Myth, and Rhetoric in Plato’s Phaedrus,” Rhetoric Society Quarterly, 36(3):
243–62.
Scott, Dominic (2006) Plato’s Meno. Cambridge: Cambridge University Press.
Williams, Rowan (2005) Grace and Necessity. London: Continuum.
395
31
ARISTOTLE
A philosophical theory of memory needs to satisfy a rather long and demanding list of desiderata.
Memory is not perception, and memory is not knowledge; but perception can give rise to knowl-
edge and to memory, memory can give rise to knowledge, and knowledge can give rise to memory.
So the three seem closely related; our theory needs to capture and explain their relations, and the
similarities and differences between them. Also, memory seems to involve a distinction between
potentiality and actuality: if you ask me to remember something, I often can remember, but typi-
cally I don’t actually remember until you prompt me to. (And notice the ambiguities in these bits of
ordinary language: “I didn’t remember when you first asked me ‘Do you remember?’, but now I do
remember, so yes, I did remember.”) Our theory should explain these puzzles; not to mention
the connected puzzle about what it is for me to “store” a memory, and how I recall memories
from that “store.” Then there is a puzzle about how the process of recall can go wrong, as it
clearly can in at least two ways: I can fail to remember and I can misremember. What do our
accounts of these phenomena say about what it takes for the process of remembering to go right?
And what exactly, to raise a presumably connected question, is memory’s physical basis? Again:
if knowledge comes in varieties—maybe (as I have argued elsewhere1) the four varieties proposi-
tional, practical, phenomenal, objectual—does memory come in the same varieties?
Aristotle’s philosophy of memory is mainly found in the short treatise in his Parva Naturalia
(“short scientific works”) called Peri mnêmês kai anamnêseôs, “On Memory and Recollection,”
which I’ll here refer to by its customary name, de Memoria (abbreviated dM). In this remarkable,
subtle, and surprisingly neglected2 treatise, Aristotle shows in the most obvious way that he is
aware of at least most of these desiderata, namely by seeking to address them.
He does so against two important backdrops. One is his Platonic heritage. The other is his
broader commitments in metaphysics, psychology, epistemology, and the philosophy of mind
and perception. To summarize what I’m going to do here: first we will try and gain some com-
mand of these two backdrops; then we turn to dM itself, taking its discussion of mnêmê and
anamnêsis in turn.
1. The Platonic background

For Plato in his earlier works, most obviously the Meno and Phaedo, memory—as “recollection,”
anamnêsis—is a pointer to our immortality, and a sign of how the soul transcends the body.
396
Aristotle
Our minds can access a priori truths because these (all of them!) are written into our minds from
eternity; in accessing these truths by recollection, we come to the securest knowledge we can
possibly have. For Plato in his later works too, memory is still recollection in something like this
sense; it is a way of accessing the Forms, and truths about the Forms.
Even if this account of memory as recollection is correct, it can hardly be a complete account,
for a very simple reason: this isn’t what ordinary people typically mean when they talk about
memory (or mnêmê, in Plato’s Greek). What they mean is, obviously enough, things like rec-
ognizing a face that you’ve seen before (Theaetetus 144c), or recalling a road that you’ve walked
before (Meno 97a). Quite possibly, there is no one thing that memory is; compare the varieties
of knowledge noticed above, and ask yourself how similar e.g., remembering facts and remem-
bering how to do things and remembering smells and remembering people are. But at any rate it is
surely clear that memory is not just the one thing that Plato calls “recollection.”
Apparently under the pressure of this criticism—it may well have been his pupil and
junior colleague Aristotle who first pressed it—the later Plato moves to a wider conception,
on which memory is a name for a variety of things, some of which are clearly very differ-
ent from what Plato calls recollection. In the words of Philebus 34a10, memory is also “the
preservation of perception,” sôtêria aisthêseôs, the retaining in the mind of an awareness of a past
perception. The Republic’s discussions of the psychological results of an education in poetry,
and the Theaetetus’ discussions of the possible relations between perception and knowledge
in the “Wax Block” and “Aviary” sections (Tht. 192a–200c), not to mention the Theaetetus’
case of the jurymen (Tht. 201a–c), all already tacitly assume the possibility of memory in this
sense—as indeed, as already pointed out, does the case of the road to Larisa (Meno 97a).
Hence Philebus 34b just makes explicit a distinction between “recollection” and “memory,”
anamnêsis and mnêmê, that Plato has already, perhaps inadvertently, talked himself into deploy-
ing. The Philebus then immediately argues that not only recollection but memory too, bodily
though it seems to be, is impossible without the involvement of the soul. At least in humans,
all desire (says Plato) presupposes two abilities: (a) to wish for things to be otherwise than they
are, and (b) to remember that they can be otherwise (Philebus 34c–36c). But no one can have
either of these abilities on a merely physical basis; both abilities, because they are abstracting and
universalizing abilities, necessarily involve the soul. So Plato concludes that mnêmê, for all its
apparent physicality, is just as much evidence for a separately existing soul, once it is properly
understood, as anamnêsis is.
In dM, Aristotle begins from this later-Platonic distinction. (The work seems deliberately
named to reference the Philebus discussion.) But Aristotle does not see Plato’s distinction as
moving us towards Plato’s conclusions, and in particular not towards Plato’s kind of dualism.
The point of dM is to repurpose the Philebus’ distinction, to point it in a different direction;
unsurprisingly, an Aristotelian one.
Before we look at how Aristotle’s argument goes in detail, we should sketch in the other
kind of backdrop I mention above, namely the wider philosophy of mind and perception within
which Aristotle’s account of memory finds its specific place.
2. The Aristotelian background

The general principle of Aristotle’s philosophy of mind is laid down by his even more general
metaphysical commitment to hylomorphism. According to Aristotle, any sublunary thing at all
can be metaphysically analyzed as matter (hylê) plus form (morphê or eidos). There is the stuff that
a thing is made of, and there is the thing that that stuff is made into. So with a statue, or a tree,
there is the matter (bronze, wood) and there is the form (Pericles, this particular beech-tree).
397
The matter is the ingredients or components or materials of the thing; the form is what the thing
is. The thing could not exist without its matter; the matter would not exist as something unless
it was this something, this thing here (a tode ti, “a particular”; in the case of a natural thing like
a tree, “an individual substance”). So any particular thing at all can equally be seen as informed
matter or enmattered form. The form is the actualization of the matter; the matter is potentially
what the form makes it actually: the matter is the “underlier” that can change into a whole
variety of things, and the form is the “supervening” nature that it does change into. The bronze
has the potential to become (inter alia) this statue of Pericles here, and the wood, leaves, bark,
sap, etc. (fill in modern scientific details as required) have the potential to become (inter alia) this
tree here. This potential is realized, precisely, in the coming-into-existence of the actual statue
or tree itself.
Strikingly, and audaciously—perhaps even recklessly3—Aristotle’s philosophy of mind in
general, and his philosophy of perception specifically, rest both of them upon applications of
this idea. Aristotle is as happy as Plato to talk of body and soul, sôma and psychê. But for perhaps
the first time in the history of Greek philosophy, Aristotle offers a clear and philosophically
sophisticated view of this relation. Aristotle’s thesis, most clearly announced in de Anima (dA)
2.1, is that soul is to body as form is to matter. The soul is the life of the body. The soul is what
makes the body perform its distinctive functions, develop its distinctive capacities, and keep
itself in existence by way, inter alia, of what we now call homeostatic feedback—self-monitoring
self-adjustment, as and when necessary, away from excess or defect and back towards the golden
mean of the species-norm.
Aristotle’s use of a hylomorphic framework doesn’t stop with the mind–body relationship.
He applies the hylomorphic model to the philosophy of perception too. According to him,
when I perceive a tree, my perceptual organs are (in the literal and in fact the original sense)
informed by the tree: for me to perceive a tree is for me to take on the form of the tree. (See de
Anima 2.5, and in particular the claim that it leads up to at dA 418a3: “The perceiver is poten-
tially what the percept is in actuality.”)
There is a Platonic background here too. The Theaetetus’ famous metaphor of the process
of perception and memory as the imprinting of an impression on a wax block is one that Plato
himself is suspicious of: it fits with an empiricist picture of perception and memory that is not
Plato’s own. Aristotle, however, accepts the wax-block picture (at dM 450b1–11 he quotes
Tht. 194c–d pretty well verbatim, in order to explain a variety of ways in which there can be
both misremembering and failing to remember). After all, when an impression is made in a wax
block, this is a clear case of one thing taking on the form of another thing, and so seems tailor-
made to go with a hylomorphic analysis.
How can this imprinting metaphor be cashed out in a serious and scientific theory of literal
perception? A moment’s thought shows that, as things stand, the metaphor is only a metaphor.
The wax block that takes on the impression of a signet-ring does not thereby perceive the
signet-ring; the mirror that reflects my shape and appearance does not thereby perceive my
shape or appearance. When I sit down, the armchair shapes itself to my shape, but that doesn’t
mean it is perceiving my backside. Aristotle may be going some way towards acknowledging
this difficulty when he notes in de Anima 2.6 (418a16 ff.) that perceiving organs and percepts
have to be “appropriate” (idia) to each other. So perception involves a case of the hylomorphic
relation; but a special case.
But special how? If a mirror cannot be said to see a tree just because the mirror reflects it,
how can the eye’s pupil be said to see that tree just because the pupil contains an image of
the tree? We need to know what makes the eye the right kind of material to receive the tree’s
398
Aristotle
form in the perceiving way. It is quite unclear how Aristotle’s theory of perception can answer
this last question. The key part of any decent response will surely consist, not in deploying the
hylomorphic framework, but in specifying details that go beyond that framework and, in truth,
have nothing specific to do with it at all.
These are the problems in the philosophy of mind and perception that, together with his
Platonic heritage, form the essential backdrop to Aristotle’s philosophy of memory in de Memoria.
3. De Memoria: (a) On memory (449b9–451a17)

As its full Greek title suggests, dM is largely structured by the contrast between its two epo-
nyms, mnêmê (considered in dM 449b9–451a17) and anamnêsis (dM 451a18–452b7); two shorter
concluding sections consider the relation of memory and recollection to time-perception
(452b8–453a4) and to physiology (453a5–b7). My main division here is between memory (3a)
and recollection (3b); the discussion of memory and time, I consider briefly at the end of (3a);
the physiological remarks we can deal with along the way.
The contrast between “memory” and “recollection,” Aristotle thinks, is a real one; but it is
not the contrast that Plato was drawing at the time he wrote the Philebus. It is, in essence, sim-
ply the contrast between retaining a perception and recalling that perception.4 But some further
stage-setting will be necessary to frame that answer.
Characteristically, Aristotle’s first observation about memory seems platitudinous, but in fact
has important consequences. It is that memory is necessarily of the past, and cannot be of the
future or the present (dM 449b9 ff.), to which expectation/opinion (elpis/doxa) and percep-
tion (aisthêsis) respectively relate. This immediately draws a clear line between what Aristotle
means by mnêmê and what Plato, a lot of the time, seems to have meant by anamnêsis. What the
slave-boy “recollects” in the most famous episode in the Meno (81e–86c) is geometrical truths
and relations. But these are not past at all—though not present either; it makes more sense to
see them as atemporal truths. So whatever is going on in the case of the slave-boy, Plato must
have been wrong to describe it, as he was happy to at the time, indifferently as a matter of
recollecting or remembering.
A second reason why “Memory is necessarily of the past” is worth saying is that it draws our
attention to a contrast that, for most of the first part of dM, interests Aristotle much more than
the memory/recollection contrast; namely the memory/perception contrast. If memory is “the
preservation of perception,” it seems to follow that perception becomes memory given a lapse of
time—as Aristotle seems to suggest at dM 449b24–30. But how does that happen? What picture
of perception makes this idea workable?
As Aristotle says at dM 449b31–32, his answer comes from his account of phantasia in
dA 3.3. Phantasia in Aristotle means, basically, the mind’s reception of phantasmata, perceptual
images. (Since it’s a technical term of his philosophy, I think it best to leave the word untrans-
lated.5) Aristotle’s idea is that when the mind is “impressed” by a perception, the process of
perception imprints an image upon the mind in something like the way that a signet-ring
imprints its seal upon the sealing-wax (dA 412b6, dM 450a31–2). The perception is the process
(kinêsis) in which the “shape” of the percept imprints itself; the memory is the condition (hexis)
of the mind that retains this “shape” afterwards, with the shape of the percept, so to speak, left
behind in the wax; and the phantasma is this shape. So again we have a form/matter relation
between perception and memory: the form and matter of the percept are both present in the
perception, but what is left behind in the memory is just the image of the percept—its “sensible
species,” its form.
399
When we consider what Aristotle actually wants to do with this picture of perception and
phantasia, it rapidly appears that it is only a picture, and that his ambitions go well beyond what
the picture entitles him to. Literal imprints in a literal wax tablet just sit there, doing nothing
very much. (Perhaps it is this sort of passivity and inertia that prompts Aristotle to open his dis-
cussion of memory by saying that the slow-witted are better at remembering, the quick-witted
at recollecting: dM 449b7–8.) The wax-block metaphor runs out because the phantasmata of
Aristotelian phantasia very clearly have a life, and a liveliness, of their own. Phantasia in Aristotle
is not just the condition of having impressions sitting there passively in your mind; dM 450a1
tells us that it is a necessary condition of any thought at all, and clearly most thought is not pas-
sivity but activity.
In this respect, Aristotle’s phantasia resembles what we call the imagination: roughly, a fac-
ulty of more or less creative visualizing. (Or otherwise evoking past perceptions. Pace some
commentators, nothing in Aristotle’s account of phantasia forces him to take it as specifically
about visual perception. Still, the priority of the seeing modality of perception in his account of
phantasia is obvious.) Phantasia involves not just inert imprints, but further processes of combi-
nation and recombination of those imprints, for example in dreams and delusions (dM 451a9),
in thought about the future, and again in intentional mental acts of visualization like those that
a geometrician—or a poet—might perform (dM 450a1 ff.). It is a very striking point about
Aristotle’s philosophy of mind that his prôton aisthêtikon, his “basic perceiving mind,” has as its
basic activity not only perceiving, but phantasia as well (dM 450a11–13). Indeed there is a sense
in which phantasia is even more basic than perceiving. For all perceiving is or involves phantasia
(there is no “blind-sight” in Aristotle). But though all phantasia (in true empiricist style) arises
from some past perception, not all phantasia is or involves (veridical, present-tense) perceiving—
as the cases of dreaming, day-dreaming, delusion, illusion, visualization, expectation, and indeed
memory all attest.
We might even suggest that perception, visualization, delusion, memory, etc. all relate to
phantasia as species to genus. Aristotle himself never makes this suggestion, and there are good
and obvious reasons why not. For a start, phantasia relates to perception as effect to cause and
as “shadow” to reality; again, the suggestion is rather too close for comfort to a possible joke-
analysis of seeing as the special case of visual delusion where the object is, well, not delusory.
The suggestion should not be taken too seriously—despite the apparent encouragement of dA
428a13—but it does perhaps shed some light. For phantasia as Aristotle presents it seems to be
the same kind of mental activity—in some good sense of “same”—whether it is online or offline,
whether it accompanies veridical perception, non-veridical dreaming, illusion, day-dreaming,
and visualizing, or indirectly veridical memory.
Actually, Aristotle says that mnêmê is a species not of the genus phantasia but of two genera,
perception and judgment (hypolêpsis). Relative to both, mnêmê is the species that we get by
adding the differentia of a time-lapse (dM 449b25); cf. my remark at the start of Chapter 30,
“Plato,” that (veridical) “memory is just the temporal dimension of knowledge.” Thus Aristotle
introduces an important duplication: there is the mnêmê that is a kind of judgment, and there is
the mnêmê that is a kind of perception. (Here, then, is a third bar to treating mnêmê as a species of
the genus phantasia.) Now dM 450a21 tells us that “all remembering involves being conscious as
well that one previously saw or heard or learnt; that is why you can’t remember without having
a sense of time,” a stricture that is already supposed at dM 449b24–9 to show that those animals
that have no sense of time, cannot have mnêmê either. (Contrast dM 453a7–14 for the claim that
animals that cannot reason (syllogizein) cannot recollect, because recollecting is a kind of reason-
ing.) But surely, all it actually shows is that animals that have no articulate and discursive sense
of time cannot have the kind of mnêmê that is a species of judgment. It doesn’t show that they
400
Aristotle
cannot have the kind of mnêmê that is a species of perception, and with it, perhaps, an inarticulate
and non-discursive sense of time (more about memory and time-discrimination in a moment).
So mnêmê, like perception and judgment and unlike visualization or daydreaming or suppos-
ing, is truth-apt: it can be right or wrong. And unlike delusion or illusion or false belief, it is not
essentially mistaken: it aims at the truth, and, as Aristotle would put it, “is such by nature as”
(pephuke) to arrive at the truth.
The largest remaining question about our two species of mnêmê, then, is the question how
they relate, causally and logically, to the judgment and the perception from which they arise.
This is the question to which Aristotle turns in dM 450b11–451a14—a remarkable piece of
analysis which deserves our close attention.
But if this [sc. the wax tablet model of memory] is how things happen in the case of
memory, then is it this [mental] affection [sc. the phantasma] that the person remem-
bers, or is it that [worldly] object from which the [phantasma] derives?
(dM 450b11–13)6
The question sets a strikingly modern dilemma about thought and its objects. Aristotle goes on:
If [what we remember] is this [sc. the phantasma], then we would never remember
anything that was not present to us. But if [what we remember] is that [sc. the object],
then how can it be that we remember what is absent and what we do not perceive, by
perceiving the phantasma[, which is present]?
(dM 450b13–16)
Aristotle is so struck by this problem that he draws out its puzzles a little further:
If there is in us something like an impression (typos) or an inscription (graphê), why

would perception of this very thing be memory of some other thing, rather than of it
[the impression]? For the person who is activating his memory directs his attention to
[theôrei] this affection—the impression; it is of this that he has perception. So how does
he [thereby] remember what is not even present? That would mean that we can hear
and see what is not present!
(dM 450b16–20)
Some recent commentators, Myles Burnyeat and John McDowell for instance, have often been
keen to rescue Plato and Aristotle from anachronistic readings of them as eighteenth-century
empiricists; the usual form of this rescue has been to read them instead as twentieth-century
Wittgensteinians. This is not the place to ask whether such readings work in general. But it is
the place to point out that here, at any rate, we have clear evidence that Aristotle does not sign
up to one key Wittgensteinian objection to “the way of ideas,” namely that it is simply mis-
placed to describe our awareness of inner images as perception.7 For in this passage Aristotle does
talk explicitly of the perception, aisthêsis, of “internal” phantasmata, and apparently takes it to be
no less genuinely perception than “external” hearing and seeing are.
Anyway Aristotle now has his dilemma about memory. Memory is either (a) of the proximal
impression, or (b) of the distal object. If (a), then memory is not open to the objection that
“you can’t perceive what isn’t there,” but does not seem to be of the distal object; if (b), then
memory is of the distal object, but seems wide open to the objection that “you can’t perceive
what isn’t there.”
401
To resolve this dilemma he says this:
Isn’t it possible for just this to happen [sc. that by remembering we perceive what is not
present]? It is like a drawing on a panel, which is both an animal and a picture [eikôn]
[of an animal]: it is the animal, and it is the picture—it is both. And yet the essence of
animal and picture is not the same; it is possible to attend to [the drawing] both as an
animal, and as a picture. Just likewise we should understand the phantasma within us.
It can be an object of attention both as a thing in its own right, and also as the image
of something else.
(dM 450b19–26)
In effect Aristotle tells us here that the key to a correct philosophical theory of memory is the
notion that we now call reference. Aristotle has no specific word for the notion, but he indicates
it fairly clearly by his use of hôs (“as,” “qua,” “under the aspect of”), and of simple Greek geni-
tives; as if we in English were to use italics to draw attention to the prepositions of and about.
Thus a phantasma, a mental image, has an existence in its own right, as a presence in our
subjective experience. But it also has an existence as the image of something. So in one sense
the objects of memory are phantasmata—and since they are there in our minds at the time of
remembering, memory is indeed of things that are present. Yet in another sense the objects of
memory are the real-worldly things of which these phantasmata are images—and since they are
not there in our minds at the time of remembering, there is also a sense in which memory is of
things that are absent.
Insofar as Aristotle’s theory of memory is successful, this subtle and ingenious distinction is
the key to its success. The distinction is crucial, too, to a proper understanding of Aristotle’s
wider doctrine that “there can be no thinking (noein) at all without phantasmata” (dM 450a1).
Aristotle’s idea is that thinking in general is in one sense a thoroughly physical and natural
process; for all it involves is the manipulation of mental images. Yet from the point of view of
what Aristotle often calls a “logical” understanding, the crucial thing is not so much the images
as the manipulation.
Compare an abacus, which from the “physical” point of view is no more than an arrange-
ment of moveable beads on immoveable bars. Yet from the “logical” point of view an abacus
is, also, a representation of a possible infinity of abstract arithmetical relations. What ties
together the two points of view—making it possible at once for mere arrangements of beads
to ascend to the lofty role of symbolizing abstract mathematical relations, and for abstract
mathematical relations to become incarnate in mere arrangements of beads—is, once more,
the concept of reference.
Here, as any modern philosopher will see, there arise difficult questions about how to under-
stand the reference relation. Apparently a phantasma P cannot refer to an original O unless O is
the cause of P, but neither can P refer to O unless P resembles O. Reference seems to involve both
causal connection and resemblance, and in the dM Aristotle uses language that implies both: he
speaks of imprinting as a kinesis, of having-been-imprinted as a pathos, but also of the resulting
impression as an eikôn. He does not answer the question how exactly these two factors interact.
That is a question that we can leave to modern philosophers working in the tradition of the phi-
losophy of reference that goes back to the writings of Frege, Russell, and Kripke. But Aristotle
certainly deserves some credit for making it a question that can at least be stated.
What Aristotle apparently does try to do—this is a speculation, but I hope a plausible one—
to make sense of memory, and in particular of the reference-relation that is involved in memory,
is to make something of his notion, already referred to, that memory and recollection both
402
Aristotle
involve a judgment or perception of time; he comes back to this at dM 452b8–453a14. If I read

him right, the idea is this. Aristotle thinks that we have, in general, a capacity for making dis-
criminations of greater and lesser in time-extent, just as we can make these discriminations with
other kinds of extent (452b8–22); and he thinks that it is a necessary part of remembering and
of recollecting that it should include such a time-discrimination (452b23–453a4). His suggestion
is, then, that the reference to something distal that is necessarily part of memory and recollec-
tion is to be understood by adding a second-order judgment about time to stand alongside the
first-order judgment or perception that memory or recollection involves: e.g., “the cat was on
the mat” (first-order) is to be accompanied by “My perception that ‘the cat was on the mat’ was
at t” (second-order).
This suggestion does not work as an explanation of the reference part of any memory-judgment,
for the obvious reason that it is regressive: the explanation of the reference-relation, as it
occurs within a judgment or proposition, cannot take the form of adding a further judgment or
proposition within which that reference-relation also occurs. If we waive that problem, the sug-
gestion may look more hopeful as an explanation of the phenomenon of memory itself. There
is undoubtedly a sense of memory in which you don’t remember that p unless you remember
when p. However, as I remarked above with regard to the case of animal memory, this sense
refers to a particularly sophisticated and articulate form of memory. It is too strong to say that
there cannot be memory at all without a clear and explicit judgment about when the content
of the memory occurred.
Aristotle turns from (a) memory to (b) recollection at dM 451a18.
4. De Memoria: (b) On recollection (451a18–452b7)

As already suggested, the distinction between memory and recollection that Aristotle wants to
make, in contrast to Plato, is simply a distinction between retention and recall: between hold-
ing something in your mind, which is what mnêmê does, and bringing something back to mind,
which is the role of anamnêsis.
What kind of “something”? Retention and recall of what? Aristotle begins his discussion
of recollection by stressing that recollection is not recollection of memories (dM 451a21 ff.).
Recollection and memory are both of the same two things, namely of knowledge and perception
(dM 451b4; cf. my remarks above about dM 449b25 on judgment and perception). But how can
knowledge and perception be available for memory or recollection? Aristotle’s answer is clearly
“by way of phantasmata,” but it takes him until 453a16 to make this fully explicit: “recollection
is a searching, in this sort of physical medium, for a phantasma.” So in memory, knowledge or
perception is retained in us by the retention of phantasmata; and in recollection, knowledge
or perception is recalled in us by the recall of phantasmata. When we retain knowledge or per-
ception in memory, it is phantasmata that we store; when we recall knowledge or perception
in recollection, it is phantasmata that we recall. Given Aristotle’s wider doctrine (see dM 450a1,
dA 427b15, dA 3.2–3 passim) that phantasia is a necessary condition of any thought at all, the
central place of phantasmata in his account of the nature of thought seems undeniable. It also
seems strikingly parallel—however much neo-Wittgensteinian readers of Aristotle may dislike
the comparison—to John Locke’s views at the opening of Book Two of the Essay.
As I say, in de Memoria Aristotle does not make it explicit until 453a16 that the field upon
which recollection (and memory) operates is a field of phantasmata; in 451b12–451a15 he talks
mostly about recollection as involving kinêseis, “processes” or “motions” [sc. of the mind].8
This restriction and unclarity in his vocabulary does not hinder him from saying some things
that mark even more clearly the parallel between his phantasia-based theory of thought and
403
“the way of ideas” that was offered, in varying forms, by Enlightenment empiricists like
Locke and Hume. What is also very evident is how very far the theory of recollection that
Aristotle presents, in this second half of the dM, is from Plato’s “theory of recollection.”
So at 451b10–14, Aristotle begins his explanation of how recollection, in his sense, happens
with the following:
Recollections happen when this motion is such by nature (pephuken) as to come about
after that one. Now if this [sequence] is by necessity, then clearly when the first
happens, the second will [inevitably] happen too; but if it is not by necessity but by
habituation, the second will happen for the most part.
(dM 451b10–14)
Just as there are two kinds of sequence in nature, the necessary and the “for the most part”
(cf. Physics 2.1), so correspondingly there are two kinds of sequences from any mental motion
A to any other mental motion B: the necessary, and the habitual. In the first kind of sequence,
the occurrence of A necessitates the occurrence of B; in the second, A’s occurrence just makes
B’s occurrence likely. Otherwise put: the connections between ideas can be necessary, or they
can be customary. (They can also, as dM 452b1–5 tells us, be accidental or random: Aristotle’s
general epistemological optimism leads him to think that the processes of recollection are usu-
ally reliable, but he does not think there is anything infallible about them. The treatise closes,
453a15–b7, with some reflections on the way bodily factors can prevent or distort accurate
recollecting.)
Aristotle continues:
It happens that some motions become more “habituated” just by one occurrence, than
others do even though they are moved many times; that is why we remember some
things that we have seen only once better than some other things that we see again
and again.
(dM 451b14–16)
Or as we might put this: some ideas strike us with force and vivacity, others much less so
(cf. 452a27–31).
Aristotle’s next remarks are:
Thus when we recollect, we move some one of the motions from before [the one
that we are seeking], [and set a sequence going from that motion], until we move the
motion immediately after which there habitually follows the one [that we are trying to get
to]. This is why we hunt in order, starting our thoughts from the present or from some
other motion, and from the similar or the opposite or the contiguous.
(dM 451b17–20)
So recollection, we learn here, can be a voluntary process—it can be a “hunting” in our

thoughts. This mental hunting is possible because ideas are connected in the ways described
above, viz. necessity and customary connection (and also by the principles of association of
similarity, opposition, and contiguity). So we can move through structures of ideas, towards the
particular idea that we are seeking. (Sorabji interestingly takes Aristotle’s rather opaque remarks
upon this subject to constitute a theory of mnemonics, and certainly mnemonics are not far
away at this point in the dM.)
404
Aristotle
So what Aristotle is presenting here is, in something pretty close to the classic empiricist
sense, a theory of ideas. My thesis is not, of course, that Aristotle’s theory of ideas is just the
same theory as Locke’s, or Hume’s, or Berkeley’s. Patently it isn’t precisely the same as any of
those theories. Notice in particular that Aristotle’s theory of ideas, unlike the Enlightenment
empiricists’, is studiedly neutral between what the Middle Ages were to call “nominalism” and
“realism”—it does not entail Aristotle’s own moderate realism about universals, but it does not
exclude it either. (The effective inclusion in Aristotle’s theory of a serious attempt to make
sense of reference, something that is strikingly missing in the Enlightenment empiricists, is
important here.) Notice also how Aristotle’s characteristic epistemological optimism comes
out again in the straightforward mirroring of the distinction between necessary and customary
connections in the world by a distinction between necessary and customary connections among
our ideas. (Cf. NE 1139b7–11 for the general idea here, that the part of us that discerns any
given aspect of the world will resemble that aspect—itself an application of the commonplace
of ancient philosophy that perception of likes is by likes.) But the affinities are clear. And if
this proposal evokes the objection that it is anachronistic to suppose that Aristotle could have
had a classic empiricist theory of ideas, the answer to that objection is clear too. Of course that
supposition is anachronistic; but there is nothing anachronistic in supposing that the classical
empiricists had read Aristotle’s de Memoria. On the contrary, nothing seems more probable.
What is also notable about Aristotle’s theory of thought and memory is, as I said above,
how studiedly anti-Platonic it is. This comes out in two ways in particular in his discussion
of recollection.
Here is the first way it comes out. Meno 80d–e famously proposed “Meno’s paradox,” a
dilemma according to which it is impossible to learn anything because it is either futile or
impossible to “inquire” or “search” for anything one might learn. For either one knows it
already, in which case inquiry is futile; or one does not know it already, in which case one can-
not know where to search for it, and cannot know when one has found it either, so that inquiry,
and hence learning, turns out to be impossible.
Plato’s answer to this puzzle is Plato’s theory of recollection, which is supposed, as we have
seen, to be a theory that covers all knowledge. By contrast, Aristotle’s full answer to Meno’s
paradox is (I speculate9) going to go something like this. First—though not in the de Memoria—
Aristotle might point out that the paradox cannot cover all knowledge: mundane examples like
looking for and finding (the location of) a lost sock, or the road to Larisa, show how little bite
the paradox has if we take it that broadly. However, second, the paradox might have some bite if
we were talking about trying to recollect things that we have forgotten; ironically for Plato, who
sees recollection in his sense as the answer to Meno’s paradox, the mental processes involved in
attempting to recollect things do look like a case where something like Meno’s paradox might
be applicable. But, third, Aristotle goes on to show that, pace Meno, even here “inquiry” and
“search” for ideas are perfectly possible; that is what Aristotle aims to show by way of his theory
of ideas, as expounded above. As he sums up this anti-Platonic conclusion:
Many times before now someone has been unable to recollect, but when he seeks he
can recollect, and he finds what he is seeking. This happens when he sets many [phan-
tasmata] in motion, until he sets in motion the right kind of process upon which the
[thing sought] will follow.
(dM 452a8–10)
This brings us to the other way in which Aristotle’s discussion of recollection is studiedly anti-
Platonic. This is that, for Plato, recollection is essentially a matter of getting hold of a priori
405
ideas: in the Meno, it means grasping the connections between geometrical ideas; in Phaedo
74a–75b, it means grasping the Forms. For Aristotle, by contrast, recollection can be the recall
of any sort of ideas at all, provided they exhibit some kind of order or connection. Certainly
Aristotle recognizes, as already noted, a distinction between those ideas that are necessarily and
those that are only customarily connected; certainly Aristotle agrees that strings of ideas are
easier to recollect when, like mathematical ideas, they have some sort of inherent order to them
(taxin tina, 452a3). He goes into considerable detail, too, about what particular mnemonic tech-
niques we might try to deploy, when we are consciously trying to recollect something that we
used to know (dM 452a10–26); the key thing, he says, is to secure a “starting-point” (452a12).
The point remains that Aristotle’s theory of recollection is far more general than Plato’s, simply
because Aristotle’s account is entirely topic-neutral. Aristotle’s account of recollection can apply
to the recollection of anything.
“So” (someone might say here) “Aristotle does not exclude, even though he does not focus
on, the possibility of a reasoned recovery, from within, of a priori rational connections like
those that the slave-boy apparently recovers in the Meno.” True. But there is a difference
between seeing such connections and recollecting them. Aristotle thinks that Plato has no reason
to claim that the slave-boy recollects them rather than just sees them, for the first time; and
since (as Aristotle began by saying) memory—and recollection—have to be of the past, our
first encounter with the truths of geometry, which are not past but timeless, will have to be a
matter of seeing, not of remembering or recollecting. Aristotle would also, I suspect, be sym-
pathetic to an objection to the slave-boy example that is familiar from generations of students:
namely that genuine recollection has to be genuinely ex autou, genuinely “from within the
thinker himself ”—and the slave-boy’s isn’t, it cannot happen at all without the prompting of
Socrates’ questions.
What Aristotle offers us, then, on the subjects of memory and recollection, is rich, ingen-
ious, and of extraordinary philosophical power and resonance. He is perhaps the first truly great
philosopher of memory in the Western tradition. Thanks to the labors of Sorabji and others,
the de Memoria has attracted some attention in the last forty years or so; but still, I might suggest,
nowhere near as much as it deserves.
Notes
1 Chappell (2014: Ch. 11).
2 “Neglected” is a relative judgment, but still—compared e.g., with the de Anima—an applicable one.
The main stimulus to modern interest in de Memoria is Sorabji (1972, 2nd edn. 2006). Bloch (2007) is
a significant and valuable study, with a fuller bibliography of recent work on dM than I have space to
provide here.
3 For the charge of unsustainable recklessness, see Burnyeat (1992).
4 Annas argues that the distinction that Aristotle has in mind is one between “non-personal” and
“personal” memory, i.e., between memories of the form “I remember that Caesar invaded Britain”
and of the form “I remember Caesar invading Britain.” Annas herself describes her hypothesis as “a
bold one” (1986/1992: 300), which seems to be putting it mildly. This essay will argue that the de
Memoria is concerned, rather, with a retention/recall distinction.
5 For useful studies of phantasia, see Frede (1992) and Schofield (1992). Schofield has much valuable
material on how to translate phantasia, which he describes as a “familial [i.e., family-resemblance?] con-
cept” (1992: 277); Schofield is on the whole rather more resistant than I am, or e.g., Sorabji (2007) is, to
seeing Aristotle as a proto-classical-empiricist.
6 All translations are my own.
7 As in the old (perhaps apocryphal) Oxford examination question: “Is perception ‘seeing’ things or seeing
‘things’?”
406
Aristotle
8 Aristotle is not the only philosopher to talk of ideas, or something very like them, as motions of the mind:
“When conclusions are too complex to be held in a single act of intuition, their certainty depends on
memory; and since memory is perishable and weak, it must be revived and strengthened by continuous
and repeated movement of thought” (Descartes, Rules for the Direction of the Mind, Rule XI). Thanks to
Paul Fletcher for the comparison.
9 He never explicitly gives this answer. A more famous text where Aristotle squares up to Meno’s paradox
is Posterior Analytics 1.1.
Bibliography
Annas, Julia (1986) “Aristotle on Memory and the Self,” Oxford Studies in Ancient Philosophy 4: 99–117;
reprinted in M. C. Nussbaum and A. O. Rorty (eds.), Essays on Aristotle’s De Anima. New York:
Oxford University Press, pp. 297–311.
Aquinas, Summa Theologica. Many editions.
Bloch, David (2007) Aristotle on Memory and Recollection: Text, Translation, Interpretation, and Reception in
Western Scholasticism. Leiden: Brill, ISBN 978-90-04-16046-0.
Block, I. (1961) “The Order of Aristotle’s Psychological Writings,” American Journal of Philology 82: 50ff.
Burnyeat, M. (1992) “Is an Aristotelian Philosophy of Mind Still Credible? (A Draft),” in M. C. Nussbaum
and A. O. Rorty (eds.), Essays on Aristotle’s De Anima. New York: Oxford University Press, pp. 14–26.
Chappell, T. (2014) Knowing What to Do. Oxford: Oxford University Press.
Frede, Dorothea (1992) “The Cognitive Role of phantasia,” in M. C. Nussbaum and A. O. Rorty (eds.),
Essays on Aristotle’s De Anima. New York: Oxford University Press, pp. 279–95.
Hardie, W. F. R. (1976) “Concepts of Consciousness in Aristotle,” Mind 85: 388–411.
Lang, H. S. (1980) “On Memory: Aristotle’s Corrections of Plato,” Journal of the History of Philosophy 18:
379–93.
Nussbaum, Martha and Rorty, Amelie (eds.) (1992) Essays on Aristotle’s de Anima. New York: Oxford
University Press.
Schofield, Malcolm (1992) “Aristotle on the Imagination,” in M. C. Nussbaum and A. O. Rorty (eds.),
Essays on Aristotle’s De Anima. New York: Oxford University Press, pp. 249–77.
Sisko, John (1996) “Space, Time, and Phantasms in Aristotle, De memoria 2, 452b7–25,” Classical Quarterly
47: 167–75.
Sorabji, Richard (1972/2006) Aristotle On Memory, 1st edn., London: Duckworth, 1972; 2nd edn.,
Chicago, IL: University of Chicago Press, 2006, ISBN 0-226-76823-6.
407
32
CLASSICAL INDIAN
PHILOSOPHY
Jonardon Ganeri
1. Memory and the self

There is a deep conceptual link between time and the self. One venerable articulation of
the relation has it that memory provides a criterion of personal identity, to which a standard
objection has long been that memory presupposes sameness of self and so cannot be used in a
non-circular analysis of it. The force of the objection is acknowledged in the move by those
theorists who argue that personal identity is instead to be analyzed in terms, not of memory,
but of the artificial notion of quasi-memory (the notion of a state that is like memory in
being produced by an earlier experience, but neutral about whose experience it was). Both
psychological continuity theorists of personal identity and their opponents agree that there is
a more basic tie between time and the self, which I will call the “self-implication requirement
on memory”:
Self-implication: In memory, a current thinker not only retrieves an action or an experi-

ence but also implicates the thinker in the memory’s content as the one who performed
the action or had the experience in the past.
This requirement has clearly been acknowledged by early modern European philosophers.
Butler puts it like this:
when any one reflects upon a past action of his own, he is just as certain of the per-
son who did that action, namely himself, the person who now reflects on it, as he is
certain that the action was at all done.
(Analogy 1736/1819: 295)
Reid only echoes Butler when he says:
I remember that twenty years ago I conversed with such a person; I remember several
things that passed in that conversation; my memory testifies not only that this was
done, but that it was done by me who now remembers it.
(Essays 1969: 341)
408
Classical Indian philosophy
And Locke commits himself to the requirement when he says that memory is “a Power . . .
[of the mind] to revive Perceptions, which it has once had, with this additional Perception
annexed to them, that it has had them before” (Essay 1795: II.x.2).
For a more contemporary formulation of the requirement, let me turn to Endel Tulving’s
work on the psychology of memory. Based on studies of victims of various sorts of memory
impairment, Tulving claims that memory is underwritten by three sorts of neurally realized
memory system: procedural, semantic, and episodic. Procedural memory enables the acquisi-
tion and retention of motor, perceptual, and cognitive skills (for example, knowing how to ride
a bicycle). Semantic memory consists in the storage of generic context-free information (that
oranges are edible; that London is the capital of Britain). Episodic memory records the “what,
where, and when” of an event: remembering walking in the park yesterday morning, remem-
bering seeing a nice pair of shoes in the shop window. Each of these memory systems supports
a relation between time and self. Procedural memory enables the preservation of individual
cognitive and behavioral skills over time. Semantic memory enables both the preservation of
character traits (such as knowing oneself to be stubborn), and also factual self-knowledge (such
as knowing one’s date of birth). Episodic memory, Tulving contends however, is unique in
satisfying the self-implication requirement, a requirement which he unpacks with the aid of
the metaphor of mental time travel. Mental time travel is the reliving or re-experiencing of an
experience one has had before, a relocating of oneself in subjective time to the past (or equally
to the future). Tulving claims that the ability to travel mentally in time is enabled by a distinct
kind of cognitive capacity, a capacity which he terms “autonoetic consciousness.” Autonoetic
consciousness is a capacity to be aware of oneself in one’s own personal past or future: it is the
name given to “the kind of consciousness that mediates an individual’s awareness of his or her
existence and identity in subjective time extending from the personal past through the present
to the personal future” (1985: 1). In its role in episodic memory, Tulving has described it as
a capacity to revisit earlier experience, “a unique awareness of re-experiencing here and now
something that happened before, at another time and in another place” (1993: 68). Tulving also
describes it as a capacity for representation, one ‘‘that allows adult humans to mentally represent
and to become aware of their protracted existence across subjective time” (Wheeler et al. 1997: 335).
We can thus reformulate the self-implication requirement as follows:
Mental time travel: In episodic memory, a current thinker not only retrieves an experi-
ence but also relives it by mentally traveling back into the past.
When Tulving says that “episodic memory differs from other kinds of memory in that its opera-
tions require a self. It is the self that engages in the mental activity that is referred to as mental
time travel: there can be no travel without a traveler” (2005: 14), I take this to mean that one
is oneself represented in the content of one’s episodic memory, and represented as the one
performing the action or having the experience. It is a familiar if controversial contention that
one cannot represent oneself in either of these ways and yet be in error as to the identity of
the person represented, that there is immunity to error due to misrecognition with respect to
the first person pronoun in statements expressing states of episodic memory (Shoemaker 1970;
Pryor 1999; Bermúdez 2012; Jordi 2014).
The source of early Buddhist discomfort with the idea of episodic memory derives from
the fear that episodic memory requires conceptual self-representation. For if episodic memory
entails conceptual self-representation, then it would seem to run counter to the essence of
the commitment to the denial of self, which is that the representation of self within experi-
ence is always a delusion. Non-Buddhist Indian philosophers use precisely this point in their
409
Jonardon Ganeri
critique of Buddhist theory. Indeed, it is a critic of the early Buddhist analysis, the M m saka
philosopher abara, living a century prior to Buddhaghosa, who comes closest to articulating
the core idea of episodic memory. He introduces a concept of “recognition” (pratyabhijñā), not
as it is normally understood to involve the identification of some perceptually presented object
with object-information retrieved from memory (and so as a matter of noetic consciousness
and semantic memory, or labeling), but rather as having to do with one’s endogenous ability to
recognize oneself in one’s recollection of past events. While ordinary object recognition can be
explained by the Buddhist theory of a continuum of memory traces, the recognition of oneself
in memory cannot be:
When a thing has been seen on one day, the notion “I saw this” arises on a later
day. And this experience refers to an inner self (pratyag-ātman), nothing else [that is,
certainly, not to a collection of skandhas]. For the present [collection of skandhas] is
different from the one that saw [the thing] earlier. Hence, there is something besides
the [collection of skandhas] to which this word ‘I’ applies . . . [Here,] we do not con-
sider the word ‘I’ that we employ to be the means for inferring another thing [besides
the skandhas]. Rather, we consider recognition (pratyabhijñā), which goes beyond the
word (sabdād vyatiriktam) [to be the means of inference]. For this is the meaning we
comprehend: “We perceived yesterday, we remember now.” Therefore, we under-
stand that we existed yesterday and we exist today. And those things which existed
yesterday and exist today, they are not such as have passed away.
(M m s -s tra-bh ya; trans. Taber 1990: 37)
Thus recognition is here a type of memory in which one remembers oneself as perceiving
some past occurrence; it is episodic memory. As Taber puts it, the force of the argument is that
“memory establishes the existence of a continuous self not insofar as it presupposes a single sub-
ject of experience that both had the remembered experience and now remembers it, but insofar
as it directly reveals one” (1990: 37). That is to say, mental time travel requires a self because
autonoetic consciousness consists in the recognition of the self in episodic memory. The sense of
personal identity across time is analyzed in terms of a recognitional capacity, the capacity to re-
identify oneself in temporally subjective experience. abara here presents in effect a substitution
instance of the self-implication requirement on memory, according to which:
Recognition: In a memory expressible as “I saw it,” a current thinker not only retrieves
a seeing but also recognizes the thinker as the one who did the seeing in the past.
An attractive feature of the analysis is that it accounts for the phenomenon of the apparent age-
lessness of the self as presented in subjective experience (Klein 2012: 503): this is just what the
claim that the “inner self” is permanent (nitya) now comes to mean, that the experiential self is
as if immortal. Taber again:
If there is such a thing as self-recognition which identifies the self as existing now as
strictly identical with a self that existed before, then we have reason to believe that a
person [self] is not a body or mind (aggregate of psychological states), since those are
things which undergo change over time. A body or mind existing now will not be
strictly identical with any body or mind existing before now. Rather, a person [self] must
be some other, distinct kind of thing that persists essentially without change over time.
(Taber 1990: 51)
410
There are further developments of a recognition-based analysis of episodic memory in the

tenth-century aiva philosopher Utpaladeva (see Ratié 2017).
One might wonder, though, whether re-cognizing oneself in experience is indeed the same
as re-living that experience. Another critic of early Buddhist theory, the Ny ya philosopher
V tsy yana, also distinguishes mere memory traces (vāsana) from episodic memory, and also
proposes a version of the argument that episodic memory requires a self; but in his case he does
not make the argument rest on a recognitional analysis of autonoesis. He says:
A memory which occurs with reference to an object not presently apprehended,

in the form “I experienced that thing”—the content of this memory is an object
previous experienced, as qualified by an experiencing subject and an experience
(jñātṛjñānaviśiṣṭaḥ pūrvajñāto ‘arthaḥ viṣayaḥ).
(Ny yabh ya ad. NS 3.1.14; trans. Taber 1990: 41)
Memory is not merely the reactivation of stored information about the remembered object
(semantic memory), but an awareness in whose content the object, the previous experience, and
the experiencing subject, all feature. What V tsy yana here proposes is an account of episodic
memory based not on the postulation of a capacity for self-recognition but on the idea that the
self appears as an ingredient in the content of awareness. On the one hand, this new proposal
has the advantage that it is less vulnerable to objections based on “immunity to error through
misidentification” considerations that threaten a recognition-based analysis, the leading idea of
which is that since it is impossible for me to misidentify myself in episodic memory, nothing like
the fallible activity of “recognition” can be involved (Shoemaker 1970). On the other hand, it
is harder now to explain what ensures that the I which allegedly features in the content of an
episodic memory is the same as the I whose memory it is.
Many philosophers who comment on the self-implication requirement on episodic memory
agree that it has metaphysical implications. Reid, in the continuation of the passage quoted at
the beginning of this chapter, carries on: “but that it was done by me who now remembers it. If
it was done by me, I must have existed at the time, and continued to exist from that time to the
present.” abara likewise continues, “For this is the meaning we comprehend: ‘We perceived
yesterday, we remember now.’ Therefore, we understand that we existed yesterday and we
exist today. And those things which existed yesterday and exist today, they are not such as have
passed away.” Tulving suggests that
until such time that we have better ways of explaining the phenomenal existences of
things such as pain, smell, and recollection of the past, we need an agent such as self for
the sake of the completeness of the story. Eventually, self may turn out to be like phlo-
giston or aether—a convenient temporary prop. But the problem today is that the story
of the mind is incomplete and awkward to tell if a concept like ‘self’ is omitted from it.
(Tulving 2005: 15)
The ambition of Buddhist philosophers in India is to provide just such an account of episodic
memory, as we will now see.
2. Episodic memory as attention

It was perhaps Nyanaponika Thera who first drew attention to the curious omission of mem-
ory in the canonical Therav da classifications of the items of mind such as the Dhammasaṅgaṇi
(Nyanaponika 1949). Padmanabh Jaini later comments that “despite the extraordinary preoccupation
411
Jonardon Ganeri
of the ancient Buddhists in explaining the process of cognition, memory is conspicuous by

its absence in the long list of mental events and concomitant mental factors” (Jaini 1992: 47),
while Sarathchandra notes that “the Abhidharma does not make any attempt to explain the
phenomenon of memory, for, having postulated process without substance, they were with-
out any known equipment for explaining it” (Sarathchandra 1958: 93). Nyanaponika suggests
that the function of memory is implicit in the role of labels insofar as they introduce the rec-
ognition of an object as the same as one seen before on the basis of the mark attached; others
have related memory to mindfulness through the etymology of the term. None, however, has
noticed Buddhaghosa’s extraordinarily perceptive discussion of the topic.
A remarkable passage in the Dīgha Nikāya does, though, treat episodic memory. A per-
son who travels to another village on returning has an episodic memory in the form of the
thought, “I came from my own village to that other one where I stood, sat, spoke or remained
silent,” and in so doing, he “in absorbed attention (samāhite) . . . directs (abhinīharati) and turns
(abhininnāmeti) his thought to his past cognitions” (D i. 81; trans. Walshe 1995). There is a clear
statement here of the self-implication requirement on episodic memory. And there is a very
interesting suggestion: that episodic memory is a kind of attention.
This idea is taken up and developed by the fifth-century philosopher Buddhaghosa. In mem-
ory I attend to what I saw before. Since contemporary scholarship has overlooked it, I will
quote his remarkable discussion of memory in full:
When [entering] purely at the mind-door, there is no stimulating of a sense-faculty’s

sensitivity. Those kinds of objects normally come into focus as seen, as heard, as smelt,
as tasted, or as touched. How? Here someone circumambulates the Great Shrine [in
Anur dhapura], which is all white-washed, painted with yellow and red ochre, etc.,
hung with many kinds of flags and banners, festooned with garlands of flowers, sur-
rounded with clusters of lamps, gleaming with the loveliest radiance and decked with
ornaments, and he pays homage with the fivefold prostration at the sixteen footstools,
and looking with reverential salutation with hands upraised palms together, he stands full
of rapt happiness with the Buddha as his object. Then after he has seen the Shrine thus
and produced happiness with the Buddha as object, at another time, wherever he goes,
while he is sitting in his day quarters or his night quarters, as soon as he adverts to it, it is
as if the Great Shrine decked with ornaments came into focus in [lit. came to the avenue
of] the eye door, and it seems like the time when he circumambulated the Shrine and
paid homage to it. This firstly is how a visible datum as object comes into focus as seen.
And when he has heard the sound of a Dhamma preacher preaching the Dhamma
with a sweet voice or the sound of a reciter vocally reciting, then at another time,
wherever he is sitting, as soon as he adverts to it[,] it is as if (viya) the Dhamma preach-
ing or the recitation came into focus in the ear door, as if at the time when he was
gladly applauding. This is how a sound as object (saddārammaṇa) comes into focus
as heard. And when he has obtained a sweet-smelling scent or flower and makes an
offering of it on a [flower] altar or at a shrine with consciousness whose object is the
odour, then at another time, wherever he is sitting, as soon as he adverts to it[,] it is
as if the odour as object came into focus in the nose door, and it seems like the time
when he was making the offering. This is how an odour as object comes into focus as
smelt. And when he has made use of special food of the various kinds, sharing it out
with his companions in the Life Divine, then at another time, wherever he gets a meal
of beans, as soon as he adverts “At that time I made use of special food of the various
kinds sharing out with my companions in the Life Divine” it is as if that flavour as
412
object comes into focus as tasted. This is how a flavour as object comes into focus as
tasted. And when he makes use of a bed or chair or a coverlet or robe that has a pleas-
ant touch, then at another time wherever he makes his bed in discomfort, as soon as he
adverts “At that time I used a soft bed and chair” it is as if that tangible datum as object
(phoṭṭhabhārārammaṇa) comes into focus as touched. So, when [entering] purely at the
mind-door, there is no stimulating of a sense-faculty’s sensitivity. These objects only
come into focus as normally seen, as heard, as smelt, as tasted, or as touched.
(Buddhaghosa, Atthas lin 72–3; trans. Bikkhu Ña amoli, modified)
Episodic memory is attention to objects from past perceptual encounters. Notice in this passage
how careful Buddhaghosa is to formulate the idea, without any reference to images or repre-
sentations, and without mention of subjects. On the other hand, he has also said that in such
circumstances the perceptual process is not engaged (there is no striking of a stimulus on the
eye, etc.). The percept is directly available to working memory because it has been perceived
before; it literally re-presents itself. In episodic memory it is as if (viya) one is perceiving again;
one relives the past experience. He describes memory as responsive to cueing: in each of his
examples some current perceptual cue triggers memory.
The work by Chun and his colleagues on memory and attention is helpful in understanding
Buddhaghosa’s idea. Chun and Johnson (2011) too argue that memory is an act of attention:
“our ability to bring past experiences to mind represents a form of internally oriented reflec-
tive attention” (Kuhl and Chun 2014: 807). Working with an encoding/retrieval distinction,
they argue that “memory retrieval is an act of selective attention,” which they call “reflective
attention,” while encoding requires a different sort of attention, externally oriented “perceptual
attention” (Kuhl and Chun 2014: 815). For Buddhaghosa, the role of selection is performed by
perceptual cueing, something in one’s present perceptual environment prompting one to select
among many potential past experiences. Chun’s formulation of the distinction between these
two kinds of attention is representationalist; thus
cognition can broadly be divided into perceptual processes, initiated by and/or directed
at external sensory information from the environment, and reflective processes, initiated
by and/or directed at internal mental representations. Perceptual processes operate on
“incoming” external stimuli (e.g., reading text, listening to a song). Reflective processes
are directed at internal representations, such as thoughts, memories, imagery, decision
options, or features of problems. That is, reflective processes operate on representations
in the absence of current external input.
(Chun et al. 2011: 520)
They emphasize, though, that “component processes of reflection are the cognitive elements
of what is often referred to as controlled/executive processing or working memory” (Chun
et al. 2011: 521), and distinguish here components of refreshing, rehearsing, reactivating, noting,
and initiating, a list that stands in partial correspondence with Buddhaghosa’s list of activities in
working memory (javana): “preliminary work” (parikamma), “approach” (upacāra), and “order-
ing” (anuloma) (Path 559 [xxi.129]). We can thus discriminate more finely between varieties of
reflective attention. Mindful attention (sati) is reflective attention in which the component pro-
cess of rehearsing (Buddhaghosa’s “ordering”) is the dominant aspect. Episodic memory, of the
sort here being considered, is reflective attention in which reactivating is the dominant aspect
(Buddhaghosa’s “approach”); indeed, for Buddhaghosa, such memory may be said to consist in
the reactivation of a past perception.
413
Jonardon Ganeri
Buddhaghosa’s careful reconstruction of a case-study involving first the circumnambulation

of a stūpa, with all the attendant perceptions in various modalities, and then the reliving of that
event when the person is in a suitable cueing environment, seems to me, however, better to fit
the paradigm of simulation in grounded cognition (cf. Barsalou 2008). That is to say, perhaps
the idea is that reflective attention utilizes simulation. He very carefully reconstructs in detail the
situation in which the original experience takes place and the situation where a memory is cued.
What he describes is the way memory involves a simulation of a previous experience triggered
by the situation. The idea is that
simulation is the reenactment of perceptual, motor, and introspective states acquired

during experience with the world, body, and mind. As experience occurs (e.g., easing
into a chair), the brain captures states across the modalities and integrates them with
a multimodal representation stored in memory (e.g., how a chair looks and feels, the
action of sitting, introspections of comfort and relaxation). Later when knowledge is
needed to represent a category (e.g., chair), multimodal representations captured dur-
ing experiences with its instances are reactivated to simulate how the brain represented
perception, action, and introspection associated with it.
(Barsalou 2008: 618)
What Buddhaghosa says is that the earlier modal state is as if again seen at the eye-door, and
comes into focus (“to the avenue”) because of having been seen. I take the avenue here to refer
to the avenue of what Buddhaghosa calls the “mind-door,” which gates working memory, and
the process to consist in a simulation of the original experience within working memory. Insofar
as multimodality is involved, this is the task of labeling (cf. Chun’s “noting”); but Buddhaghosa
does not put the idea in terms of representation. He seeks to develop a non-representationalist
theory of long-term storage.
The difference in phenomenology between an original experience and a relived one is the
consequence of two different ways of attending to the same event. In both cases it is transparent,
and shows up not in content but indirectly. Just as one knows what is the phenomenal char-
acter of one’s current perceptual experience in virtue of attending to the world itself, so too, if
episodic memory is attending to the past, can one know what is the phenomenal character of
one’s current recollection of a past event. So the phenomenology of time is explained by the
transparency of episodic memory, considered as a distinctive kind of attention. In the terminol-
ogy of Dokic (2014), this is a “two-tiered” account of episodic memory: episodic memory is
factored into a component consisting in a first-order memory of a past happening together with
a metacognitive “feeling.” The claim is that the second component consists in a proprietary
kind of phenomenology specific to episodic memory, rather than in a second representation, of
oneself as having had the experience in the past, and that autonoesis is a distinct kind of atten-
tion, subjective attention to the past. Dokic comments that
the episodic feeling is much more specific than the ‘feeling of pastness’, which Bertrand
Russell famously claimed sets memories apart from perceptions and imaginings. Unlike
the feeling of pastness, the episodic feeling has a metacognitive dimension; it somehow
concerns the relationship between our present memory and our past experience. The
episodic feeling is also more specific than other metacognitive feelings, such as the
feeling of familiarity . . . Only episodic memories feel episodic.
(Dokic 2014: 420)
414
Buddhaghosa’s phenomenological description of the autonoetic attention in which episodic

memory, for him, consists, is indeed richer than a mere “feeling of pastness”; the metacogni-
tive feeling involved in autonoetic attention is a sense of revisiting, an as-if-ness, in which the
remembered event is as if happening before one’s very eyes once again.
References
Barsalou, Lawrence (2008). “Grounded Cognition,” Annual Review of Psychology 59: 617–45.
Bermúdez, José Luis (2012). “Memory Judgements and Immunity to Error Through Misidentification,”
Grazer Philosophische Studien 84: 123–42.
Buddhaghosa. Fount: The Fount of Meaning (Atthasālinī). Path = The Path of Purification (Visuddhi-magga). P li
texts are available in the Cha ha Sa g yana edition (Yangon, 1956), digitized in various scripts (http://
tipitaka.org) and in roman transliteration (www.shemtaia.com/BU/cspattha.shtml). Referencing is to
pagination in the P li Text Society editions. Ñ amoli, Bhikkhu (ms.), transl. The Fount of Meaning
(Atthas lin ) by Buddhaghosa cariya (Island Hermitage Library, Dodand wa). Partially typed, partially
handwritten, ms. on Asl. 36–114.
Butler, Joseph (1736). “Of Personal Identity,” in The Analogy of Religion (London: James, John and Paul
Knapton); First Appendix (First Dissertation); reprinted in John Perry (ed.), Personal Identity (Berkeley,
CA: University of California Press, 1975), pp. 99–105.
Chun, Marvin M. and Johnson, Marcia K. (2011). “Memory: Enduring Traces of Perceptual and Reflective
Attention,” Neuron 72: 520–35.
Dokic, Jérôme (2014). “Feeling the Past: A Two-Tiered Account of Episodic Memory,” Review of
Philosophy and Psychology 5: 413–26.
Jaini, Padmanabh (1992). “Sm ti in the Abhidharma Literature and the Development of Buddhist Accounts
of Memory of the Past,” in Janet Gyatso (ed.), In the Mirror of Memory: Reflections on Mindfulness and
Remembrance in Indian and Tibetan Buddhism (Albany: State University of New York Press), pp. 47–60.
Jordi, Fernández (2014). “Memory and Immunity to Error through Misidentification,” Review of
Philosophical Psychology 5: 373–90.
Klein, Stan B. (2012). “The Self and Its Brain,” Social Cognition 30(4): 474–518.
Kuhl, Brice and Chun, Martin (2014). “Memory and Attention,” in A. Nobre and S. Kastner (eds.), The
Oxford Handbook of Attention (Oxford University Press), pp. 806–36.
Locke, John (1975). An Essay Concerning Human Understanding, ed. with an introduction by Peter H.
Nidditch (Oxford: Clarendon Press).
Nyanaponika Thera (1949). “The Omission of Memory in the Therav din List of Dhammas: On the
Nature of saññā,” in his Abhidharma Studies: Researches in Buddhist Psychology (Colombo: Frewin and Co.).
Pryor, Jim (1999). “Immunity to Error Through Misidentification,” Philosophical Topics 26: 271–304.
Ratié, Isabelle (2017). “Utpaladeva and Abhinavagupta on the Freedom of Consciousness,” in The Oxford
Handbook of Indian Philosophy (New York: Oxford University Press).
Reid, Thomas (1785/1969). Essays on the Intellectual Powers of Man (Edinburgh: John Bell, 1785; reprinted
Cambridge, MA: MIT Press, 1969).
Sarathchandra, Ediriweera R. (1958). Buddhist Psychology of Perception (Colombo: The Ceylon University
Press).
Shoemaker, Sydney (1970). “Persons and Their Pasts,” American Philosophical Quarterly 7(4): 269–85.
Taber, J. A. (1990). “The M m s Theory of Self-Recognition,” Philosophy East and West 40(1): 35–57.
Tulving, E. (1985). “Memory and Consciousness,” Canadian Psychology/Psychologie Canadienne 26(1): 1–12.
Tulving, E. (1993). “What Is Episodic Memory?” Current Directions in Psychological Science 2(3): 67–70.
Tulving, E. (2005). “Episodic Memory and Autonoesis: Uniquely Human?” in Herbert S. Terrace and
Janet Metcalfe (eds.), The Missing Link in Cognition: Origins of Self-Reflective Consciousness (Oxford:
Oxford University Press), pp. 3–56.
Walshe, Maurice (trans.). (1995). The Long Discourses of the Buddha (Boston, MA: Wisdom Books).
Wheeler, M. A., Stuss, D. T., and Tulving, E. (1997). “Toward a Theory of Episodic Memory: The
Frontal Lobes and Autonoetic Consciousness,” Psychological Bulletin 121(3): 331–5.
415
33
INDIAN BUDDHIST
PHILOSOPHY
Monima Chadha
1. Introduction
Buddhist philosophy puts forward a revisionary metaphysics. The central claim is that our
ordinary way of thinking and our ordinary conceptual scheme, in which the self (minimally
conceived of as the referent of “I”) occupies a prime position is mistaken. Instead the Buddhists
propose an intellectually and morally preferred picture of the world that lacks such a self.
Anātmavāda, or the no-self doctrine, is interpreted in various ways within the classical Indian
Buddhist tradition, and among its Hindu critics. As a matter of fact, anātmavāda (insofar as we can
speak of it in the singular) is still a matter of debate among contemporary Buddhist philosophers.
However, most modern scholars agree that Buddhist doctrine of no-self is not merely aimed
at rejecting a given theory of self in the Indian debate—that the self is an immaterial, eternal,
and (essentially?) conscious entity and they also agree that the no-self theory minimally dictates
that the referent of “I” is not a persisting entity. My discussion in this chapter will focus on the
Abhidharma view, since it is generally thought of as the best representative of Indian Buddhist
philosophy. Abhidharma tradition arose around the third century bc as a first attempt in classi-
cal India to systematize and organize the metaphysical and epistemological theses in Buddha’s
teachings in the Nikāyas which were in the form of dialogues.1 The Abhidharma philosophers,
I argue, read the no-self view as the no-subject or no-ownership thesis. This raises serious
questions about the coherence of the Buddhist view. For the purposes of this chapter, I shall
focus on an account of episodic memory and its ownership. I shall begin my discussion with
Vasubandhu’s “Refutation of a Theory of the Self,” because I take him to be an exemplar of the
Buddhist-Abhidharma revisionary metaphysics. Furthermore, in his later writings Vasubandhu
has a strategy to sidestep the Strawsonian criticism of the no-ownership view. I will focus on his
arguments against the Ny ya-Vai e ikas (Section 4 of the “Refutation of a Theory of the Self”)
as that discussion revolves around the self as the subject/owner of experiences and thoughts.2
The Naiy yiakas exhort their Buddhist opponents to address questions like: how, although we
are not selves, can we apprehend an object, or remember it?; How, without a self, can there be
an agent of action or a subject that experiences their results?; How, without a self to possess it,
can there be a mind that conceives an “I”?; And, how, without a self, is there an underlying sup-
port for desire, cognition, feelings of pleasure and pain, etc.? Vasubandhu defiantly holds ground
against his opponents and refuses to posit a self as a subject of experience. Rather, he cleverly
416
Indian Buddhist philosophy
shifts the explanatory burden away from the self/person to a mind (or minds) conceived of as
a series of causally connected mental states. The task of this chapter is to analyze Vasubandhu’s
valiant effort to defend the no-self doctrine against the Ny ya-Vai e ikas in order to bring to the
fore the Buddhist model of mind. The impetus to develop an accurate descriptive metaphysics
of the mind is, I believe, provided by the revisionary Buddhist metaphysics which denies the self
as a subject/owner of experiences, memories, and thoughts.
The discussion in Section 4 of Vasubandhu’s text starts with exploration of episodic memory.
The Ny ya-Vai e ikas argue that the no-self theory cannot explain our episodic memory expe-
riences. The issue raised by episodic memories is then generalized by the Ny ya-Vai e ikas to
demand an explanation of ownership of other mental experiences and actions and also the results
of one’s actions. I will rehearse the Ny ya argument and Vasubandhu’s response in Section 2 of
this chapter. In Section 3, I shall look at recent empirical work in psychology to suggest a new
way of thinking about episodic memories. This new way, surprisingly, hypothesizes an account
of episodic memory that can easily be adapted to a Buddhist explanation of episodic memories.
In Section 4, I show how Vasubandhu, particularly in his post-Abhidharma writings, provides a
satisfactory account of memory and as well side-steps the criticisms raised by the no-ownership
theorists.
2. Vasubandhu’s no-self doctrine

In the opening section of the “Refutation of a Theory of the Self,” Vasubandhu gives a simple
proof for the no-self theory: how do we know that the word “self” is only a designation for a
series of skandhas, and that no self exists in and of itself?
We know this because no proof establishes the existence of a self apart from the skandhas, no
proof by direct perception, nor any proof from inference. If the self were a real entity, separate
like other entities, it would be known either by direct perception as are the objects of five-sense
consciousnesses and the objects of mental consciousness, or by correct inference, as are the five
sense-organs (the translation is from Pruden 1988: 1313, slightly modified).
There is no proof for the existence of the self by direct perception, nor by inference.
Vasubandhu elaborates further that we can know objects of the five senses and the objects
of mental consciousness by perception. And we can know about the existence of the five
external sense organs on the basis of inference from the fact that even in the presence of all
other causes of perception—e.g., external objects, light, attention, etc.—the blind and the
deaf cannot perceive. Thus we infer the existence of the sense organs as a cause whose pres-
ence, together with other factors, brings about a perception. However, we cannot perceive a
self, nor are there any considerations that would lead us to infer or postulate a self; so we can
conclude that there is no self.3
The question then is whether in denying the existence of the self, the Buddhist philosopher
is denying the subject of experience as well? For this we need to look closely at Vasubandhu’s
attempt to deal with the concerns raised by the Ny ya-Vai e ikas. Uddyotakara in his com-
mentary (NyāyaVārttika) on Nyāya-sūtra 1.1.10 presents an inference to contest the simple proof
provided by Vasubandhu. His celebrated Ny ya “memory argument,” first presented as nega-
tive inference (kevala-vyatireka), is: my present desire and a certain past experience are unified
insofar as they concern the same object; I recognize that the thing I desire now is of the sort I
experienced to be a cause of pleasure in the past. Recognition requires a persisting unitary agent,
the referent of “I” since that which does not have the same agent is never recognized. For example I
can never recognize my friend’s cognitions; therefore, recognition cannot be explained without
417
Monima Chadha
postulating a persisting unitary agent, i.e., a self. The contention is that if the self does not exist
and the psychological states are momentary, there can be no recognition of an object experi-
enced in the past. Vasubandhu’s reply to this argument is that memory is generated in a series of
mental states immediately after a special kind of mental state arises. What is this special kind of
mental state? According to Vasubandhu, a mental state is the immediate cause of the memory if
it is inclined towards the object of memory in virtue of attending to a concept of the object or
an object which is similar to or associated with the object to be remembered, or simply by habit
or a prior resolution. Vasubandhu adds that the immediate cause should not be inhibited by a
psychosomatic change arising from sickness, grief, distraction, etc.
The Naiy yikas’ main complaint against this Buddhist account is that it allows that one mind
can perceive what another remembers, and that implies that the Buddhist account is wrong: it is
impossible that Jill remembers what Jack has perceived. Vasubandhu in reply says that such exam-
ples cannot be used to reject his account, since there is no appropriate connection between a mental
state in the continuum of Jack and a mental state in the continuum of Jill, since these two minds
are not related as cause and effect within a continuum. A mental state, say, that remembers Obama
winning the 2008 presidential elections, arises from another mental state that perceived that event in
2008. Memory, according to Vasubandhu, is the result of a chain of memory impressions occurring
in a series of mental states initiated by an original experience and giving rise to a memory experi-
ence in a later mental state in the same series. The Naiy yikas fail to understand how this account
is supposed to work. They ask: who is it that remembers? When we say someone remembers, they
grasp an object with the help of memory; but there isn’t anyone in Vasubandhu’s account who can
be the agent or the owner of the memory. Vasubandhu’s reply is that there is no agent grasping the
object; there is just the occurrence of memory which is fully explained by the special kind of mental
state which is the cause of the memory. Since no separate act of grasping is required, no self as an
agent is required. We say Jill remembers when we perceive a memory occurring in the continuum
of aggregates that we call “Jill.” Vasubandhu adds that this account of memory of an object should
also suffice to explain recognition of an object since a recognition arises from a memory.
This complaint about requiring an agent for the act of grasping a memory is generalized by
the Naiy yikas as: we need an agent (self) as the owner of experiences, the doer of actions, the
subject that experiences the results of karma, etc. Indeed the Naiy yikas would say that every
activity, whether it be a conscious activity as in apprehending an object or a physical activity that
of walking, requires an agent. The claim that every activity (signified by an active verb) depends
on or requires an agent (signified by a noun to which the active verb is attached) is assumed by
the Naiy yikas. The rationale for this grammatico-metaphysical claim derives from an assump-
tion that the categories of classical Sanskrit discourse mirrors categories of entities represented in
the world. Vasubandhu simply dismisses this assumption as a mere convention; he does not even
address the need for an agent of the memory, for he believes, as do all Buddhists, that experiences
do not require a bearer. As the Buddha said, “Misery doth only exists, none miserable. No doer
is there, naught save the deed the found” (Visuddhi Magga, Chapter 8, Section 15). The Buddhist
argument for this claim is phenomenological and can be nicely summarized in Hume’s words,
When I turn my reflexion on myself, I never can perceive this self without some one
or more perceptions; nor can I ever perceive any thing but the perceptions. ‘Tis the
composition of these, therefore, which forms the self . . . Suppose it to have only one
perception, as of thirst or hunger. Consider it in that situation. Do you conceive any
thing but merely that perception? Have you any notion of self or substance? If not, the
addition of other perceptions can never give you that notion.
(Hume 1739–40/1888/1978: 634)
418
Vasubandhu is perhaps right to dismiss the assumption made by the Ny ya-Vai e ikas and the
Grammarians in the classical Indian tradition. Even if we accept the Buddhist contention that
there is just remembering rather than an agent that remembers, still the question “who is owner
of a memory?” remains. Vasubandhu does not reject this question, nor does he try to defend the
“no-ownership” view (a view dismissed as internally incoherent by Peter Strawson).4 Rather
he attempts to sidestep the no-ownership view, in his response to the Naiy yikas, Vasubandhu
shows that ownership can be analyzed in terms of causal relations. It will be instructive here to
examine Ny ya objection and Vasubandhu’s response as stated in the Abhidharmakośa-bhāsya:
If a self does not exist, [they ask] who remembers? [They claim that] What is meant by
[saying that someone] “remembers” is [that an agent] “grasps an object [of perception]
with [the help of] a memory [of the object].”
But is [an agent] grasping an object [in this case] anything other than [the occurrence
of] a memory [in a continuum of consciousnesses? Surely, it is not. No separate act
of grasping is required, and consequently no self as the agent of this act is required,
to explain the occurrence of a memory of an object. If they ask] what produces the
memory [of the object if there is no self, we reply that] the producer of the memory,
as we have [already] said, is the special kind of mental state that causes a memory.
Although we say Caitra remembers, we say this because we perceive a memory that
occurs in the continuum [of aggregates] we call Caitra.
If the self does not exist, [they ask] whose is the memory? [They say that] the meaning
of the use of the possessive case [indicated by the use of “whose”] is ownership. It is
the owner of a memory in the way that Caitra owns a cow. [In their view] a cow can-
not be used for milking or for carrying anything and so on unless it is so owned, [and
in the same way a memory cannot be directed at an object unless it is so owned] . . .
[In your example] what is called “Caitra” is called the owner of the cow because we
are aware of a single continuum of a collection of [phenomena] causally conditioning
[other] phenomena [within the same continuum] and assume a causal connection [of phe-
nomena within this continuum] to the occurrence of changes of place of, and alterations
in, [the continuum of the collection of phenomena we call] a cow. But there is no one
thing called Caitra or a cow. Therefore, there is, [even in the Tirthikas example] no rela-
tion between the owner and what it owns other than that between a cause and its effect.
(Duerlinger 2004: 87–8; translation of Section 4 of the
“Refutation of a Theory of the Self”)
The Naiy yikas suggest that the self is the owner of the memories in just the way that Caitra is
an owner of a cow. Vasubandhu’s response is to say that the owner–owned relation between the
psychophysical continuum called “Caitra” and the psychophysical continuum called “Cow” is
just a causal relation between the continuums. This result is then generalized to explain that the
ownership relation (between an owner and what is owned) can be analyzed in terms of causal
relations (between a cause and its effect). Subsequently, this generalized account is itself extended
to answer the question: Who is it that apprehends an external object of experience, or an internal
state like pain, and so on? And again, who is it that performs an action, like walking? Vasubandhu
takes great care to address each of these questions in detail. But the strategy that he uses is essen-
tially the same as in the explanation of owner of memory experiences. So, when we say that
Devadatta apprehends or walks, we seem to imply that there is an owner of experiences who is
conscious or performs the action of walking, but this is wrong because an owner is just the cause
419
Monima Chadha
of the action which in turn is no more than an earlier mental or physical state in the continuum
of states that we call “Devadatta.” Thus Vasubandhu claims that the existence of a relation
between an owner and what is owned is nothing but the relationship between cause and effect.
Ganeri explains why this general analysis of ownership is found wanting by the Naiy yikas:
they offer a dispositional account of ownership defining it as the “power to use the object as one
wishes” (2007: 176–7). He argues that since it is possible for a capacity to go unexercised, there
may be no causal connections corresponding to it. Note, however, that Ganeri’s pessimism is
not justified as there is a prima facie case for metaphysical connection between dispositions and
causes: both are suited to (counterfactual) conditional analyses (Handfield 2009: 3). The coun-
terfactual conditional analyses of cause are of the form, “If it were to be the case that a (cause),
b (effect) would follow.” An analysis of cause in such a form can be offered to cash out causes
in terms of possibilities, even though there are no actual causal connections corresponding to it.
While there is reason to believe that Vasubandhu’s explanation is incomplete, there are no
sufficient grounds to dismiss his explanation of ownership out of hand. The owner of memory,
Vasubandhu argues, is an earlier, albeit special kind of, mental state which is its cause. And, when
we generalize this to explain what it is to apprehend, or be conscious of, an object, we have
to say that the owner or possessor of consciousness of an object is the cause of the conscious-
ness. In an ordinary perceptual experience, such a cause is the organ of perception, the object
of perception and attentiveness (Duerlinger 2003: 250). Though this sounds odd, Duerlinger
explains that Vasubandhu is not claiming that the meaning of “owner/possessor” and “owned/
possessed” is “cause” and “effect”; rather the point is that our notions of an owner/possessor and
what is owned/possessed are conceptual fictions, and thus only conventionally real. But when
we dig deeper to find out what, if anything, underlies these notions of owner/possessor, we
find that they correspond to those of cause and its effect (Duerlinger 2003: 251). That said, the
explanation of memory in terms of cause and its effect is incomplete because Vasubandhu does
not say how the relation between a specific cause and its effect is to be known. A more general
version of this problem was pointed out by the tenth-century Vai e ika philosopher, r dhara,
in a discussion with the Buddhists on memory:
[Buddhist:] As a result of there being a causal connection, a later memory [is a mem-
ory] of what was experienced at an earlier moment. The son does not, however,
remember what was experienced by the father; this is because there is no causal con-
nection between the cognitions of father and son, and their bodies, though admittedly
so [connected] are not [themselves constituted of] consciousness.
[ r dhara:] This is not well-reasoned, for in the absence of a self, there would be no
determinate notion (niścaya) of a causal connection. At the time of the cause, the effect
has yet to occur, and when its time comes, the cause has gone. Aside from the two of
them, some unitary perceiver is denied; so who would observe the causal connection
between those two things occurring in sequence?
(Quoted in Ganeri 2007: 177)
r dhara’s argument, Ganeri explains, is that in order to form a conception of a causal sequence,
one must be able to experience a sequence of events, as a sequence. However, in the Buddhist
Abhidharma picture of reality, all there is is the flow of momentary mental and physical states
(skandhas); there is as such no possibility of witnessing a drawn-out sequence (Ganeri 2007:
177). I believe that the Abhidharma view can be defended. While it is true that the relation
between cause and effect cannot be directly perceived because the cause ceases to exist before
420
the effect is perceived. However, the relation between cause and effect could, in principle, be
known by inference. For this to work, the Buddhist has to provide an account of how these
specific causal relations may come to be known by inference. I explore such an account on
behalf of the Abhidharma Buddhist philosophers in the last section. But first, in the next sec-
tion I present a model of memory from recent empirical psychology which shall be useful for
explicating the Buddhist account in the last section.
3. Episodic memories
The Naiy yika argument against the Buddhist seeks an explanation of what is called “episodic
memories” in the contemporary literature. This is evident from the examples they offer, “that
source of pleasure in the shape of garland, sandal paste and the like, which I remember is the
same that I am perceiving now,” “I am seeing the thing I saw before,” etc. The Naiy yikas
focus on examples of episodic memories because they think that every memory contains not
just an awareness of the object previously experienced but the awareness of the object as previ-
ously cognized by an agent of cognition. This may seem like a very “narrow” view of what we
think of as memory. However, it is interesting to note that contemporary psychologists and
philosophers are concerned that what we label now as memory is too broad and the range of
phenomena to which it currently applies is in serious need of pruning. Stanley Klein (2015)
argues that this narrow view is indeed right, that only episodic memory is, strictly speaking,
memory. In response, Kourken Michaelian (2015) agrees that some pruning is necessary, but
he argues that Klein goes too far. Michaelian claims that the arguments mounted by Klein
(including historical, conceptual, and empirical) support what he calls the intermediate view of
the scope of memory to include episodic and semantic memories as genuine forms of memory
but not procedural memory. I agree with Klein that debate should not be dismissed as seman-
tic hair-splitting because it has important consequences for the metaphysics of memory and
empirical research concerning memory. Furthermore, the reason I mention this contemporary
debate here is to set aside the concern that Ny ya demand for an explanation of episodic mem-
ories is unfair as it depends on an overly restrictive view of memory.5 However, my task here
is to attend to the empirical analyses of episodic memory to ask whether a causal account of
the sort that the Abhidhharma Buddhists can hope to offer has the ingredients for an adequate
explanation of episodic memories.
Klein, following Tulving (1985, 1993) argues that episodic memory and semantic memory
cannot be differentiated by content per se (2015: 16). Tulving (1985: 3) suggests that it is the
autonoetic consciousness “that confers the special phenomenal flavour to the remembering of
past events, the flavour that distinguishes remembering from other kinds of awareness, such
as those characterizing perceiving, thinking, imagining or dreaming.” The crucial features of
episodic memory (i.e., spatial, temporal, and self-referential) can also be on display in semantic
memory experience (Klein 2013: 2). For example, the content of semantic memory is capable
of contributing to a representation that includes spatial and temporal information, e.g., I know
that South Australia became the second jurisdiction in the world (after New Zealand in 1893)
to extend the vote to women and the first to also allow women to stand for Parliament in 1894,
even though I do not remember the event in which I acquired this knowledge. This exam-
ple shows that the self-referential element is not entirely the privilege of episodic memory.
Furthermore, a rather dramatic case presented by Stuss and Guzman demonstrates that there is
no principled reason why the content of semantic and episodic memory systems should differ.
The case of patient J.V. who suffered profound retrograde amnesia as a result of illness was
able to successfully relearn specific spatial and temporal details of his personal past. However,
421
Monima Chadha
these relearned memories are independent of a sense of personal intimacy (Stuss and Guzman
1988: 21). The authors suggest that the apparently absolute loss of personal memories can be
explained by either a retrieval deficit or a storage depletion. Stanley Klein claims that cases like
that of J.V. suggest that “the ‘time, space and self’ criterion for distinguishing between semantic
and episodic memory is insufficient for the task for which it was designed” (2013: 2). The case
of patient M.L. shows a similar pattern, but it also shows that amnesiacs suffering from a loss
of episodic memories because of injury also suffer from a compromised autonoetic awareness.
M.L. was able to retrieve a few pre-morbid personal experiences, but these memories were not
endowed with a sense of re-acquaintance with or reliving one’s own past. Developmental dis-
orders of episodic memory also show similar patterns, some content is preserved (or relearned,
depending on the age of onset) without any reliving or re-acquaintance with one’s past. Patient
M.S. who suffered neurological insult during childhood was unable to episodically retrieve
events in his life. M.S. was able to learn and report some facts about his life post-morbidity.
For example, he knew he had spent a recent winter away from home and could describe some
specifics but the acquired content lacked a sense of personal reliving (Broman et al. 1997).
Based on this, Klein argues that what distinguishes episodic memory is not the nature of its
content or the hypothesized module in which it is stored; rather memory becomes episodic in
virtue of retrieved content associating with autonoetic consciousness (Klein 2013: 3). The take
home message from this is that episodic memory is crucially identified in terms of the manner
in which content is given in awareness rather than the content per se. Klein (2015) argues that
episodic memory is the only kind of memory that properly deserves the name “memory” but,
as I have said before, that is not my concern here.
My concern here is with the fact that retrieved content gets classified as episodic by virtue of
its being acted upon by a contemporaneous act of autonoetic consciousness, which act confers a
sense of temporal subjectivity to it in the present (Klein 2013: 4). This fact has implications for
the debate between Vasubandhu and his Ny ya opponents. Autonoesis is the “unique aware-
ness of re-experiencing here and now something that has happened before at another time
and place” (Tulving 1993: 16). The important point to note here is that the special subjective
phenomenology, the “mineness” or the sense of ownership which accompanies episodic mem-
ories, is an add-on in the present (thanks to autonoetic consciousness). The content of episodic
memory is not encoded, stored, or retrieved in a special way; it is subjectively atemporal like
knowledge and belief. The content acquires subjective temporality in virtue of a concurrent
act of autonoetic awareness. Episodic memory does not entail a persisting self as the agent and
owner of experiences, but only a sense of the self as the agent of an action or the subject of expe-
rience. Vasubandhu can account for the sense of self, as we shall see in the next section, provided
that it is only a sense, an appearance. Vasubandhu holds that the self is really an illusion, in fact
a false mental construction based on a failure to “see things as they really are.” But first, there
is another point about the relation between episodic memory and autonoesis in the psychology
literature that deserves our consideration.
There is reason to think although content and autoneotic awareness are normally conjoined
in episodic memory, their relation is a contingent matter. Klein (2015: 18) calls this the rela-
tional interpretation. This interpretation is illustrated by the case of R.B. (for details, see Klein
and Nichols 2012) who as a result of a serious accident suffered head trauma that resulted in
several cognitive and memory impairments including retrograde and anterograde amnesia for
events in close temporal proximity to the accident. In particular, R.B. suffered a new form of
memory impairment; he was able to remember particular incidents from his life accompanied
by temporal, spatial, and self-referential knowledge, but he did not feel that these memories
belonged to him. His testimony is clear: he is recalling scenes, not facts. R.B. could clearly recall
422
a scene of being at the beach with his family in his childhood, but the feeling was that the scene
was not his memory; it felt as if he was looking at a photo of someone’s vacation. R.B. could
remember events from graduate school, could relive the experience, but “it did not feel like it
was something that had been a part of my life.” R.B. did not intellectually doubt his memories,
he believed that they were his memories because there was continuity of memories that fitted a
pattern leading up to the present time. But even that did not help change the feeling of owner-
ship. R.B. was also able to intentionally call such memories into awareness. The case of R.B.
(and of J.V. and M.S. mentioned earlier) shows that the feeling of ownership is merely a contin-
gent feature which is best explained as a relation between two functionally independent systems
contributing jointly to the experience of episodic memory (Klein 2013: 12). The hypothesis is
that memory storage and retrieval systems get coupled with autonoetic consciousness, which
is responsible for ownership of memories.6 The important point to learn from this case is that
the sense of ownership of memories is not an intrinsic feature of episodic memories—it can
be disassociated. More evidence, provisionally supporting the hypothesis that the content and
autonoesis can be disassociated comes from the fact that separate networks are associated with
content storage (structures in the medial temporal lobes) and autonoetic awareness (structures in
the frontal lobes). Content associated with medial temporal lobes on retrieval may or may not be
experienced as episodic memory depending on whether or not it gets associated with autonoetic
awareness, for the process of transformation is located in the frontal lobes. The point is that the
same content presented to awareness can be taken as “my past” or “not my past,” depending on
the functional integrity of the mechanisms that enable autonoetic awareness (Klein 2015: 19)
4. Buddhist explanation of episodic memory

The hypothesis put forward in the last section would be welcomed by Buddhist philosophers.
The subjective element is attached to the memory in the present, and that allows the Buddhist
philosopher to explain the special subjective feeling associated with memory without having to
postulate a kind of memorial content which is imbued with a past self. So for the very least, we
do not need a persisting entity to explain episodic memory. But we still need to say more about
how Vasubandhu can explain memory ownership in terms of causal relations.
The early Abhidharma philosophers, Vasubandhu included, struggled with the idea of explain-
ing the subjective phenomenology, i.e., the mineness or ownership of experiences including
memories. The canonical Abhidharma account of mind reduces it to fundamental atoms consti-
tuted by six kinds of awareness or consciousness (vijñānas). Five of these correspond to the five
sense organs (sight, touch, hearing, smell, and taste) and the sixth is a mental cognition (mano-
vijñāna). This picture of consciousness is counterintuitive to many ordinary facts of experience,
e.g., phenomenal unity of experiences, sense of self, the feeling of continuity, etc. Asa ga, the
noted Yog c ra philosopher and author—or at least the compiler—of Yogācārabhūmi7 intro-
duced storehouse consciousness (ālaya-vijñāna) and ego-consciousness (kliśṭa-manas or simply
manas) in response to these limitations. The first is a constant and neutral baseline consciousness
that serves as a repository of all basic habits, tendencies, and karmic latencies accumulated by the
individual, providing some degree of continuity to mental states. The second can be thought of
as an innate sense of self arising from the apprehension of storehouse consciousness as being a self
(Dreyfus and Thompson 2007: 112). From the Buddhist point of view, this sense of self, though
mistaken, is a mental imposition of unity where in fact there is only a multiplicity of interre-
lated mental and physical events: that is why it is kliśṭa, literally afflicted. Vasubandhu endorses
Asa ga’s conception of the mind/mental states in his later work (e.g., Tri śikā-kārikā, or Thirty
Verses) and develops it further. It is important to note that this later model involves a significant
423
Monima Chadha
shift in Buddhist thinking. Vasubandhu appreciates that once the self is relegated to the realm of
illusion, metaphysically speaking, the work delegated to the self has to be farmed out to another
faculty. Thus there arises a need to enrich the sparse model of mind with the six conscious
states postulated by the early Abhidharma philosophers. However, it is important to note at
the outset that Vasubandhu never compromises on the basic tenet of impermanence: vijñāna
(consciousness) is always to be thought of as continually evolving streams of mental moments
or elements. The enriched model of mind is a more complex system of mutually interacting,
though continually changing, mental states. Mind cannot be thought of as a persisting entity.
Thus, the storehouse consciousness and the ego consciousness are not unlike the momentarily
arising discrete sensory perceptions in this and other respects, as we shall see below.
The question then arises: how does this enriched post-Abhidharma model which includes
the storehouse consciousness and ego-consciousness among its components explain episodic
memory and its ownership? I will show that the new model of memory with autonoetic con-
sciousness as a separate subsystem that accounts for the special feeling of “personal ownership”
fits in nicely with the post-Abhidharma model of mind. However, first we need to get clear
about the role and nature of the two new consciousnesses and their interaction with the other
six consciousnesses. The Buddhist terms “vijñāna” and “citta,” generally translated as “conscious-
ness,” have many different connotations in the Buddhist texts, depending on specific contexts.8
Vijñāna may indicate three things: a conscious state, an instrument of consciousness, and the
activity that produces a conscious state. These three meanings can be associated as well with
the storehouse consciousness and ego-consciousness as conceptualized by Yog c rins. They also
apply equally to all the six vijñānas accepted by all Abhidharmikas. For example, caksurvijñāna,
or eye consciousness, as a conscious state has forms (in the sense of sights) as its object, the
eye as the instrument and the activity of eye-perception that produces consciousness of the
forms. On similar lines, as a conscious state, storehouse consciousness is an implicit awareness
of one’s immediate environment and also a continuous, albeit implicit, perception of one’s
own body—indeed more than just the material body; it also includes the implicit awareness
of the predispositions (cognitive and affective conditionings persisting from the past). As an
instrument of consciousness, storehouse consciousness functions as the receptacle of all seeds
(sense-impressions, memory-impressions, afflictive tendencies, emotions, and karmic residues).
As an activity, storehouse consciousness shapes current cognitive experiences in that it provides
a background awareness which is continually produced and is ever present “as an ever renewed
background of mental presence out of which the more focussed and manifest forms of cogni-
tive awareness arise” (Dreyfus 2011: 150). There is a two-way dynamic between storehouse
consciousness and ordinary perceptions: storehouse consciousness insofar as it contains seeds or
predispositions produces conscious states (almost always in association with sense faculties and
their objects) say, of seeing a mango, which, in turn, accumulate further seeds, say, desire for a
mango, into it. In contrast to the early Abhidharma model, here we are presented with a holistic
model of the mind—wherein perceptions and other mental dispositions are tied together in a
continuous feedback cycle. Manas or ego-consciousness also fits into this tripartite scheme. As
a conscious state, it takes the storehouse consciousness as its object and treats it as a continu-
ing self. This is a mistake, of course, because the storehouse consciousness itself is an evolving
stream of momentary states. As an instrument, ego-consciousness provides the support, just like
the sense organs do in respect of the five sensory consciousnesses, for the sixth kind of con-
sciousness, i.e., mental cognition. As an activity, ego-consciousness attaches a sense of personal
ownership or mineness to conscious states. In this sense, it accounts for the persistence of the
sense of self, a sense which always exists in ordinary conscious states except in the saintly states.
Here is how it is explained in the text:
424
Supported by the eye, etc. and form, etc. eye perception, etc. arises. Immediately
afterwards consciousness perception (manovijñāna) becomes aware of the object
(visaya). [This consciousness perception has manas as its support.] From completed
awareness arises lust and the other passions. Then action (karma) arises, which is
called samskāra of meritorious, sinful or neutral character . . . In this way the pas-
sional elements produce sensory impressions, memory impressions and other vāsanās
in the storehouse perception.
(Galloway 1980: 12; slightly modified)
In reality, the manas has the storehouse perception for its phenomenon. Gunaprabha:
“This means that it phenomenalizes [sees] the storehouse perception as a self.”
Vasubandhu: “It is that which is associated with the constant delusion of self-view,
of self-egoism, self-love and lust for self and so on.” Gunaprabha: “It is explained
as operating always, and arises as good, bad and indifferent. His saying ‘It is of one
class’ means [in contrast to what it means for the storehouse perception that it has
a passionate (klista) nature.] ‘It is continually produced’ means that it is momentary.
It operates always, but ‘It is not present in an Arhat, on the Noble Path, or at the
time of the cessation-attainment.’ In the last two it is prevented from producing
vāsanās; when one rises out of them, the seeds arise again from it. In Arhatship they
cease completely.”
(Galloway 1980: 18; slightly modified)
This way of conceptualizing the two new kinds of consciousness introduced by Yog c rins
has the advantage of giving a uniform tripartite analysis of all various kinds of conscious states
and also explains the interactions between these conscious states. But more importantly, the
tripartite analysis of ego-consciousness shows that it is both the cause as well as the owner
of a memory. Manas or ego-consciousness seems to be the most likely candidate for the
owner of a memory since it appropriates the seeds (memory impressions, etc.) from the store-
house consciousness and attaches a sense of personal ownership to these states. But, note that
ego-consciousness in the sense of an instrument is also a cause of the memory impression
(together with sensory perceptions), since manas in this case is the instrument or support for
an earlier conscious perception (manovijñāna), which, in Vasubandhu’s words, is a completed
perception of the object following from sensory perceptions. Ego-consciousness is responsible
for the sense of mineness associated with the earlier experience which is now remembered in
the sense of reliving that earlier experience in a memory episode. Ego-consciousness, or rather
a part thereof,9 in this sense is both the cause and the owner of the memory. This way of think-
ing about the interaction between the different mental states or consciousnesses explains his
specific comment that “the cause of the memory is the owner of the memory.” This account
also shows that ownership is not an intrinsic feature of the memory impression in the way that
it is stored or retrieved, rather the sense of mineness or personal ownership is an add-on in the
present (thanks to manas) after the memory has been appropriated by it. I am suggesting here
that ego-consciousness, or a part thereof, plays the role of autonoetic consciousness in that it
imposes a sense of mineness or personal ownership to memory-impressions it appropriates from
the storehouse consciousness by placing them in a particular time in the causal history. But this
placement is a mistake, for history is just that: history. It no longer exists and appropriating the
causal history of the mental stream and particular time moments in it as part of one’s personal
past is just a trick conjured by the ego-consciousness (kliśṭa-manas) to deceive the ever-evolving
mental stream into believing that there is a continuously persisting self.
425
Monima Chadha
Thus the account of mind developed by Vasubandhu in the post-Abhidharma period can be
thought of as an evolution of the earlier model of mind in response to the concerns raised by
the Naiy yikas. And this account seems to be well supported by contemporary psychology. At
least in the context of memory and its ownership, the work of the concept of self is done by
elements of the model of mind developed during the Abhidharma and the post-Abhidharma
period of Buddhist philosophy.
Notes
1 Indian Buddhist philosophy can be roughly divided into three major traditions: the Nikāya, the
Abhidharma and the Mah y na.
2 The “Refutation of a Theory of Self ” forms the ninth chapter of his Abhidharmakośa and Abhidharma-
kośa-bhāsya which are perhaps the most important texts in the Abhidharma corpus. the Vasubandhu
defends the Buddhist no-self doctrine not only from the continuing onslaught of Hindu philosophers,
especially Ny ya-Vai e ikas, but also rejects the theory of persons proposed by other Buddhists, namely
Pudgalav dins and Madhyamikas.
3 Duerlinger’s (2003) is the most detailed reconstruction of Chapter 9 and its arguments. It will become
evident that my reading of the text differs from Duerlinger’s. He suggests that the Vasubandhu argument
is that we are not selves and that we ultimately exist. I cannot see any reason to the attribute such a view
to Vasubandhu.
4 See Strawson (1959: 95–8).
5 And as a corollary for Klein to note that there is cross-cultural support for the view of memory he
defends in his 2015 paper.
6 Stanley Klein; personal communication.
7 Yogācārabhūmi is recognized as the authoritative text for the Yog c ra tradition, but it was not written by
a single author. It is a compiled collection of works written over the centuries.
8 “Citta” is another term used sometimes as a synonym for manas and vijñāna, though I do not mention it
in the context of this discussion.
9 The qualification is important. Manas is what explains the sense of self emerging from distinct men-
tal states and memories are only part of it. But as manas is the instrument for conscious perception
(manovijñāna), it can also be thought of as responsible for attaching a sense of mineness or personal own-
ership to concurrent perceptual cognitions.
Bibliography
Broman, M., Rose, A. L., Hotson, G., and Casey, C. M. (1997). Severe Anterograde Amnesia with Onset
in Childhood as a Result of Anoxic Encephalopathy. Brain, 120: 417–33.
Chadha, M. (2013). The Self in Early Ny ya: A Minimal Conclusion. Asian Philosophy, 23(1): 24–42.
Dreyfus, G. (2011). Self and Subjectivity: A Middle Way Approach. In M. Siderits, E. Thompson, and
D. Zahavi (eds.), Self, No-Self? Perspectives from Analytical, Phenomenological, and Indian Traditions
Dreyfus, G., and Thompson, E. (2007). Asian Perspectives: Indian Theories of Mind. In M. Moscovitch,
E. Thompson, and P. Zezalo (eds.), The Cambridge Handbook of Consciousness. Cambridge: Cambridge
University Press.
Duerlinger, J. (2003). Indian Buddhist Theories of Persons: Vasubandhu’s Refutation of the Theory of a Self.
London: RoutledgeCurzon.
Galloway, B. (1980). A Yogacara Analysis of Mind, Based on the Vijnana Section of Vasubandhu’s
Pancakhandaprakarana with Gunaprabha’s Commentary. Journal of International Association for Buddhist
Studies, 3(2): 72–5.
Ganeri, J. (2007). The Concealed Art of the Soul. New York: Oxford University Press.
Handfield, T. (ed.). (2009). Dispositions and Causes. New York: Oxford University Press.
Hume, D. (1739–40/1888/1978). Treatise of Human Nature (L. A. Selby-Bigge (ed.), 2nd edn.). Oxford:
Clarendon Press.
Klein, S. B. (2010). The Self: As a Construct in Psychology and Neuropsychological Evidence for Its
Multiplicity. Wiley Interdisciplinary Reviews: Cognitive Science, 1(2): 172–83. doi: 10.1002/wcs.25.
426
Klein, S. B. (2013). Making the Case that Episodic Recollection Is Attributable to Operations Occurring
at Retrieval Rather than to Content Stored in a Dedicated Subsystem of Long-Term Memory. Frontiers
in Behavioral Neuroscience, 7. doi: 10.3389/fnbeh.2013.00003.
Klein, S. B. (2015). What Memory Is. WIREs Cogn. Sci., 6: 1–38. doi: 10.1002/wcs.1333.
Klein, S. B., and Nichols, S. (2012). Memory and the Sense of Personal Identity. Mind. doi: 10.1093/
mind/fzs080.
Michaelian, K. (2015). Opening the Doors of Memory: Is Declarative Memory a Natural Kind? WIREs
Cogn Sci, 6: 475–82. doi: 10.1002/wcs.1364.
Pruden, L. (1988). Abhidharmakośabhāyam English Translation of Poussin, Louis de la Vallée., 1923–1931/
1980. L’Abhidharmakośa de Vasubandhu, 6 vols., Bruxelles: Institut Belge des Hautes Études Chinoises. Berkeley,
CA: Asian Humanities Press.
Strawson, P. F. (1959). Individuals: An Essay in Descriptive Metaphysics. London: Methuen.
Stuss, D. T., and Guzman, D. A. (1988). Severe Remote Memory Loss with Minimal Anterograde
Amnesia: A Clinical Note. Brain and Cognition, 8(1): 21–30.
Suddendorf, T., and Corballis, M. (1997). Mental Time Travel and the Evolution of the Human Mind.
Genetic, Social, and General Psychology Monographs, 123: 133–67.
Tulving, E. (1985). Memory and Consciousness. Canadian Psychology, 26(1): 1–12.
Tulving, E. (1983). Elements of Episodic Memory. Oxford: Clarendon Press.
Tulving, E. (1993). What Is Episodic Memory? Current Directions in Psychological Science, 2: 67–70.
Wheeler, M., Stuss, D., and Tulving, E. (1997). Toward a Theory of Episodic Memory: The Frontal Lobes
and Autonoetic Consciousness. Psychological Bulletin, 121: 331–54.
427
34
CHINESE BUDDHIST
PHILOSOPHY
Chung-Ying Cheng
1. Background from two Indian Buddhist traditions

Memory was approached and treated in Indian and Chinese Buddhist traditions in different
ways than in the West, and they break new grounds for cognitive research in memory in mod-
ern days. Their approaches are both pragmatic in yielding practical norms for life, and at the
same time analogical to our common-sense understanding of things in nature.1 They may not
be systematic and have to be seen as phenomenological, but not as analytical.2
In early Indian Buddhism, there is the teaching of the three seals from Buddha which states
that “all things are impermanent, all dharmas have no self, and the nirvana is quiet.”3 Founder
of Buddhism Sakyamuni Gautama (563–411 bce) denies ultimate personal selfhood for human
persons. His ground for doing so is that the life process of a human person is marked by suf-
fering through birth, old age, illness, and death. To stop suffering, one way is to save life by
means of a divine savior one can trust; the other way is to destroy the seeds of life so that one
would remain in complete extinction of actions by one’s own efforts.
The former is the Western theological position, whereas the latter is the Buddhist position;
both positions are rooted in a deep experience of life in one way or another. For Buddhism,
one must be awakened to the fact that things have no essence and the world is in flux (samsara)
without permanence. In fact, the impermanence of the world can be seen as a matter of the
lack of essence of things. We may indeed speak of causes and effects embodied in the doctrine
of karma (action-deed), based on Buddha’s enlightened insight. This means that Buddha must
reach a point of understanding called “enlightenment” (wu悟).4
Karma here means results from one’s actions in life and it always involves doing the wrong
thing or the right thing. Doing the right thing will advance one to a higher state of life, while
doing the wrong thing will downgrade one to a lower state. But this is no real solution for the
suffering of life. The real solution comes from deconstructing an impermanent self caused by
desires, emotions, and knowledge claims, and realizing the emptiness of life and being.
It was Nagarjuna (Longshu 龙树 in Chinese, 150 ce) who came to hold the emptiness of
all things because things as entities are impermanent and because they are not to be identi-
fied objectively nor subjectively, but completely subject to the karmic force. This emptiness
(sunyata in Sanskrit and kong 空 in Chinese) is neither existence nor non-existence. Things
arise by way of gathering causes which do not abide nor have continuity of their own.
428
Chinese Buddhist philosophy
Thus the so-called “reality” is just a matter of dependent conditions (pratitya samutpada)
emerging from sunyata and known by conventions.
As to how this view of reality can be made compatible with common sense regarding things
and people, it is both a question and a challenge. To answer this question and challenge, we
encounter the subsequent idealistic school of consciousness only (Yogacara/Vijinavara) in
Indian Buddhism as founded by Asanga and Vasubandhu (known as Shiqin世亲and Tipo提婆
in Chinese, fl. in the fourth century ce).
In my view, both the tradition of Madhyamika and the tradition of Yogacara are derived
from the fundamental insights of Gautama regarding the nature of reality and self. According to
Madhymika, reality is sunyata in which there is no self and there is no consciousness for itself to
be identified in any manner. What is identified conventionally in language and consciousness
is neither real nor unreal, and we simply should not cling to it as if it is truth. On the other
hand, what is conventionally presented in language and consciousness still has its conventional
reference and meaning which appeal to our consciousness and is capable of generating further
consciousness on our part. In this sense, we have to account for their happening and their rel-
evance in our life. In this account, memory functions in defining and developing a conventional
selfhood of a person.
According to Asanga and Vasuhandhu, one forms one’s self by projection of one’s con-
sciousness, which is given in the formation of one’s mind. Although accepting sunyata as the
ultimate reality, they take experienced consciousness as a fact, and this fact points to a force
of consciousness making in oneself like the world-making Maya in early Indian Brahman
religion. Once consciousness arises, it would have the latent power of perfuming (vasara) the
void space of sunyata, and thus beclouding the ultimate reality through the illusions of our
consciousness. “Perfuming” will induce further experiences of consciousness through the
formation of seeds (bija) of consciousness which leads eventually to different individual centers
of consciousness. This original consciousness which is the source of all later consciousness and
their effects is called alaya consciousness (storehouse consciousness藏识), literally meaning
“consciousness hidden and stored.”
By taking this view, we could recognize the reality of mind or consciousness on a level
different from the level of the ultimate truth, namely the sunyata. Unlike Descartes, the alaya
consciousness is conceived to support all individual minds and the consequent selfhood of
human persons. Alaya consciousness seems to be individuated, for each person’s identity and its
changes are explained by each person’s own alaya consciousness.
2. Formation of the human self and the logical status of memory

In the Platform Sutra of the Sixth Patriarch, Huineng (慧能 638–713 ce) in Chinese Chan Buddhism,
we have the following description of the formation of human mind and consciousness:
All things are included in your own natures; this is known as the storehouse (alaya)
consciousness. Thinking hence turns up consciousness, and thus the six conscious-
nesses are produced, and the six dusts (phenomena) are seen emerging from the six
gates (sense organs). The three sixes make eighteen. From the errors of your self-
nature the eighteen errors arise. If your self-nature is correct, then the eighteen correct
things arise. If it contains evil activities, then you are a sentient being; if it contains
good activities, then you are a Buddha. From what do activities stem? They stem from
the ‘confrontations’ (dui 对) that face your own nature?5
429
Chung-Ying Cheng
It is important to observe that the storehouse consciousness is conceived as part of the true
nature of a person which in itself is empty. But when the storehouse consciousness becomes
separated from one’s true nature (self nature), then there arises the activities of consciousness
which lead to the formation of the objects of senses and the organs of senses. The dualistic
confrontation of the objects with the subject-senses reflects the confrontation of the true nature
with the consciousness of the true nature which turns out to be separated from the true nature.
This means that the world of conventional reality is a result of the separation of nature and
consciousness. It must be granted that it is the emergence of consciousness which makes this
separation possible. Hence consciousness must be regarded as a different principle of being than
the principle of the ultimate being, namely the principle of sunyata which can be said to have a
power of transformation.
The second observation from the above is that human self is presupposed in our conscious-
ness of things. For it is based on the assumption of self as a principle of organization and
integration that all relevant experiences of a person can be said to be related or interrelated.
Hence it is a principle of inner unity which could lead to self-consciousness: one could speak of
oneself as the underlying entity which supports and shapes the self into a centered self. But if we
take the principle of the ultimate reality seriously, we must say that there should not be such a
principle of inner unity of the self, for there is no self to start with. Now if we wish to speak of
self as an existing entity, we must build such an entity from basic experience of the five senses.
This of course still presupposes some fundamental principle of consciousness from which the
self can be built up from experience. In the aesthetic experiences of the five senses, we may not
immediately experience a separation between the subject and the object. So we must assume
that it is something behind the five senses, like our thinking which by nature is intentional.
Then, to explain the intentional thinking of a person, we have to conceive of a person as hav-
ing a mind as self-owning entity. We may further speak of the self as an enduring subject which
could be the source of action. At this point, we may say we have arrived at a deeper sense of
self. The grounding of such a self is then the alaya or storehouse consciousness which acts as the
ultimate potential source of all human actions.
3. Formation of memory in virtue and context of formation of self

Huineng uses the word “dui” (对) which Philip Yampolsky translates as “confrontation” in
the presentation of our experienced world.6 In one sense, it is indeed confrontation insofar as
we see how experience and consciousness is intended to confront self nature or the ultimate.
However, the word “confrontation” is misleading, as it suggests an antagonistic opposition.
There is also the non-antagonistic relation of pairing between the truth of sunyata and being of
a self. In fact, they could form a pairing relationship of polarity which would lead to better and
enriched understanding of both reality and self. Thus Huineng is able to speak dialectically of
36 pairings (dui 对) which covers formation of objective nature, such as heaven and earth, sun
and moon, and then categories to do with language and facts, and so on. The purpose of doing
so is to make sure that mind in oneself could make sense of this world, even though this world
is empty in the ultimate sense.
Based on this contextual background we can now raise the question as to how the self is
formed by perception and memory, and how the latter has to take a turn to interpretation for
correct action and perhaps for enlightenment toward the ultimate reality.
There are two ways in which memory can be said to be formed by self and the self can be
said to be formed by memory. In effect, this means that there is a mutual formation relation-
ship between memory and the self which must be logically determined by the fundamental
430
requirements of the two Indian Buddhist traditions: the Madhyamika and the Yogacara. For
the former, there is neither self nor memory. For the latter, consciousness must be sustained
and maintained in some causal way. Now as memory comes from perception in conscious-
ness and continues as a form of perception of continued consciousness which forms the
content of the emerging self, one can thus see how memory requires a container (the self) to
hold its existence while the self can use memory to enhance its self-understanding. Of course,
we have to admit that memory, once derived from the perception, can be long or short,
depending on where it was deposited in the brain. The brain can be conceived as a modern
scientific symbol or index for the self-nature (自性zixing) in the Buddhist psychology. We
thus come to my Yogacara-based thesis: There is no self without memory and there is no
memory without a human self.
In the terminology of Yogacara, we can see that the perceptive activities constitute the first
five consciousnesses, namely the visual, the auditory, the olfactory, the gustatory, and the tactile,
which have a wide contact with the appearing world. These five sense-consciousnesses are based
on the thinking or mental consciousness which is called the sixth consciousness, or conscious-
ness of thinking or mind. Specifically, the mind consciousness (yi-consciousness意识) has the
function of making distinctions between the inner and the outer, between the past, present, and
future of time, and thereby expressing the activities of thinking, reasoning, judging, remember-
ing, and anticipating. Sometimes, these activities are further classified into the following four
types according to some account: type 1 consciousness which clarifies the first five conscious-
nesses; type 2 consciousness appearing in meditation; type 3 consciousness, which gives rise to
memory, expectation, and imagination, and type 4 consciousness in dreams, loose and vague
and even unconnected.7
It is clear that memory cannot be as clear as perception but merely reflects perception with a
similar content of non-presence. It is perceptual consciousness with objects moved away in time
and space. However, all these activities are rooted, and to be unified in the seventh conscious-
ness which is referred to as mana-consciousness or self-consciousness. Self-consciousness could
be explained as some deeper mind which ties the previous six consciousnesses together to create
a sense of continuity and identity. It is therefore also referred to as the I-consciousness (我识). It
can be also described as the Holding Will underneath the previous six consciousnesses, in that it
may initiate decisions and actions which have causal effects on the self. If we wish to speak of the
interaction between the self and the world, it is the seventh or the I-consciousness, that takes an
active role of making responses, whereas the first five consciousnesses basically remain receptive
and passive and the six consciousness makes an inner movement within the I-consciousness.
However, this is not the end of the story, for we have the eighth consciousness to deal with, in
the ontology of memory and self in Yogacara.
What is the eighth consciousness? It is the hidden consciousness which is called the alaya
consciousness or storehouse consciousness, as mentioned above. It is in the alaya consciousness
that epistemological capabilities of senses and teleological dispositions of mind and will are to be
ontologically rooted or grounded. It is therefore the source of power which makes the previous
seven consciousnesses possible. In a sense, all the prior consciousnesses are creations of alaya-
consciousness and their activities must be conceived as taking place within the framework of the
eighth consciousness. Besides, the eighth consciousness can be seen as the underlying deep and
permanent self, which is impervious to life and death, and thus is across the life and death of an
individual sentient being. It is also capable of storing all the perfumed seeds so that it may issue
them in the form of objective entities and subjective selves in the world.
In fact, it is what the mana (I-consciousness意识) clings to that it may control the first six
consciousnesses and lead to thinking of objects of the world as objects without knowing. It is
431
Chung-Ying Cheng
therefore a matter of reification resulting from functioning of mana or empirical self. Mana does
not know that the objects and events of the world are in fact the projections of the alaya con-
sciousness, and thinks that there are such things outside one’s mind. This leads to the occurrence
of various forms of karma and thus entrenches the human self in an eternal cycle of transmigra-
tion or samsara (lunhui 轮回).
From the above, it is clear that the mana consciousness plays a dual function in the forma-
tion of the sense of self in our conventional language. It maintains the unity of the empirical
self and makes it an agent in the world on the one hand. On the other, it holds an unbreakable
persistence of the self based on its unconscious clinging to the alaya consciousness which enables
the metaphysical selfhood and its recycling to take place according to the laws of karma. The
alaya consciousness is no doubt extremely powerful as it hides all its seeds and causes formation
of empirical entities of things, including sentient beings such as humans. In fact, it contains two
kinds of opposite functions of human existence: one kind is the opposition of the object versus
the subject; the other is that of the cause versus effect. It is by way of producing these double
oppositions that the alaya consciousness is able to hold all sentient beings as items or pawns in a
cycle of samsara or flux.
According to Vesubandhu, the alaya consciousness may contain both evil and impure seeds
and good and pure seeds. This assumption is important for enlightenment of a human being,
as one could transform seeds of evil into seeds of good through a process of self-cultivation
and thus gain release out of the cycle of life and death as artificially created illusion from alaya
consciousness. Here we must recognize the important difference between mana consciousness
and alaya consciousness: whereas alaya consciousness is cosmological, the mana consciousness
is essentially sentient (human) and epistemic, but subject to the laws of karma. In a sense, the
alaya consciousness is practically the same as the Maya which, in the Brahman tradition, creates
a grand illusion in which the human can be trapped. But as a human being, one could strive
to get out of this grand illusion and thus transcend all perception and memory and thus gain
freedom. When a human being is able to do this, he is said to reach a state of enlightenment
called wu (悟) or awakening.
We have thus described how memory and self are related from a Yogacara tradition of the
Two Brothers. In this perspective, memory is treated as removed perception which forms the
receptive content of the human self. This point is not explicitly mentioned in the Yogacara
literature, but it is logically obvious that we must recognize memory as result of empirically con-
sumed perception in a posited human self with a metaphysically assumed identity of underlying
alaya consciousness. Of course, the alaya consciousness has endowed the empirical self with
more content than memory, as the human self can be seen as an agent with imagination and will
apart from memory, but memory no doubt provides the necessary epistemic basis for identifying
the empirical human self as it is tied to time and to the effect of causal action.
With regard to the latter point, it is clear that memory, once formed, will also function as
the source of induction, imagination, and expectation of the future which makes interpretation
of human self-identity meaningful. We may indeed speak of three forms of identity of a human
person: the temporal identity based on short memory, the long-term identity based on long or
life-long memory and its narrative.8 Finally, there is assumed the metaphysical identity of the
human self which transcends temporary and life-long identity, forming a karmic identity which
spans across generations of life-times due to the presence of alaya consciousness. In this sense,
we may speak of memory as independent of any empirically accessible way of remembering. It
requires a special way of penetration for retrieving which stages or what grounds it has devel-
oped. In this sense, the Yogacara position on memory has gone far beyond the cognitive science
model of data processing.
432
One relevant remark concerns our idea of history based on the Yogacara view: history can be
considered an aggregate or group memory of important events and matters of community life,
with such a group memory objectively recorded and presented. History is therefore possible on
the ground that private memories are made intersubjectively verifiable and objectively accessible
through a process of reification by the human mind or reason.
4. Awakening of the non-self from memory-self

Whereas the Chinese school of consciousness only shares with the Indian Yogacara tradition
the basic constitution of the human self and the role of perception and memory in the for-
mation of the human self, Chinese Buddhism has the distinction of developing theories for
resolving the problem of human identity in a unique way, namely the way of enlightenment
(wu 悟), in the spirit of the Daoist paradigm of wu-dao (悟道enlightened of the way). The term
wu 悟 is an ancient word explained by Shu Chen (first century ce) in Shuowen as “awake”
and is equivalent to the word “jue” 觉, literally wakefulness in distinction from sleepiness or
drowsiness. To be awake or to awaken from sleep is to have a clear consciousness and be able
to have control of one’s mind. By analogy, by becoming awakened or enlightened, one is made
to see better and realize truth better and thus we can speak of truth understanding as a matter
of being awakened or enlightened to truth. Then there are two things to be said about this
wakefulness or being awakened: first, it is a deeper truth to be awakened to—for the Daoist, it
is to understand the dao which is behind and inside all things according to Daoism. It is a truth
which cannot be spoken nor articulated, as language would form a block in one’s making a
statement or saying or naming of the dao. It is immediate and total understanding which defies
descriptive language.
Hence to be awakened to the dao is to have an immediate and total grip or participation in
the dao as the ultimate reality. Second, as the term wu 悟ideograph indicates, wu is the enlight-
ening action or state of the mind, for it is in one’s mind that we can attain a deeper and deeper
sense of enlightenment or illumination from the dao. Neither Daodejing nor Zhuangzi used this
word, but its original meaning makes it possible for later scholars and philosophers to speak of
being awakened to the dao (wu-dao悟道) and thus to the self-nature of oneself in emancipation
from the grand illusion of our consciousness and conventions.
In light of the above, the gist of the concept of wu is to have reached the ultimate reality
with our mind, in our mind for the illumination of our nature as identifiable with the ultimate
reality. There is no implication for how one may reach the ultimate reality nor implication
in which manner the ultimate reality is to be reached. Later in the development of the chan
(禅meditation), one can speak of gradual enlightenment as versus sudden enlightenment, which
suggests two different models of reaching the ultimate reality. On the gradual model, one strives
to reach the peak of a hill by gradually climbing. Before you reach the peak, one views the
landscape part by part until you see all the parts of the landscape which is the state of wu. On the
other hand, there is the model of sudden enlightenment, which consists in immediately coming
to see the whole scene which differs radically from what you are familiar with. There must be a
breakthrough point where the whole is seen as different from the part or parts.
Of course, in the sudden enlightenment model, one needs not to deny that gradual movement
and progress may not be necessary as a preparation or a base for sudden and total breakthrough.
Besides, there is no necessity that either suddenism or gradualism must be fitting for every indi-
vidual, since each individual occupies a unique position in the indefinite space and temporal flux
of the alaya consciousness. Experience seems to indicate that, whether through gradual efforts
or some form of shock therapy or incidental contingency, enlightenment frequently happens
433
Chung-Ying Cheng
like a flash of lightning and for this, the theory of sudden enlightenment is no doubt an attrac-
tive theory, the attractiveness of which is shown in the widespread prevailing of the Linji Chan
tradition since Mazhu Daoyi (马祖道一 709–88 ce).
We could apply these two models to memory retrieval when we try to remember things
which have taken place in our experience. There two models suggest that we could have a sud-
den memory as we may have targeted an end and yet could not reach it by slow work. The idea
of sudden enlightenment is to look for an extraordinary way of reaching the end. We could also
have gradual memory as far as we could reach our goal by either the method of free association
or by the method of elimination. Whatever we wish to remember forms part of our genuine
experience, we have to use sudden memory or gradual memory to recover or recoup what we
have once experienced as truth. What then is the basic principle which makes either possible?
In light of what Chinese Chan Buddhism teaches such as realized in Huineng, the human self
is in fact on the one hand defined by our various grades of memory plus our present percep-
tion and anticipation of the future, while on the other hand, it is conceived as against a non-self
nature which is rooted in the ultimate reality called sunyata. The non-self of the self is precisely
this sunyata, which could fall into the memory-self with the activities of the alaya conscious-
ness. When the Chan master speaks of the awakening to the genuine self-nature, it is intended
to make the non-self-self of the self to be illuminated by either sudden enlightenment or by
gradual enlightenment.
5. Buddha nature and the ontological implication for the memory-self

Given these basic remarks, we may have to face the philosophical issue as to whether there is
a level of consciousness which transcends the eighth consciousness, the alaya consciousness.
Again, the answer is that it depends on how we actually understand the alaya consciousness.
The fact that we can become aware and inspired by a heart of wisdom to seek enlightenment
has to be understood as coming from the depth of our existence, which must include the eighth
consciousness. In fact, the eighth consciousness needs not to be seen as an independent entity or
state which has no source of validity and thus functions like seven prior consciousnesses. Insofar
as it is conceived as having seeds which perfume the later development of the seven conscious-
nesses, it could also be conceived to have the power not to germinate the seeds. If it is also
conceived to have produced impure seeds, it can be also conceived to have produced pure seeds.
In this manner, we may conceive the eighth consciousness as having a creative nature, which is
not dependent on any other nature or is antagonistic to any other nature. On the contrary, we
can conceive it as a self-nature capable of creative self-change which explains the existence of
the sentience of beings and which is capable of making responsible choices between the pure
and the impure. It is therefore something we can call the world of possibilities which has a dis-
position to actualize in ways which is undetermined.
With this understanding and interpretation of the eighth consciousness, the eighth con-
sciousness is both the body and the function of the self as it is. This means that it is internally
creative and not just causative and there is no break between the body and its function. In this
sense, perhaps we cannot speak of a ninth consciousness. On the one hand, there is indeed the
issue of source. Can we identify the source of this body and function of alaya? Again, as we feel
no need to posit for the external arising of the eighth consciousness, we may indeed take the
source to be the ultimate, namely the sunyata; we again have no need to posit a ninth conscious-
ness between the sunyata and the alaya consciousness, for the sunyata as the source simply could
internally make the alaya consciousness a pure seed consciousness. Hence we have no need to
posit for a ninth consciousness.
434
A good consequence of this argument is to make it possible for the achievement of the pure
seed alaya through awakening and effect, so that one would not necessarily fall back to the
source, which would be the extinction of all consciousness. This possibility of achievement of
the pure seed alaya, I believe, is required for achieving Buddhahood. This is important, because
we could then conceive Buddhahood as a positive concept rather than a negative concept.
As a negative concept, Buddhahood is equivalent to nirwanna, whereas as a positive concept
Buddhahood can in fact be the basis for becoming a bodhisattiva or even be able to lead a form of
existence perhaps to be labeled as sambhogakaya body, or reward body among the three bodies
in the Yogacara tradition.
In light of the above discussion, I wish to assert that the idea of alaya consciousness can be
actually further explained as the open and creative dialectical unity of the pure and the impure:
it is in fact reflected in the “Doctrine of One Mind with Two Functions” (two gates) in the
Chinese Buddhist book Awakening of Faith in Mahayana (大乘起信论).9 In this work, it is
asserted that one can enact purity of understanding and action and engage with impurity of the
secular world at the same time.
From this it is clear that it is up to one’s mind to make distinctions and differences or not
to make such distinctions. This could mean that one’s mind is between awakening and non-
awakening and it could become awakened and thus achieve a positive status of awakening.
One could suggest that this unity of purity and impurity is indeed the source of inspiration for
all major Chinese Buddhist traditions or schools, including Tiantai (天台), Huayan (华严) and
Chan (禅) Buddhism. But this does not mean that there are no fundamental differences among
them. With these differences in mind, one can finally explore how the problem of ontological
status of memory is to be conceived and settled.
6. Chinese Buddhism as ways of transcending memory

Based on the recognition of the source as the sunyata as described above, we have three schools
of Chinese philosophy which could be said to provide guidance and resources for thinking
out the problem of self identity and its dissolution. The first is the Tiantai School represented
by Zhiyi (智顗 538–97 ce): the second is the Huayan School as represented by Facang (法藏
643–712 ce), and finally, the third, the Chinese Chan School as represented by Huineng.
In the Tantai School, according to Zhiyi, the occurring mind has to deal with the dissolu-
tion of the illusionary self in terms of three efforts of the mind: it has to recognize or to have
an understanding of the true nature of Buddha or the Buddha nature, so that one could come
to actually experience it and then see it as it is. Following this understanding and vision or
knowledge of this nature, one can use one’s virtuous practices to achieve the Buddha nature.
Of course, it does not mean that there is an end outside for us to seek out. It means simply that
one can come to have a true mind not falling into illusions, desires, and traps and thus realize
the inner Buddha reality (the enlightenment) in oneself, as the inner truth of Buddha nature
has never separated from oneself, but is generally hidden from us in our being deluded by false
objects (or all objects as false).
Next, we come to the Huayan School’s approach to the transformation of self into the
realization of the Buddha nature. Again, under the strong influence of the Dacheng Qixinlun
(大乘起信论), Huayan focuses on how the true mind could activate and present the ultimate
reality implicit in the alaya consciousness, so that it will exhibit a world of non-obstruction of
principles and events in which no illusion is to be found. The basic idea is that one’s concrete
experience of mind at any time and in any place could invoke the true mind of the ultimate,
so that one can become liberated from the vagaries of mind and thinking. Huayan is able to do
435
Chung-Ying Cheng
this because it has taken the dharmadathu as the reality, so that the mind, once liberated from
illusions, will immediately and directly present the world of the ultimate dharma world which is
described as non-obstruction of events and laws. The question is: which is the initial factor, the
dharma world or the mind seeking emancipation? The answer is that it is the true mind which
never stops working at ultimate reality and yet it takes the opportunity or critical occasion to
rise up and shake off the misty cloud of desires and local perceptions so that the great world of
dharmadhatu will be presented.
The difference of the Huayan approach from the Tiantai approach is that in the Huayan
it is the true mind’s immediacy of presentation which reveals one’s enlightenment, whereas
the Tiantai approach requires that continued efforts be made to resist evil and to choose good
in one’s practice and action. In this manner, one may say that the Huayan approach is more
like the sudden enlightenment of the Chan school as taught by Huineng, whereas the Tiantai
approach is more like the gradual enlightenment process of the Shenxiu (神秀 606–706 ce), a
contemporary and co-disciple of the Fifth Patriarch Hongren (弘忍 601–75 ce) at the time of
Huineng (慧能638–713 ce).
Finally, we may ask what would be the unique approach of Chinese Chan? The answer is
that the Chan teaching is more like the Huayan’s than the Tiantai’s, in its approach to retrieval
of the Buddha nature of a person. Like Tiantai, Chan is strongly under the influence of the
Dacheng Qixinlun, in that it takes one’s self-nature as having the power to seek out the ultimate
reality of the self as Buddha nature. How to do it? Huineng, the founder of Chan Buddhism,
takes “straightforward meditation” (yixingsanmei 一行三昧) or the meditation with true mind in
any moment of life whether walking, dwelling, sitting, or sleeping. The so-called yixingsanmei
is to concentrate on removing false ideas and desires so that the mind remains unmoved and
ready for presenting or revealing the ultimate reality. Huineng describes his teaching as teaching
of no-thought as principle, no-form as main body, and non-abiding as root. He thus brings out
the analogy of lighting to enlightenment.
My description of the three Chinese Buddhist approaches shows that they share many things
in common; we may, for example, list the following five main points:
(1) They all take the human mind to be formed by perception and memory of things in the
world and that one’s interaction with things in the world are important for forming one’s
natural self. This natural self however is not simply determined by one’s actions in this
present life of myself but is also determined by karma from past lives. Yet none of the three
stress this later point, which is in sharp contrast with Indian Buddhism.
(2) All three presuppose the Yogacara theory of alaya consciousness as the base and potential
for all prior seven consciousnesses of the self, including the five aggregates of senses. All
three schools make it clear that the crucial problem of life is to seek a way of emancipa-
tion and dissolving one’s habitual self through the self-effort of self-conscious thinking and
knowing and action.
(3) The action of enlightenment is to be motivated by our self-nature which is supposed to
contain the ultimate nature of our being so that we can achieve our ends of enlightenment
if we make efforts from our self-nature or original mind.
(4) The action of enlightenment is not anything extraordinary but rather a natural and simple
practice of following precepts, meditation, and doing good things and refraining from
doing bad things.
(5) Once enlightened, one need not to be separated from this world nor this life but one could
lead an enlightened life of practicing the good of our nature and perhaps to continue to
enlighten other people.
436
If we wish to speak of the differences among the three typical Chinese Buddhist philosophies,
it can be said that the Tiantai approach wants to make it extremely clear and necessary that one
must have a vision of the ultimate and make crucial decisions in choosing good and not doing
evil. For the Huayan approach, however, the most important thing is to return back to the
original and ultimate reality through activation and inspiration of one’s internal nature. For the
Huayan, to go to the very beginning of one’s vision of the ultimate would have a great impact
on our understanding our identity as no-identity, and thus achieving our Buddha nature. In
this sense Huayua is the most extreme Buddhist school among all the Buddhist schools, both
Indian and Chinese, in striving to reach for the original and primary illumination of mind of
the historical Buddha himself. The Buddha nature becomes the display of the Buddha world of
events and laws in perfect order.
Finally, the source idea for the Chan Buddhists beginning with Huineng is strong, and this
idea of the source of one’s Buddha nature precisely points to the self-nature of oneself. Chan
stresses the practice of separation in non-separation and non-separation in separation, and there-
fore does not wish to reject one’s self-identity within the Buddha nature and vice versa. The
most essential point of the Chan approach to the human self is immediate and concentrated
practice and action and a determination to do this throughout.
7. Concluding remarks
In the above, I have traced the issues raised by our experience and understanding of memory
in light of a common-sense phenomenology of experience which provides a basis for reference
to the understanding of a philosophical reflection and explanation of memory from Indian and
Chinese Buddhist views on the self and non-self of the human person. We have seen that the
concept of self is tied up with the concept of memory in terms of the Buddhist philosophy,
whether Indian and Chinese. But I pointed out that the Buddhist philosophy of self is not
intended to answer questions of memory in the very beginning; instead we have to relate the
Buddhist views to the questions of memory from an analytical point of view based on a histori-
cal understanding of their development. I assert that memory in present-day understanding in
cognitive psychology is a good start for the study of the Buddhist philosophy of self and non-self
and the related Buddhist theories in turn provide a useful and meaningful theoretical hypothesis
for understanding of memory and its ontology.
In short, we come to see memory as a matter of consciousness based on perception, imagi-
nation, and reconstruction in the context of the deeper layers of consciousnesses. The most
interesting question is that memory is not the center of the Buddhist view; rather, the central
concern is how to interpret and eventually to get out of memory and memory-self. This means
that we have much to learn from the Buddhist views, particularly the Chinese Buddhist views in
dealing with retrieving identity and interpreting identity of the human self in general, in which
memory by experience can be seen to play a revealing role.
Notes
1 Here I mean that for making metaphysical explanations appeals are made to analogies and models of
events in nature. For example, the idea of “perfuming seeds” in Yogavara in Indian Buddhism and “origi-
nal face before birth” and “voidness of things” in Chinese Buddhism are basic for explaining relationships
of memory to human selfhood and vice versa.
2 A good example of this is found in the oral statement of Buddha himself after he became enlightened in
the Sanskrit discourse Majjhimankaya, translated in English as The Middle Length Discourses of the Buddha,
in Nanamoli and Bodhi (1995: 105), as quoted by Yao Zhihua in Zhihua (2008: 4).
437
Chung-Ying Cheng
3 See all kinds of Abidharma Sutras. Even in 大智度论卷22, CBETA T25 1509, .223 B3–10. This same
teaching was repeated in the Chinese Buddhist Sutra 妙法莲华经玄义, vol 8.
4 See later explanation in this chapter.
5 Cf. Huineng (1967: 171). I have made some changes in the translation.
6 See Philip Yampolsky’s translation of Liuzu Taijing in Huineng (1967).
7 See Fang (1991: 149), as based on Asanga’s Baifa Mingmenlun (百法明门论).What is called “loose conscious-
ness” (独散意识) is intended to refer to memory as basically having no determinate and well-connected
reference to things in the world or to other consciousness of the human self.This is the meaning we could
see if you wish to make sense of memory in comparison with other types of consciousness. Of course it is
not as solid as perception, but it is less free than dreams.
8 For this, one can consult Ricoeur (2004).
9 This text was said to be translated from Sanskrit by monk Paramartha (499–589 ce) in the year 553. The
work was attributed to an Indian author named Asvaghosa 马鸣 in Chinese. But the way of writing and
its content suggested that it was composed from a Chinese Buddhist author in that period of time and
has nothing to do with Paramartha.
References
Fang, L. (1991) Buddhist Philosophy (Fojiao Zhexue), 2nd edn. Beijing: Renmin University Press.
Nanamoli, B. and Bodhi, B. (trans.) (1995) The Middle Length Discourses of the Buddha: A Translation of the
Majjhima Nikāya. Boston, MA: Wisdom Publications.
Ricoeur, P. (2004) Memory, History and Forgetting. Chicago, IL: University of Chicago Press.
Huineng, L. (1967) The Platform Sutra of the Sixth Patriarch. Trans. Yampolsky, P. New York: Columbia
University Press.
Zhihua, Y. (2008) On Memory and Personal Identity. Journal of the Centre for Buddhist Studies, Sri Lanka,
VI: 219–34.
438
35
AUGUSTINE
Lilianne Manning
Grant me, O Lord, to know and understand . . . whether first to know you or call
upon you . . . I will seek you, O Lord . . . .
(The Confessions, 1.1.11)
1. Introduction
When Aurelius Augustinus Hipponensis (354–430 ce), referred to as St. Augustine or Augustine,
converted to the Christian doctrine, he set himself the aim of understanding God, although, as it
is explicit in the quotation above, he did not know where to start. The answer, ten books later
in his Confessions (Conf.), is that the only place to search for God is in his own memory:
I shall mount beyond this power of my nature, still rising by degrees towards Him who
made me. And so I come to the fields and vast palaces of memory, where are stored the
innumerable images of material things brought to it by the senses.
(Conf. 10.8.12)
The aim of the present chapter is twofold: (i) to describe some of Augustine’s personal char-
acteristics, which in interaction with his cultural context, may explain the modernity of his
writings on memory despite the interval of 1,600 years, and (ii) to develop the central com-
ponents of his theory of memory, taking into account that Augustine was interested in the
memory process, which he considered to be the unique method to know God.
In Section 1, his capacity of introspection and his personality are commented on against the
background of two philosophical currents, Neo-Platonism and Christianity (the only true philo-
sophical theme for Augustine). In Section 2, we see that the formation of mental images of the
external world is the fabric, so to speak, of recollections, but that this general memory process raises
important questions and some problems when Augustine wonders about memory-images of God,
emotions, and forgetting. The section ends with his notions about the self.
As mentioned above, Augustine reflected and wrote on memory not to know about memory,
not even to know about his own recollections, but because he was adamant that by remember-
ing his relation to God, he would be following the only way of knowing what a human being
can know about their maker.
439
Lilianne Manning
Besides knowing God, Augustine also aimed at both understanding the soul—bearing in
mind that he considered memory to be part of the soul—and awakening his fellow Christians’
thoughts and emotions towards God. His surviving works attest to how committed he was in
the pursuit of his objectives. His work includes many books, apologetic and exegetical works,
hundreds of lectures, letters, and sermons. Not surprisingly, the seventh-century theologian,
Isidore of Seville expressed his deep admiration, addressing him directly: “Augustine, he lies who
says he has read you entirely . . . for you Augustine, glow with a thousand volumes” (translated
version: Barney et al. 2006).
It has been suggested that the first precedent of the twentieth-century memory taxonomy
could be situated in book 10 of Augustine’s Confessions (Cassel et al. 2013; see this chapter’s
Conclusion), and that books 10 and 11 of the same work, contain the foundations of the cur-
rent notion of “mental time travel” (Manning et al. 2013; see the Conclusion). In the present
chapter, an attempt is made to focus on Augustine’s memory theory with no other reference to
the present day than the current reader’s impression of remarkable modernity when following
Augustine’s introspective conclusions about memory processes. The present text is decidedly set
to “listen” to the author in his context and from there, try and describe (a minute) part of the
results of his complex meditations in his search of God. Therefore, before tackling his concept
of memory, this chapter starts with a brief reminder of some of our author’s characteristics that
seem interesting to better contextualize his work on memory.
Among Augustine’s “thousand volumes,” in Isidore of Seville’s parlance, the Confessions and
most particularly book 10 are consensually acknowledged to represent, par excellence, his work
on memory. However, it is also well-known that from book 10 to the end of the Confessions
there are many issues that were Augustine’s topics of meditation in an earlier book, On the
Trinity. The present text is therefore largely based on passages from these two volumes, the
Confessions and On the Trinity (Trin.), which are thought to represent the work on memory in
Augustine’s theological vision of what it is to be human (Clark 2005).
2. The man behind the writer

Augustine’s biography is widely accessible: as a recent example, a brief but fairly comprehensive
list of his biographical facts can be found in Cassel and colleagues’ article (2013). The present
subheading briefly focuses on two personal features, crucial for his quest on memory: (i) his
capacity of introspection and the role played by Neo-Platonism in his self-analysis, and (ii) the
complementary characteristic, i.e., passionate (almost obsessive) way, in which he pursued his
introspective activity. It is possible that these personal features explain, at least partially, why
what he wrote on memory, 16 centuries ago, still pulsates with the imprint of modernity.
Due to human reason’s mutability, rational comprehension of God will always be partial,
stated Augustine. Intensely aiming at this partial knowledge of God, he decided that it could
be achieved by a correctly guided self-knowledge. This decision meant to rely entirely on the
understanding of his memory and, in that endeavor, his likely natural capacity and also his
training to carry out self-analysis were crucial. The influence of Neo-Platonism on Augustine’s
conceptions, in general, has been amply debated. Did he follow the Platonists’ teachings to use
and therefore understand his memory to find God? Augustine encountered Neo-Platonism
in the spring of 386, as he mentioned eleven years later, in the Confessions (Conf. 7.6.9).
Undoubtedly, Augustine shared with Neo-Platonism the notion of order and the hierarchical
places that God and the soul occupy in that ordered world. Importantly, however, Augustine
differs from the Greek tradition in his attempts to understand memory from his human perspec-
tive, and independently of any ontological aim. Despite this difference though, the influence of
440
Augustine
Neo-Platonism is noticeable in the means chosen to understand memory (O’Daly 1987). Both
Plotinus and Augustine advocate for the human being’s inner life. It is the interior man’s sense
that will yield the responses about God as Plotinus explained in the Enneads: introspection coin-
cides with contemplation of God. What seems uniquely Augustinian is the Christian perspective
of self-examination. The Confessions, particularly but not exclusively, were written as a speech
addressed to God, asking for help, and expressing full confidence in him. This allowed him to
reveal freely and naturally his inner life and self-awareness (Clark 2005), by means of scrutiniz-
ing his inner being.
It appears therefore that Augustine’s method to understand and eventually conceptualize
memory, resulted from both the influence of Neo-Platonism and the fact that his introspective
technique was grounded in the Christian tradition, and more precisely, rooted in the contents
of the Psalms, which were ubiquitous in his everyday life after his conversion.
The role of the Psalms (being a powerful declaration of loving God), in interaction with his
capacity of introspection, enhanced Augustine’s naturally passionate character, and this combi-
nation probably holds an important key to understanding the context in which he achieved such
an imposing work on memory.
At the age of 19, after having read Cicero’s Hortentius, Augustine commented (in De Beata
Vita) how intensely he desired to devoting himself to the learning of wisdom (philosophy). He
remembers again this first passion for wisdom, in the Confessions (3.4.8): “How did I then burn,
my God, how did I burn to wing upwards from earthly delights to You.” By the same token,
an important transformation is manifested: the love for wisdom of his youth is amalgamated in
this passage of the Confessions, with his love for God.
Finally, aside from his most salient personal features, it is interesting to note Augustine’s use
of the metaphor as being his literary style to convey his meditations. In this characteristic style
of the ancient Roman writers, his metaphors to explain memory are rich and varied; sometimes
they are spatial and architectural, evoking palaces, halls, vast courts, avenues, or fields, and some-
times they are associated with the human body. Related to the latter, memory is seen as the
stomach of the mind or the mind’s eye, and the will’s eye to direct attention in order to retrieve
memories (see “Mental images of emotions” in Section 3).
In summary, his outstanding capacity of analysis, including self-analysis of his sensations, per-
ceptions, thoughts, and memories, likely enhanced by long-standing Neo-Platonist notions, and
his passionate personality, highlighted by the daily manner in which Christians praise God, were
the bases to construct the first model ever of human memory, amply using the metaphoric style.
3. Augustine’s conceptualization of memory

As it was mentioned above, it would be a mistake to think that Augustine wrote hundreds of
pages because he was “interested” in memory. He was not. What he deems important is not
what he remembers from his past, but how he remembers the past (Clark 2005). The mental
activity of remembering paves the way to reflect on our existence in relation to God, and
how reflecting in that manner leads to knowing God. This notion is his guiding theme, and
it was already explicit in his Soliloquies, as the central subjects of the two books that form the
volume: knowing his creator and knowing his own soul. Interestingly, O’Daly (1987) remarks
that the dichotomy of these two kinds of knowledge is only apparent due to the influence of
the Neo-Platonist notion that knowing one’s own mind is tantamount to knowing its source.
(A different matter is the a priori knowledge of God, which is commented on below.)
The theoretical frame of Augustine’s concept of memory is that memory that is able to
know God is part of the intelligent soul (intelligent memory), and different from the sensitive
441
Lilianne Manning
soul (sensitive memory). Augustine makes this distinction explaining that the latter (anima), is a
vital and inferior part of the soul (animals have anima), while the former (animus) is the rational
superior part of the soul, exclusively present in the human being.
In On the Trinity (10.10.13), it is said that animus comprises the superior form of memory,
together with the faculty of understanding and the will. Knowledge and science of many things
are contained in memory and understanding, he stated, while will allows us to enjoy or use them.
Memory, operating as the link between the senses and intellect, is the basis of Augustine’s
reflections on the whole processes of storing images, remembering, and forgetting. The start-
ing point is, therefore, to consider sensory perception and the workings of memory to use the
products of perception.
Mental images
In Augustine’s terms, the representations of the physical world are mental images formed thanks
to the senses and stored in our memory. Moreover, to the countless images brought by the
senses, Augustine adds the memory’s own imaginative creation. He illustrates this notion in a
passage from On the Trinity (11.8.13), in which he describes the recollection of having seen only
one sun together with the possibility of multiplying this memorized unique image in order to
“see” two, three, or more suns, and memorize the images.
We think owing to memory-images; they are the necessary link between the objects or
events themselves, and the objects or events as being perceived. However, although memory for
the physical world corresponds invariably to perceptions, we need our conscious will to form
the mental images and retain them:
for it happens even when reading . . . that I have read through a page or a letter and did
not know what I was reading, and so had to read it again. For when the attention of
the will is centered on something else, then the memory is not so applied to the sense
of the body as the sense itself is applied to the letters.
(Trin. 11.8.15)
When the perceived object is withdrawn, he says, its image “remains in the memory, to which
the will may again turn its gaze in order to be formed by it from within, as the sense was formed
by the sensible body that was presented to it from without” (Trin. 11.3.6): this results in the
individual recollecting, retrieving memories. The “will turning its gaze” triggers recollections,
which implies that we cannot recollect without wanting to recollect, but willingness cannot
suffice, if the image is not there any longer, no recollection will take place. But what triggers
the will to direct its eye to some place in the mind to direct attention and retrieve memories?
Augustine’s response is that when people say that future events are seen, it is not the events
themselves, but their signs that are seen. Those signs are present “and from them things to come
are conceived by the mind and foretold” (Conf. 11.18.24). What triggers the will to direct atten-
tion in one way or the other seems to be expectation based on present signs. In attentive actions,
intentional will would be the directional path between memory and expectation.
Recollection is nothing else than the perception of mental images. However, mental images
cannot be called true. In fact, they differ from perception in an essential point, which is that only
the objects can be perceived by other than the perceiver himself. It follows that memory-images
and therefore recollections, being impossible to be verified as real objects, are less reliable.
Moreover, remembering does not imply a direct contact with the past event and therefore
memories of past events can be distorted. However, memory distortions are not lies, they have
442
Augustine
evidential solidity, but only for the rememberer. (A passage in Trin. 15.9.16, describes these
distortions. They are called “false memories” in current neuropsychological literature and are
the subject of intensive research and abundant publications.)
We may have the impression that perceptions are more reliable than the images formed
through them, since the external object can be verified. Can perceptions be called true?
Objects of sight, which are presented from without, where the eye may be deceived,
as it is deceived when it sees the bent oar in the water, and when the navigators see the
towers moving, and thousands of other things which are otherwise than they appear.
(Trin. 15.12.21)
Our senses can also be mistaken, though in a lesser degree than the mental images that we
remember.
Memory-images are neither a precise nor an accurate reproduction of the external reality;
they are rather the result of reconstruction.
Are there mental images of God?

Only the things that always exist, as eternity itself, he says, do not need mental images. Augustine
asks God where, in what quarter of his memory he is, since he, Augustine, has searched through
his body images, affect recollections, and passions, he has searched even in the memory of his
memory as far as memory remembers itself, and did not find his image. God is not an image.
However, if he were not in his memory, Augustine could not search for him (Conf. 10.25.36).
He wonders where did he get the knowledge of it (it being “the happy life,” beata vita, which
is God), that he should so desire and love it? God must be in his memory . . . but how, he
does not know (Conf. 10.20.29). Augustine compares the images of a series of different non-
corporeal things such as numbers, eloquence, or emotions with remembering the happy life to
conclude that only God has no images in the memory (Conf. 10.21.30).
According with these comments, Augustine’s basic concept of memory, as mental images
created through senses and recollections made possible through the will, is not directly related to
his search for God. This observation leads him to one of his most complex propositions: divine
illumination. In fact, to explain a priori knowledge and the presence of God in his memory, he
seems to adhere to Plato’s theory of reminiscence. However, considering other passages of On
the Trinity, it can be seen that Augustine clearly refutes Plato’s theory. Thus, in On the Trinity
(12.15.24), he starts by contesting Plato’s example describing the slave-boy’s knowledge of
geometry (in the Meno). In Augustine’s view, rooted in his context of Christianity, it is not a
previous life experience that could account for reminiscence, but “a sort of incorporeal light of
a unique kind” (Trin. 12.15.24). It is divine illumination that acts in memory, and memory is
viewed as desire of the happy life (in opposition to remembrance of past lives). The happy life
being the knowledge of truth, Augustine’s elucidation appears qualitatively different from Greek
explanations of reminiscence. By the same token, Augustine’s meditations on memory in this
perspective, make it obvious that his theory should not be conceived as a simple palace of the
mind, a storehouse of a great variety of items. The palace of the memory would apply to mental
images of objects and the contents of teachings such as language, numbers, laws, etc., but not
to his ultimate goal. His model of memory should be seen rather as having at its core the rela-
tionship between the knowledge of God, through divine illumination, and the knowledge of
his own memory, through the introspective method applied to sensations, perceptions, images,
and recollections. To go a step further, it has been stated that self-knowledge and knowledge of
443
Lilianne Manning
God stand in a dialectical relationship (Crouse 1999). Memory turns inwards to itself, in order
to ascend to the knowledge of God; however, that knowledge, in turn, involves a transforma-
tion of the self-identity.
Mental images of emotions

We do not necessarily relive an emotional state, since “I remember past fears without fear, and
past desire without desire” (Conf. 10.14.21). However, what Augustine wishes to understand
is how those emotional recollections are recalled, since they have lost precisely the quality by
which they entered the memory. The “representations” of emotions must be in our memory,
owing to the fact that at some point we felt the emotions themselves. Recollecting emotional
states must follow the same processes than recollecting other types of contents. Although not
physically present, emotions have mental images. In a passage of the Confessions (10.15.23), he
explains how he knows that the memorized image of an affect exists and is accessible, despite
the fact that the initial feeling is absent. When he names “pain” without feeling pain, he does so
thanks to the storage of an emotion of pain; otherwise, he says, he would be unable to distin-
guish pain from pleasure. However, this account is not sufficient to understand how emotions
are represented in our memory and Augustine’s attempts to clarify this difficulty resulted in one
of his most astonishing metaphors, namely, memory seen as being the stomach of the mind.
He thought that, most particularly regarding the recollection of emotions, it was necessary to
consider some sort of “digestion.” Both the stomach and the memory store food and memories,
respectively. Bitter and sweet food and affects of joy and sorrow are stored but they no longer
taste/feel. Augustine’s metaphor allowed him to solve the problem of how we can think of our
emotions without reliving (“tasting”) the emotions themselves, and how different and opposed
emotions can be present simultaneously (Vaught 2005).
Forgetting
The problem of forgetting is that accepting the fact of having forgotten implies remembering it:
Are we to understand from this that when we remember forgetfulness, it is not pre-
sent to the memory in itself but by its image: because if it were present in itself it
would cause us not to remember but to forget? Who can analyse this, or understand
how it can be?
(Conf. 10.16.24)
Remembering that we have forgotten is comparable to recollecting emotional states in that

the reason itself of the recollection, why it is recollected, is forgotten or not relieved. Thus,
as commented on above, the emotion itself that originated the emotional state is absent
in recalled emotions since the person may be sad while recollecting happy moments, and
the thing itself is absent in forgetfulness when remembering that we forget. However, the
problem is more difficult to resolve in the case of forgetting, since the presence “through
its image” solves the problem for emotions, but does not solve it for forgetting. The reason
is that forgetting, unlike emotions, does not have an original presence, no “original” thing
could have existed to create at some point certain traces in the mind.
In the case of forgetting, he is confronted with a double problem: how to explain the image
of forgetting itself, and since this seems impossible, how to account for remembering forget-
ting. Using solely his introspection, these problems seemed unsolvable. Therefore, bearing in
444
Augustine
mind that Augustine’s meditations on this topic are meant, as are all his reflections, to lead him
to know God, he resorted to a theological reasoning point of view. Augustine concluded that:
by forgetting God, as though forgetting their own life, they had been turned into death,
i.e., into hell. But when reminded they are turned to the Lord, as though coming to life
again by remembering their proper life which they had forgotten.
(Doctrinal Treatises 13.13.17)
What is important about forgetting is that one kind of forgetfulness means death.
Memory and the self

Our identity is based on what we remember of ourselves, and also what we know of ourselves
with no particular recollection of a memory (e.g., I know I am a kind/aggressive person, but don’t
remember the events that formed that knowledge). What we know and what we remember about
ourselves extends in time because, owing to the ephemeral and fleeting nature of the present, our
self is grounded in our past and future. Memory becomes in Augustine’s construction, the pivotal
entity through which we can think meaningfully of the self across time, i.e., phenomenological
continuity: “The present of things past is memory, the present of things present is sight, the present
of things future is expectation” (Conf. 11.20.26).
Augustine states that whatever is in the memory it is in the mind, the mind is what we are,
our self-identity, which implies being conscious of ourselves. Having established this sequence,
Augustine asks what is the self or what the meaning is of the personal identity of which we are
conscious:
Great is the power of memory, a thing, O my God, to be in awe of, a profound and
immeasurable multiplicity; and this thing is my mind, this thing am I. What then am I,
O my God? What nature am I?
(Conf. 10.17.26)
The following quotation could be seen perhaps, as an indirect answer to these questions: “It is
an inner knowledge by which we know that we live” (Trin. 15.12.21). It is worth recalling here
that for Augustine, to be alive and to have a soul are tantamount. Although the soul cannot be
perceived by the senses, our constant awareness of its existence is the awareness of being alive:
“For what is so intimately known, and what knows itself to be itself,2 than that through which
all other things are likewise known, that is, the soul [animus] itself?” (Trin. 8.6.9). Moreover,
when the mind knows itself, it does so because it is present to itself, in the present moment (see
Trin. 10.10.12). This is the memory in which the subject knows himself, and from this memory
he visits his own past, and projects into future actions, events, and expectations. More impor-
tantly for Augustine, from this memory the creature can address his creator and construct his
relationships with him across time.
In summary, Augustine’s central notion of the self is that its formation is possible thanks to
the recollection of our experiences and thanks to the fact that we are conscious of our self-
knowledge. This is possible only in the extended temporal dimension that becomes present at
any given moment, therefore the importance of memory. We are no longer who we were, but
we remember who we were and therefore who we were in the past becomes a present represen-
tation of ourselves. We are not yet who we hope to become, and again thanks to the memory
445
Lilianne Manning
of ourselves, expectations are present. Ultimately, as the most basic expression of the memory
of our self, it is defined by the knowledge of being alive, which is equivalent to having a soul,
and the action of the mind that recognizes itself as itself.
4. Conclusion
The present chapter suggests that Augustine’s works on memory result, among other factors,
from an outstanding capacity for introspection and a passionate pursuit of his own memory
examination. The latter, put to the service of God, guided the former. In this context, the influ-
ence of the Neo-Platonist doctrine modulated probably the selection of his central topics. A
cultural specificity, the use of metaphor, was widely practiced in the writings of this period, but
never in this manner. In fact, Augustine’s writing style is considered a turning point in Western
literature (Clark 2005), owing to the fact that for the first time ever, the author writes about how
he sees himself to himself; the style is entirely personal, emotional and relational. Augustine’s
personal features and literary style—applied to the thorough and meticulous description of
memories and reflections of memory—resulted in the brilliant work that captivates us today,
and gives us the disorienting impression of modernity and freshness.
Augustine’s theory of memory states in the first place, the formation of memory-images
brought by the senses, that is, the sensible memory. The introduction of the Christian notion
of divine illumination opposing that of previous life experiences, explains why there are no
memory-images of God. Moreover, he tackles subtle questions such as the recollection of
emotional states and memory-images of emotions, and that—unresolved—of forgetfulness
and recollection of forgetfulness, for whose explanation he adopted a theological perspective.
With regard to the self, his reflections produced a coherent, articulate, and far-reaching
description of phenomenological continuity and the necessary kind of memory to be closer
to the knowledge of God.
His powerful mind, through the pursuit of moral excellence, prepared for nineteenth-century
psychology as the science of consciousness (Klein 1970), and foreshadowed the twentieth-
century neuropsychology of memory (Cassel et al. 2013). The latter authors hypothesized that
Augustine’s work on memory was the first complex categorization proposal. Although he did
not aim at organizing memory systems, describing different types of memory meant ipso facto the
construction of the first taxonomy of memory. Moreover, Augustine’s descriptions are close to
those of modern neuropsychology initiated by Tulving (1972).
Finally, Manning and colleagues (2013) proposed that Augustine’s work, due to the uni-
versality of its contents, i.e., memory and time, heralded the twenty-first-century notion of
“mental time travel” (Suddendorf and Corballis 1997; Tulving 1985). Augustine’s medita-
tions on the continuity of the self are close to ours and their influence on our way of reflecting
on subjective time must have been important, though never acknowledged.
To conclude, it can be said that although Augustine wished to know only God and his
own soul, nothing more; in the end, his aims meant aspiring to know everything about human
memory.
Notes
1 Augustine’s citations indicate the name of the work, and the book, chapter and subheading numbers.
2 In 2000, Gallagher suggested that even if all of the unessential features of identity disappear, we still know
and feel that there is a primitive and immediate “something” that we are willing to call a self.
446
Augustine
References
Augustine of Hippo (1910) The Soliloquies. Trans. Cleveland, R. E. Boston, MA: Little, Brown & Company.
Augustine of Hippo (1944) De Beata Vita: The Happy Life. Trans. Brown, R. A. Washington, DC: The
Catholic University of America.
Augustine of Hippo (2002) On the Trinity: Books 8–15. Trans. McKenna, S. Ed. Matthews, G. B.
Augustine of Hippo (2006) Confessions. Trans. Sheed, F. J. Ed. Foley, M. P. Intro. Brown, P. Second edn.
Indianapolis, IN: Hackett Publishing.
Augustine of Hippo (2007) Doctrinal Treatises. In Schaff, P. (ed.), Nicene and Post-Nicene Fathers: First Series
Vol. III. New York: Cosimo Classics.
Barney A. S., Lewis, W. J., Beach, J. A., Berghof, O. (2006) The Etymologies of Isidore of Seville. Cambridge:
Cassel, J. C., Cassel, D., Manning, L. (2013) From Augustine of Hippo’s memory systems to our modern
taxonomy in cognitive psychology and neuroscience of memory. Behavioral Science, 3: 21–41.
Clark, G. (2005) Augustine: The Confessions. Bristol: Phoenix Press.
Crouse, R. (1999) Knowledge. In Fitzgerald, A. and Cavadini, J. (eds.), Augustine through the Ages: An
Encyclopedia. Cambridge: William & Eerdmans Publishing Company.
Gallagher, S. (2000) Philosophical conceptions of the self: implications for cognitive science. Trends in
Cognitive Science, 1: 1421.
Klein, D. B. (1970) The earlier medieval background. In Klein, D. B. (ed.), A History of Scientific Psychology.
London: Basic Books, pp. 124–58.
Manning, L., Cassel, D., Cassel, J. C. (2013) St. Augustine’s reflections on memory and time and the cur-
rent concept of subjective time in mental time travel. Behavioral Science, 3: 232–43.
O’Daly, G. (1987) Memory. In Idem, Augustine’s Philosophy of Mind. Berkeley, CA: California University
Press, pp. 131–50.
Suddendorf, T., Corballis, M. C. (1997) Mental time travel and the evolution of the human mind. Genet.
Soc. Gen. Psych., 123: 133–67.
Tulving, E. (1972) Episodic and semantic memory. In Tulving, E., Donaldson, W. (eds.), Organisation of
Memory. New York: Academic Press, pp. 381–403.
Tulving, E. (1985) Memory and consciousness. Canadian Psychology, 26: 1–12.
Vaught, C. (2005) Access to God in Augustine’s Confessions. Books X to XIII. New York: State University
of New York Press.
447
36
AVICENNA AND AVERROES
Deborah L. Black
1. The conceptual and historical framework:

the internal senses
Avicenna and Averroes developed their accounts of memory within the broader framework
of their theories of the internal senses, a cluster of faculties localized in the brain which were
posited to explain a variety of sensory operations. The internal senses originate as an attempt
to refine and systematize Aristotle’s account of imagination (phantasia) and to harmonize it
with later developments in medical theory stemming from the physician Galen. As Avicenna
observes, “What [Aristotle] brings together here under the term ‘imagination’ can be divided
into a number of active powers, such as estimation and cogitation, and retentive powers, such
as the formative faculty and memory.”1 Yet the theories developed under the rubric of the
“internal senses” also include many new elements not present in Aristotle’s psychology. One
significant source for many of these developments was the Arabic version of the Parva natu-
ralia, which is not a unadulterated translation of Aristotle, but an adaptation that incorporated
materials from Neo-Platonic sources.2 On the basis of these diverse sources, Avicenna and
Averroes developed their own competing accounts of the nature of memory.
Two features of the internal sense tradition in particular are of significance for understanding
these accounts of memory. The first is the ambiguity that arises from the fact that Avicenna and
Averroes use the term “memory” (dhikr/ḥif ) for a particular internal sense faculty within their
systems. But this faculty is not responsible for all of the operations that we would now consider
to be forms of memory, and both Avicenna and Averroes also use these same Arabic terms to
denominate a variety of cognitive activities which involve the cooperation of many internal
senses. So to understand their theories of memory, we must draw on features of the entire inter-
nal sense tradition, and not just on their accounts of the memorative faculty.
The second is that one of the key concepts from the internal sense tradition is also funda-
mental to both Islamic philosophers’ theories of memory, namely, ma nā (pl. ma ānī). This term
literally means “meaning” or “idea,” but it has traditionally been translated as “intention,” in
keeping with its standard medieval Latin rendition.3 Unfortunately this term is notoriously
equivocal, although the various meanings are linked to one another. Within the broader cogni-
tive psychology of classical Islamic philosophy, the term “intention” has both a generic and a
specific meaning, akin to the English term “perception.” “Perception” can be used for any act
448
Avicenna and Averroes
of cognition, whether intellectual or sensory, but it is often used simply for sense perception
in contrast to other types of cognition. In a similar way, “intention” can be used generically to
mean the object of any cognitive act, as is reflected in our current use of the term “intentional.”
But in the context of the internal senses, an intention is the object of the internal senses of
estimation or cogitation and memory. This special sense of ma nā is found in the Arabic Parva
naturalia itself, but Avicenna and Averroes each adapt the term to their own theories. To under-
stand what memory is for both these authors, then, we will need to understand their respective
accounts of what an intention is.
2. Avicenna
Memory in Avicenna’s system of internal senses

Avicenna establishes his system of the internal senses by appealing both to empirical evidence and
to a variety of principles that reflect the explanatory function that each faculty is meant to serve
in his psychology. Two of these principles have a direct bearing on his view of memory. The
first is the claim that receiving and comprehending sensory information is a distinct operation
from retaining and recalling it. Since faculties are principles of action, retentive faculties, such as
memory, must be distinct from receptive or perceiving faculties.4 The second principle is a basic
Aristotelian one: that different types of objects must be cognized by distinct faculties. Avicenna’s
application of this principle is novel, however, since it involves his understanding of “intentions”
as cognitive objects distinct from the sensible qualities or “forms,” that are immediately available
to the five external senses. Avicenna identifies these intentions as properties of sensed objects
that escape the perceptual capacities of the five senses because they are not essentially material.
Nonetheless, these intentions are not abstract universals, since they are always represented as
features of concrete, sensible particulars. Avicenna’s favorite example of an intention, which
became ubiquitous in the West, was an animal’s instinctive perception of features of its environ-
ment necessary for its survival, such as the sheep’s perception of the wolf as a hostile predator.5
But Avicenna extended the notion to apply to any perceptual objects that could not neatly be
classified as proper or common sensibles,6 and he allowed that humans have sensory as well as
intellectual perception of such intentions.
Avicenna envisaged the internal senses as a set of two perceptive and retentive pairs,
corresponding to the two types of sensory objects, forms and intentions. In addition to these four
faculties, Avicenna also posits an active faculty of imagination which is able to create new images
from what has been stored in the two retentive faculties. Thus we end up with five internal
senses:
(1) the “common sense,” which receives and coordinates the perceptions of the sensible forms
from the external senses;
(2) the retentive imagination,7 or “formative” faculty, which is the storehouse for the sensible
forms received by the common sense. Imagination thus functions as the most basic form of
sense memory;
(3) the estimative faculty (wahm) which is responsible for the perception of intentions;
(4) the memorative faculty (dhikr/ḥif ), the storehouse for these intentions;
(5) the compositive imagination (al-mutakhayyilah), the active faculty which combines and
divides intentions and forms. When it is controlled by reason in humans, it is called instead
the “cogitative” or “thinking” faculty (fikr).
449
Deborah L. Black
In the narrowest sense, then, memory (4) for Avicenna is the faculty for storing intentions;
slightly more broadly construed it consists of both retentive faculties, imagination (2) and
memory (4). But when memory is understood to include the full range of activities that
involve retaining information, recalling it, or restoring it, all of the internal senses play an
explanatory role.
“Working memory”: perception versus retention

At first glance, one might think that the faculty of common sense would play no role in
Avicenna’s account of memory. After all, the common sense is simply the capacity to collate
and distinguish the information supplied by the five senses while the object is being observed.
Yet Avicenna argues that in the act of perception itself, the common sense displays a capacity for
something akin to what we now call working memory. Avicenna establishes this point through
an analysis of the conditions necessary to explain how we can perceive a raindrop tracing a line
and then a circle on some surface as it falls:
consider the state of the raindrop which is seen as a straight line, and then the state of the
straight line that rotates, whose end is seen as a circle. It is not possible for a thing to be
perceived as either a straight line or a circle unless it is seen several times. The external
sense cannot see it twice, for it sees it where it is. Once, however, it is impressed in the
common sense and ceases before the form in the common sense is erased, the external
sense would have perceived it where it is, whereas the common sense would perceive it
as if existing where it had been and as existing where it has come to be, thereby seeing
either a circular or straight extension. This, however, cannot be related to the external
sense at all. As for the faculty of representation, it perceives both things and forms a
representation of them even when the thing ceases and is absent.8
On Avicenna’s view, the external senses are only able to perceive objects that are actually
physically present. Since each act of seeing the raindrop involves a discrete affection of the
eye, the repetition necessary for continuity cannot be explained by the external sense itself.
What is needed is some way to connect the successive perceptions of the raindrop’s position
at each instant. The common sense, as the faculty in which the traces of the external senses are
impressed as they are being received, collects these diverse perceptions to form unified wholes.
A similar account of something like working memory is found in Augustine, who observed
that in order even to hear a poem or piece of music, memory must link the individual sounds
together.9 Yet unlike Augustine, Avicenna does not consider this linking activity a form of
memory, but rather, part of the perceptual process itself. Indeed, the whole point of this
example is to differentiate true retention from the ability to hold on to successive sensory
impressions. While the lingering of these impressions in the common sense ensures the continu-
ity of perception, the common sense is still dependent on the continued presence of the object.
If it does not capture the fleeting impressions as they pass, it will not perceive the line or circle
they form. The common sense, then, is not retentive because its objects must still be present to
it; it simply prolongs reception by providing continuity and flow.
The retentive faculties

As we have seen in our overview of the internal senses, there are two purely retentive facul-
ties in Avicenna’s psychology: imagination, which preserves the proper and common sensibles
450
perceived by the external and common senses, and the memorative faculty, which preserves the
intentions that are perceived by the estimative faculty. The perceptual content stored by the
imagination is fairly straightforward: it comprises any images we have of proper sensibles, such
as our remembering a particular shade of purple, a musical tune, or the scent of a skunk, along
with those of the common sensibles, such as our images of triangles or our ability to recall the
feel of a breeze moving the air. But a bit more needs to be said about how Avicenna under-
stands the intentions that are the objects of the estimative faculty and stored in the repository of
memory. As noted above, Avicenna’s best-known examples of intentions are properties related
to the instinctual appetites of non-human animals, such as the sheep’s perception that the wolf is
hostile. But intentions comprise much more than this, and it is likely that Avicenna introduces
his account of intentions with such examples because they illustrate the basic motivation for
recognizing intentions as distinct components of sensory objects. Since these affective proper-
ties are neither proper nor common sensibles, and yet animals perceive them, they cannot be
intelligibles, but rather they must be the object of another sensory power, which Avicenna calls
estimation (wahm).
Intentions nonetheless cannot simply be identified with affective properties like the wolf’s
hostility. Avicenna characterizes an intention as any object that is perceived concomitantly with
a sensory quality while not being “essentially material” in its own right.10 In this way intentions
also account for the various types of perception that Aristotle called “incidental,” such as “see-
ing” Diares’s son or that the yellow liquid is sweet. Avicenna’s conception of the estimative
power itself is more complicated still. While he argues for its existence based on the discovery
of intentions as perceptual objects, the fundamental role of estimation is as the faculty of sensory
judgment which directs the activities of all the sense powers: “Estimation is the greatest judge
(ḥākim) in the animal, for it judges by way of an imaginative arousal (inbi āth takhayyulī) without
this being verified.”11 That is, the judgment of estimation is non-propositional and pre-rational,
consisting merely in the recognition of some sensory object as accompanied by an intention.
Thus in the case of Avicenna’s standard example, the sheep uses the intention of hostility to
judge that the wolf is harmful and this arouses a flight response. As a general rule, then, when-
ever an animal (or a human making a pre-intellectual judgment) acts in some way on the basis
of his perception, the intention will be whatever cognitive object must be posited, in addition
to the basic sensory information it has, to explain its reaction.
Remembering and recollecting

Avicenna often describes the faculties of imagination and memory as treasuries or storehouses
(s. khizānah) insofar as they perform a purely retentive function.12 But when we call the store-
house of intentions “memorative,” we indicate the readiness or disposition that faculty has to
stabilize or represent its contents, or to retrieve them after they have been lost.13 “Memory”
then primarily connotes the ease with which a perceiver is able commit an object to one of the
storehouses of memory and make use of its contents at will. Avicenna gives the most attention
to the third activity of recollecting or retrieving a lost memory, which involves a “desire and
seeking” of lost memories that Avicenna takes to be uniquely human. Non-human animals do
remember things on Avicenna’s view, but they have no voluntary control over their memories.
As Avicenna puts it, “if they remember, they remember”; if they forget, they would have no
desire to seek what they have forgotten. Avicenna’s main reason for thinking that recollection
is purely human is that it entails the ability to draw inferences that something one once knew
has vanished, and inferential reasoning is a rational capacity. Nonetheless, even in the case of
recollection, the primary psychological mechanisms in Avicenna’s account are wholly sensory,
451
Deborah L. Black
since they are ultimately reducible to the judgmental functions of the estimative faculty—albeit
estimation “embellished with reason.”14
Avicenna describes the process of recollection as follows:
[Recollection] occurs when the estimation applies itself to its imagining faculty and
begins to make the forms in the imagination appear one by one, so that it is as it were
observing the things whose forms they are. If the form which it perceived along with
the intention that has disappeared occurs to it, the intention will then appear to it as it
once appeared to it from outside, and the retentive faculty would stabilize it in itself as
it had preserved it, and thus one would remember. It may [also] be that the progression
is from the intention to the form, in which case the remembered thing sought would
not be related to what is in the storehouse of retention, but to what is in the storehouse
of the imagination.15
In this account of how the process of retrieving lost memories works, Avicenna’s splitting of
sensory objects into two main components—sensory forms and intentions—is key. Avicenna
views remembering as an activity that usually begins with a partial recall of either the form or
the intention, which then provides the impetus for reconstituting the perception in its entirety.
Avicenna outlines two scenarios to parallel the two possible starting points: (1) one can begin
with a physical description—a cluster of sensory forms—and seek the intention to which it
belongs; or (2) one can begin with an intention and seek its proper description.
Some concrete examples will help to understand what Avicenna has in mind.16 While
Avicenna doesn’t explicitly identify the intention with the individual here (as Averroes later
will), his account implies that the intention often represents a specific individual as the remem-
bered object. In the first scenario, one begins by imagining the sensory qualities of some past
perceived object so as to reproduce the experience of direct observation. Once enough images
have been collated so as to make a sketch of the forgotten object, its intention becomes reunited
with it, and the overall representation of the thing with its sensible properties is available to be
committed to memory again. For example, I might look at a photograph—a depiction of sensible
qualities—and try to remember whose picture it is—the intention. Is it a picture of Mary or of
Jane? In the second scenario, I would begin with the “intention,” and attempt to reconstruct its
sensory appearance. For example, suppose I am thinking about my high school classmate named
Marianne: I have Marianne’s “intention” but I can’t remember what she looked liked, so I try to
reconstruct her sensory appearance. I may be able to access my stored images of her, but if not, I
would need to re-experience her, perhaps by getting out my old yearbook. In that case, a “return
to sensation,” as Avicenna puts it, would be required.
This account of how remembering works also helps to explain why Avicenna thinks that we
need to posit two separate storehouses for forms and intentions. After all, even if one accepts
Avicenna’s argument that we need a special faculty to perceive intentions, one might wonder
why we would need to store them in a different place from the images that originally accom-
panied them. The answer lies in part in Avicenna’s account of how a multiplicity of faculties is
compatible with the unity of perception. Avicenna argues that when there is evidence of either
interference or independence in the various activities of a living being, those activities must
issue from distinct faculties which are nonetheless united by a single nexus. If I can do x without
doing y, then x and y must flow from different principles or faculties. But if x and y can also
interfere with each other, then they must both be rooted in an ultimate unifying principle, my
soul or self.17 This suggests that for Avicenna, if one can remember a sensible quality or physi-
cal description while forgetting the intention that goes with it, the form and the intentions are
452
retained by different faculties, whose contents are available to a higher unifying faculty, either
the estimative, or, in the human case, reason.
Animal memory and experience

While recollection is a distinctively human form of memory, Avicenna does think that other
animals too remember. In their case, however, this happens only when there is a natural trigger;
they do not actively seek to recollect what they have forgotten. Thus my cat may hear the sound
of a can opening and remember its association with the “intention” of eating; but if my cat is
hungry, he simply seeks out whatever sounds and smells directly indicate the presence of food
to his senses—he cannot try to remember what colors and shapes describe the tin of cat food.
Despite these restrictions, Avicenna nonetheless holds that animals also have the ability to
accumulate the multiple particular observations stored in their memory and thereby acquire
something akin to experience (tajribah).18 Avicenna explicitly recognizes the way that animals
adapt to their environments, a process which he interprets as one of acquiring “intentions”
beyond those that they know naturally or by instinct. This happens when they retain the mem-
ory of a past association between a set of sensory qualities belonging to some object and the
pleasant or painful effects they suffered from that object. Thus, as Avicenna notes in the follow-
ing passage, dogs will learn to fear objects that have been used to beat them even though they
are not naturally or inherently harmful:
Another division [of animal estimation] pertains to something akin to experience.

This is such that when pain or pleasure befalls an animal or if there comes to it a sen-
sory benefit or a sensory harm connected with a sensible form, whereby the form of
that thing and the form of what is connected with it are imprinted in the formative
imagination, and the intention of the relation between the two and the judgment
concerning it are imprinted in the memory, then the memory in itself and by its
inborn nature will attain this [intention] . . . For this reason dogs fear mud, wood
and other things.19
This account of animal experience also helps us to explain what Avicenna means by saying
that remembering requires re-establishing the relation of the intention to the image. In cases of
natural estimative judgments (e.g., the sheep’s perception of the wolf as hostile), the intention
and form are perceived simultaneously and each stored in its respective storehouse; the relation
between the two is inbuilt. In cases of animal experience, the animal repeatedly experiences
some pleasure or pain along with a sensory image, causing the estimative faculty to associate the
two. The relation produced by that repeated association thus becomes an additional intention
that can be stored in the memory. When the animal encounters that same sensible form again,
it remembers the past intention associated with it, and reacts accordingly.
Memory, recollection, and intellect

One of the positions for which Avicenna was often criticized by Western medieval authors was
his claim that there can be no memory in the intellect. This view rests in part on arguments set
forth in Avicenna’s broader account of how abstract knowledge is acquired, which is outside
the scope of our concerns. Avicenna’s basic argument for his point does, however, bear upon
his general account of memory, since it relies on the principle that the perception or actual
consideration of an object, and its retention, are mutually exclusive. In the case of the senses,
453
Deborah L. Black
this principle leads to the positing of dedicated retentive faculties with their own organs in the
brain. But there can be no such dedicated retentive faculty within the intellect for storing non-
occurrent thoughts, since on Avicenna’s view, the intellect is an immaterial power which does
not use an organ. So there is no physical “place” to put concepts when we are not thinking of
them. Any abstract intelligible in the intellect must be an actual object of thought. Intellectual
memory must be explained, then, not in terms of storage, but dispositionally. To learn abstract
universals is to develop a habit to be able actively to consider them at will: “So when it is said,
‘So and so knows the intelligibles,’ its meaning is that he is such that whenever he wishes he can
make its form present.”20
Even though there is no faculty of memory in the intellect, Avicenna’s view still requires
an explanation of how we can trigger our dispositional knowledge in cases where we seem to
forget what we have learned, for example, if I “forget” a mathematical proof because I am nerv-
ous or haven’t thought about it for some time. I might discover that I really don’t understand it
at all any longer, in which case it would require relearning it from scratch, so that the notion of
“memory” no longer applies at all. But if I still understand the relevant concepts, then I will be
able to recollect it through the rational direction of my internal sensory apparatus.
Avicenna explains in some detail how the process of recollection differs from learning
(al-ta allum), that is, acquiring new knowledge, which also depends on the cooperation between
the internal senses and the intellect. Both recollection and learning involve a psychological
“transfer” from an occurrent perception to one that it is not actively under consideration.
Whereas recollection seeks what is “related to a thing which was existent in the soul in past
time,” learning seeks a transfer “from the known to unknown.”21 Moreover, while both learn-
ing and recollection rely upon a causal link between the starting point and the desired goal, in
the case of recollection the connection is not a necessary one, whereas in learning it is.22 In the
original act of learning, the intellect follows the necessary and essential connections between
concepts; by contrast, the vehicles of recollection are subject-relative and depend upon the
prior associations of the one recollecting—they are merely “signs” of what is sought. Avicenna
gives the example of a book that calls to mind the teacher with whom I studied it. While that
book will bring her to mind for me, it will not have the same mnemonic effect for someone
else. Indeed, there is no guarantee that it will produce the sought-after recall even in my case.
By contrast, the tools of logic produce certain knowledge by necessity: if I know a set of true
premises and derive a conclusion from them by logically valid methods, I must necessarily assent
to that conclusion.
3. Averroes
Averroes’s account of the internal senses

Averroes’s main account of memory is found in his early Epitome of the Parva naturalia, a
discussion to which he also alludes in his late Long Commentary on De anima. In contrast to
Avicenna, and in keeping with the Arabic version of the Parva naturalia, Averroes recognizes
four internal senses: the common sense, the imagination or formative faculty, the cogitative or
discriminative faculty, and memory. In the Epitome of the Parva naturalia, Averroes implicitly
eliminates Avicenna’s faculty of estimation in animals; in his later Incoherence of the Incoherence,
he explicitly rejects it as a superfluous innovation.23 Thus the faculty of memory is no longer
a storehouse for the intentions perceived by estimation, becoming instead the percipient of
intentions that have been extracted from images by the cogitative faculty, a uniquely human
internal sense power.
454
Memory, intentions, and individuals

While Averroes, like Avicenna, links memory to the perception of intentions, his understanding
of what an intention is differs from that of his predecessor. For Avicenna, an intention is any
non-sensible, non-material property that is perceived along with sensible forms—the paradig-
matic case being affective properties such as hostility. Averroes, by contrast, identifies intentions
as representations of the individual as such: “For that which the imaginative faculty perceives
of an ostensible individual Zayd is simply what the painter describes of him in [the faculty] that
retains, whereas that which the memorative faculty perceives is simply the intention of this
description.”24 When we perceive the individual man Zayd, the images retained in the imagi-
nation are representations that describe his external sensible qualities—for example, that he has
dark hair, is of a certain weight, has a deep voice, and so on. The perception of these qualities
as belonging to this particular concrete individual involves a further representation, and this is
what intentions contribute on Averroes’s view.25
But why does Averroes assign the perception of these intentions to memory? It seems that
Averroes is highlighting the fact that remembering depends upon the ability to recognize an
object as the very same one that I have perceived in the past. When I say, “I remember Mary,”
the salient feature is that I am identifying the object of my perception as Mary. But this depends
in turn on my ability to associate this particular description with this concrete individual. In
order for the image of a cat to be not just any image, but that of my cat, Fuscus, it must combine
the sensible qualities of Fuscus—his grey fur and yellow eyes—with a reference to a specific
cat. Thus, although remembering presupposes the prior activities of sensation and imagination,
another faculty is needed to grasp the intention. And since that faculty plays the central role in
the act of remembering, it is appropriate to term it the “memorative” faculty, although as we
shall see it has other functions in Averroes’s psychology. Averroes holds that our ability to per-
ceive the intention in turn depends on a process of analysis or separation (taḥlīl/tafṣīl), and this
requires the introduction of another power, the cogitative or discriminating faculty, an exclu-
sively human internal sense closely aligned with intellect.26 Averroes describes the intention that
the cogitative faculty abstracts for memory as the essential “core” or “fruit” of the sensed object,
since it is what makes it the individual thing that it is. By contrast, the external sensible forms
represented by its image are like “rinds,” which are peeled away to reveal that individual core.27
By shifting the focus of his account of memory to the perception of the concrete individual,
Averroes downplays the idea that memory is principally concerned with our awareness of the
past as past. We don’t remember something simply because we are aware of it as a past object
of our perception, even though having perceived an object in the past is a necessary condition
for memory. Instead, we remember it because we have identified it as this individual through
the abstractive process, and then we later recognize it as such when we recall it. If for some
reason I fail to recognize some past object of my perception as a particular individual, I am not
truly remembering it, but merely observing or imagining it anew. Memory, then, has more to
do with my awareness of the identity of an object than with my awareness that it was an object
for me at some past time.
While it may seem counterintuitive at first glance to sever memory from the awareness of
the past, there is some plausibility in Averroes’s position. My memories need not be explicitly
subjective or self-referential. Some complex memories will include my awareness of myself as
a feature, but others may be fully focused on the objects. Obviously, if they are my memories
then I will have experienced them in the past, but the phenomenal content of my memory may
have no reference to my subjective experience at all. And it is the recognition of this point that
allows Averroes to downplay the link between memory and the experience of the past.
455
Deborah L. Black
Animal memory?
Whereas Avicenna’s account of memory was closely linked to his concept of the estimative fac-
ulty in both humans and animals, Averroes’s account implies that animals do not have a distinct
faculty of memory. The possession of the cogitative faculty is a condition for the presence of
the memorative faculty, since memory receives intentions only after the cogitative faculty has
extracted them from the images that encase them. But the cogitative faculty is unique to humans
on Averroes’s view. Although he talks about animal memory in the Epitome of the Parva natu-
ralia and differentiates animals that remember from those that do not, in both his early and his
late works, Averroes denies that animals can perceive intentions.28 Their perceptual capacities
extend only to the external sensible features or “rinds” of physical objects:
As for the properties of perception in these five senses of animals, they are not of a
single mode. This is because in a human, they perceive the differences of things and
their proper intentions, and these are what hold the rank in relation to the thing
sensed of the core in relation to the fruit. In animals, however, they perceive only
the things which are external, these being that whose relation to the things is the
relation of rind to the core of the fruit.29
In his later Long Commentary on De anima, Averroes reiterates the same point, noting that the
ability to grasp individual intentions seems to be “proper to the senses of a human,” insofar as it
is dependent on the abstractive operations of the cogitative faculty.30
Averroes’s overall view seems to be that animals have an inchoate and instinctive grasp of the
intention insofar as it is contained within its sensible rinds. But they are unable to separate that
intention from the image that surrounds it, and that seems to suggest that they have no ability
to pick out this individual from that one—except insofar as the two individuals have different
external sensory qualities or rinds. This limitation of animal perception to the recognition of
external features is considerably weaker than Avicenna’s account of animal estimation and the
capacities for experience and learning that build upon it. What Averroes seems to be saying
is that higher animals “remember” in the sense that they can identify a sensory object which
they perceived in the past through its rinds alone. My cat recognizes me by my scent, voice,
and appearance, and he is able to associate those sensory qualities with being fed and groomed.
But he does not consciously recognize me as a concrete individual substance, since he cannot
extract my intentional core. That requires a cogitative faculty under the influence of intellect.
Once again, then, mere retention of past sensory information, and even the recognition of it as
something sensed in the past, is neither sufficient for memory nor essential to it. Memory rests
at bottom on the capacity to recognize an individual as this individual, not merely as a collection
of sensory qualities.
Recollecting and creative imagination

For Averroes, committing an object to memory and recalling it are complex acts that require
the positing of a separate faculty of memory, although they also rely on the activities of other
internal senses. By the same token, the faculty of memory itself is involved not only in activi-
ties that pertain to the retention and remembering of things one perceived in the past, but also
to other activities that have nothing at all to do with memory. Chief amongst these are acts of
creative imagination, that is, constructing an image of some sensory object that one has never
experienced, which Averroes takes to be dependent on the same mechanisms as recollection.
456
Averroes defines recollection as the act of “voluntarily seeking the intention after having
forgotten it,” and then “making it present after its absence.” This takes place through the reverse
process to the one that produced the original memory, which Averroes calls “composition” or
“synthesis” (tarkīb). Like analysis or separation, synthesis depends on the cogitative faculty, in
this case to reunite the image with the intention and recover the forms which had once cov-
ered that intention.31 Creative imagination involves a similar synthetic operation in which we
represent objects that we have never actually sensed in the past upon having their attributes
described to us. The objects of such acts of creative imagination need not be simply fictional,
and Averroes himself is most interested in cases where someone is able to imagine something
accurately based simply on its description. Averroes’s favorite example of this process is someone
being able to imagine what an elephant looks like without having seen one before.32 Averroes
adds that all acts of synthesis, both those used for recollection and those used to imagine what
one has not actually experienced, require the obedience of the internal sensory faculties to
reason, in contrast with the basic analytic operations which are a function of the sensory soul
alone. Since both processes require the cogitative faculty to combine the image and intention,
Averroes cannot be attributing these two abilities to animals. Rather, he is pointing out that the
basic operation of abstracting the intention is an automatic one performed by the internal senses
of human beings, whereas the acts of recollection and creative imagination require a voluntary
and conscious effort on our part.
This close association between recollection and creative imagination is intriguing and
follows naturally from Averroes’s acceptance of memory as a function of the abstraction of
individual intentions. But it is also somewhat puzzling, raising further questions about the
nature of the individual intention. In the case of creative imagination, it is unclear what
would function as an intention for an object that does not represent an actually experienced
individual. Can the soul concoct intentions in the memory that do not come from past
experience? Perhaps the created image is like a shell without any fruit in it, to use Averroes’s
metaphor. But that would fail to account for Averroes’s claim that the creative act involves a
synthesis of the same constituents that were originally separated out by the cogitative faculty,
that is, both the image and the intention. So it seems that when someone who has never seen
an elephant imagines one, she would have to attach some intention to it, perhaps that of some
other animal that she has perceived in the past. If that is the case, then it seems wrong to take
the intention as representing some unique property of the individual in question—it doesn’t
seem to be anything like “Zaydness,” for example, but something more like “this human.”
And of course, if the creative act requires memory to supply an intention for the image, this
raises serious questions about the markers for the veridicality of memories. But perhaps that
is the very point of the analogy between memory and creative imagination: the fact that both
rely upon parallel synthetic operations is one way to explain why some people confuse real
memories from imagined ones.33
Memory and intellect

Unlike Avicenna, Averroes does not explicitly raise the question of whether there is memory
in the intellect. In the Epitome of the Parva naturalia, he asserts that we can only recollect uni-
versals with reference to the “imagined things upon which they depend.” By the same token,
universals can also be forgotten if one has no image in which to think them.34 This view reflects
Aristotle’s claim that the soul cannot think without images, a central dictum of the De anima
that is reiterated at the beginning of the De memoria. It is a view that is at the foundation of
Averroes’s theory of the intellect, especially his mature and much-maligned doctrine of the
457
Deborah L. Black
unicity of the Material Intellect, according to which both the intellectual principles posited
by Aristotle, the material (or potential) and active (or agent) intellects are a separate substance
shared by all human knowers.
On Averroes’s understanding of the unicity of the intellect, the universal intelligibles that
form the content of our abstract thinking do not reside in us individually, but instead have as
their subject a separate Material Intellect to which we “conjoin” whenever we actually think.
Nonetheless, these intelligibles still depend for their content on the images in the internal senses
of individual human knowers, and neither the separate intellect nor the individual human is able
to think abstract thoughts about the natural world without their mutual conjoining. For this
reason, neither separate intellect nor the individual human serves, even metaphorically, as the
storehouse for our intelligibles in Averroes’s psychology. For Averroes as for Avicenna, insofar
as learning implies the ability to recall at will what we have learned, it should not be understood
as a process whereby individual minds amass a personal hoard of intelligibles. Learning is instead
to be understood dispositionally as an ability to tap into the activities of both the Agent and
Material Intellects whenever we wish, “because these two activities are reduced to our will,
namely, to abstract intelligibles and to understand them.”35 Averroes, then, does not deny the
existence of intellectual memory; it simply does not occur to him to associate the concept of
memory with the intellective aspects of the mind. Memory is an exclusively sensory capacity,
which impinges on the realm of abstract thought only incidentally.
Acknowledgment
This research was supported by the Social Sciences and Humanities Research Council of Canada.
Notes
1 Avicenna (1978: 97), commenting on De anima 3.3.428b11ff.
2 See Pines (1974); Daiber (1997: 36–41). Rotraud Hansberger studied this text in her PhD thesis and is
currently preparing an edition of the text. For accounts see Hansberger (2008, 2010, forthcoming).
3 Hasse (2000) introduced “connotative attribute” as an alternative translation. While this fits many of
Avicenna’s examples, it is not clear that all the properties that count as ma ānī are connotative, and this
translation will not work for Averroes’s account of intentions.
4 Avicenna (1959, 1.5, 44; Avicenna 1981, 31).
5 Avicenna (1959, 1.5, 43; 2.2, 60–61; 1981, 31).
6 The common sensibles are properties such as magnitude, shape, and motion, which are not exclusive to
any one of the five senses but perceptible to two or more of them.
7 Al-khayāl. Sometimes this is rendered as the “imagery” faculty (McGinnis-Reisman) or the “representa-
tive faculty” (Rahman).
8 Avicenna (1959: 1.5, 44–5; cf. 4.1, 154); Avicenna (2014: Part 3, c. 9, 99–100); see also Di Martino
(2007: 21–2).
9 Di Martino (2007: 21–2).This point is made by Augustine in several places, but the best known is in the
discussion of time in book 11 of the Confessions.
10 This description is not meant to identify estimative intentions with intentions in the generic sense,
that is, intentional objects. Intentions are understood to be real, though non-material, features of the
external world.
11 Avicenna (1959: 4.3, 182).
12 Avicenna (1959: 4.1, 164, 168).
13 Avicenna (1959: 4.1, 167–8).
14 Avicenna (1959: 4.3, 185).
15 Avicenna (1959: 168–9).
16 See the passage below (n. 22) for one of Avicenna’s own examples of a book calling to mind the teacher
with whom one studied it.
458
17 Avicenna (1959: 5.7, 252–4); Avicenna (1981: c. 15, 64–8).

18 Avicenna (1959: 4.3. 184–5). For Avicenna’s innovative account of experience (tajribah) in humans, see
McGinnis (2003); Janssens (2004).
19 Avicenna (1959: 184–5). Avicenna’s account and example of animal experience has a long legacy, most
notably recurring in Leibniz’s identification of animals as “empirics,” in a number of texts, such as
Monadology, §§26–7 and Principles of Nature and Grace, §5.
20 Avicenna (1959: 5.6, 244–7). Avicenna’s account of this dispositional knowledge is related to his under-
standing of the Active or Agent Intellect of Aristotle’s De anima 3.5, which Avicenna interprets as a
separate, immaterial mind.
21 Avicenna (1959: 4.3, 185–6). This is Avicenna’s standard formula for the goal of logic.
22 Avicenna (1959: 186).
23 Averroes (1972: 2.1, 39); Averroes (1954: 333–41).
24 Averroes (1972: 41).
25 For the background in the Arabic Parva naturalia, see Hansberger (forthcoming).
26 Averroes (1972: 43).
27 Averroes (1972: 42–3). The fruit-rind metaphor is derived from the Arabic Parva naturalia.
28 See Averroes (1961: 22, 24).
29 Averroes (1972: 33; 1961, 20 modified).
30 Averroes (2009: 2.64, 176).
31 Averroes (1972: 41).
32 Averroes attributes this point to Aristotle himself in Averroes (1972: 45). The elephant example prob-
ably derives from al-F r b ’s Book of Letters (al-F r b 1969: §169, 169–70), where the examples are an
elephant and a camel. (Thanks are due to David Wirmer for the F r b reference.)
33 In De memoria 451a5ff. (Aristotle 1975). Aristotle raises the problem of how we know whether we are
actually remembering something, and mentions people affected by madness “who spoke of their images
as if they had actually taken place, and as if they actually remembered them.”
34 Averroes (1961: 22).
35 Averroes (2009: Bk. 3, comm. 18, 351).
Related topics
• Aristotle
• Augustine
• Aquinas
Bibliography
Al-F r b 1969. Kitāb al-ḥurūf. Alfarabi’s Book of Letters, ed. Muhsin Mahdi. Beirut: Dar el-Machreq.
Aristotle 1975. On The Soul, Parva naturalia, On Breath, trans. W. S. Hett. Cambridge, MA: Harvard
University Press.
Averroes 1953. Commentarium magnum in Aristotelis De anima libros, ed. F. S. Crawford. Cambridge, MA:
Mediaeval Academy of America.
Averroes 1954. Tahafut al-Tahafut (The Incoherence of The Incoherence), 2 vols., trans. Simon Van Den Bergh.
Averroes 1961. Epitome of “Parva naturalia,” trans. H. Blumberg. Cambridge, MA: Mediaeval Academy
of America.
Averroes 1972. Talkhīṣ kitāb al-ḥiss wa-al-maḥsūs, ed. H. Blumberg. Cambridge, MA: Mediaeval Academy
of America.
Averroes 2009. Long Commentary on the De anima of Aristotle, trans. R. C. Taylor and T.-A. Druart. New
Haven, CT: Yale University Press.
Avicenna 1959. Avicenna’s De anima, Being the Psychological Part of Kitāb al-Shifā , ed. F. Rahman. London:
Avicenna, 1978. Al-Ta līqāt alā awāshī Kitāb al-Nafs (Marginal Glosses on De anima), ed. A. R. Badawi in
Arisṭū inda al- Arab, 2nd edn. Quwait: Wak lat al-Matbu at, pp. 75–116.
Avicenna 1981. Avicenna’s Psychology, trans. F. Rahman. Westport, CT: Hyperion.
459
Deborah L. Black
Avicenna 2014. Remarks and Admonitions: Physics and Metaphysics, trans. S. C. Inati. New York: Columbia
University Press.
Black, Deborah 1993. “Estimation in Avicenna: The Logical and Psychological Dimensions.” Dialogue 32:
219–58.
Black, Deborah 1996. “Memory, Time and Individuals in Averroes’s Psychology.” Medieval Theology and
Black, Deborah 2000. “Estimation and Imagination: Western Divergences from an Arabic Paradigm.”
Topoi 19: 59–75.
Bloch, David 2007. Aristotle on Memory and Recollection: Text, Translation, Interpretation, and Reception in
Western Scholasticism. Leiden: Brill.
Daiber, H. 1997. “Salient Trends of the Arabic Aristotle.” In Gerhard Endress and Remke Kruk (eds.), The
Ancient Tradition in Christian and Islamic Hellenism. Leiden: Research School CNWS, School of Asian,
African, and Amerindian Studies, pp. 29–41.
Di Martino, Carla 2007. “Memory And Recollection In Ibn S n ’s And Ibn Rushd’s Philosophical Texts
Translated into Latin in the Twelfth and Thirteenth Centuries: A Perspective on the Doctrine of the
Internal Senses in Arabic Psychological Science.” In H. Lagerlund (ed.), Forming the Mind. Essays on the
Internal Senses and the Mind/Body Problem from Avicenna to the Medical Enlightenment. Dordrecht: Springer,
pp. 17–26.
Hansberger, Rotraud 2008. “How Aristotle Came to Believe in God-Given Dreams: The Arabic Version
of De divinatione per somnum.” In. M. Ashtiany (ed.), Dreaming across Boundaries: The Interpretation of
Dreams in Islamic Lands. Cambridge, MA: Harvard University Press, pp. 50–77.
Hansberger, Rotraud 2010. “Kitāb al-ḥiss wa-l-maḥsūs: Aristotle’s Parva naturalia in Arabic Guise.” In
C. Grellard and P.-M. Morel (eds.), Les Parva naturalia d’Aristote: Fortune antique et médiévale. Paris:
Sorbonne, pp. 143–62.
Hansberger, Rotraud forthcoming. “Representation of Which Reality? ‘Spiritual Forms’ and ‘Ma ānī’ in
the Arabic Adaptation of Aristotle’s Parva Naturalia.”
Hasse, Dag N. 2000. Avicenna’s De anima in the Latin West. London: Warburg Institute.
Janssens, Jules 2004. “‘Experience’ (tajriba) in Classical Arabic Philosophy (al-F r b -Avicenna).” Quaestio
4: 45–62.
McGinnis, Jon 2003. “Scientific Methodologies in Medieval Islam.” Journal of the History of Philosophy 41:
30–27.
Müller, Jörn 2015. “Memory in Medieval Philosophy.” In Dmitri Nikulin (ed.), Memory: A History.
Pines, Shlomo 1974. “The Arabic Recension of Parva naturalia and the Philosophical Doctrine Concerning
Veridical Dreams.” Israel Oriental Studies 4: 104–53.
Taylor, Richard 1999. “Remarks on Cogitatio in Averroes’ Commentarium Magnum in Aristotelis De
Anima Libros.” In J. A. Aertsen and G. Endress (eds.), Averroes and the Aristotelian Tradition. Leiden:
Brill, pp. 217–55.
Wolfson, Harry A. 1935. “The Internal Senses in Latin, Arabic, and Hebrew Philosophic Texts.” Harvard
Theological Review 28(1935): 69–133.
460
37
THOMAS AQUINAS
John O’Callaghan
1. Introduction
This chapter will focus on two aspects of Aquinas’ treatment of memory: its sources, primarily
in Aristotle and Ibn Sina,1 and the distinctive position he takes on intellectual memory that he
draws in partial opposition to Ibn Sina. Memory in the proper sense involves a corporeal faculty
for the retention of temporally indexed sensible forms—temporally indexed particular sense
information employed teleologically in the life of perfect animals. Sense memory is memory per
se, while intellectual memory, a habitual state of knowledge in the incorporeal faculty of intel-
lect, is only memory per accidens. The chapter will proceed in two stages, considering sense
memory first and then intellectual memory.
There are several important settings for understanding Aquinas’ account of memory—
primarily his Commentary on Aristotle’s On Memory and Recollection, and his Commentary on
Aristotle’s Metaphysics, but also his development of his general account of the powers of the
soul in the Summa Contra Gentiles, the Summa Theologiae, and the Disputed Questions on the Soul.
Thematically, he works out his account in his dispute with Ibn Sina on the correct interpreta-
tion of Aristotle on the intellect, particularly on intellectual memory, but also his discussion
of the human as an image of God, engaging Augustine’s theory of mind from the latter’s
De Trinity. So Aquinas’ positions on memory, sensory and intellectual, provide a striking
instance of the interplay of philosophical and theological concerns in his thought.
2. Sense memory
Aquinas wrote twelve commentaries on the works of Aristotle, some complete and some incom-
plete.2 Given the temporal proximity of their composition to that of corresponding discussions
in the Summa Theologiae, at least one purpose for them was to serve as a review of Aristotle to
be employed in writing the Summa.
One commentary bears the title Sentencia Libri De Sensu et Sensato, but consists of two
commentaries. The first is on Aristotle’s De sensu et sensato, while the second is a commentary
on De memoria et reminiscentia (Torrell 1996: 341). It was written slightly after his Commentary
on Aristotle’s De anima and likely while Aquinas was composing question 78 of the first part
of the Summa on the “special powers of the soul,” the question in which he addresses the
461
John O’Callaghan
faculty of memory. The treatment of memory in 78.4 as one of the “interior” senses follows
immediately upon the treatment of the “exterior” senses in 78.3, paralleling the dual structure
of the commentary, with memory and reminiscence treated immediately after sensation and
what is sensed. Memory is distinguished in both discussions as among the “internal senses,” as
opposed to the five “exterior senses” of touch, taste, smell, vision, and hearing.
Considering memory per se—sense memory—Aquinas’ treatment in 78.4 is concise and
parallels closely his Commentary on Memory and Recollection. He first distinguishes what he refers
to as those powers that are precursors to the functioning of the intellect from the intellectual
and appetitive powers. The intellectual powers comprise the agent intellect responsible for
abstracting intelligible forms from sense experience of material objects and the possible intellect
in which the abstracted intelligible forms reside and through which intellect gives expression to
understanding in act.3 The appetitive powers are sense appetite and rational appetite or will, the
faculties through which some object is desired as pleasing to the senses, sense appetite, or desired
as an understood good, rational appetite.
Sense memory is located among the cognitive faculties that are precursors to intellect, the
cognitive sense faculties that intellect normally presupposes for the information upon which
it acts in abstraction (agent intellect) and then gives expression to in understanding (possible
intellect). Employing the particular information within sense experiences, the agent intellect
produces a general intelligible form within the possible intellect that renders the latter capable
of its operation: general understanding of what things are. The sense faculties that the agent
intellect relies upon are divided into the external senses and the internal senses. The external
senses are the familiar five senses of touch, taste, smell, vision, and hearing; they are immediately
engaged and activated by their proper objects in the actual presence of corresponding sense
qualities of material objects.
Aquinas distinguishes these external senses from the internal senses by the teleological func-
tion of the latter. He explains that the life of a “perfect animal,”4 consists in moving toward or
avoiding material objects not actually present but that become present or distant by the move-
ment. This follows some cognitive apprehension that informs the appetitive powers. But absent
the material object desired or feared, there must be apprehension of that desired or undesirable
object for the movement to begin.
The external senses only apprehend their objects in the presence of those objects, and only
apprehend sense modalities like odor versus color versus solidity or softness, and so on. But sen-
sation involves the apprehension of unified sensible wholes, not simply disparate sense qualities,
the apprehension that color is not odor is not solidity or sweetness, and the apprehension of
magnitude, motion, and time (Aquinas 1985: cap. 2, #139–47). As opposed to the proper sense
qualities that pertain to the five sense powers, Aquinas refers to the sensible form, by which
he means the unified whole of sense experience—the flower in its shape, odor, and color, as it
sways in the breeze, and so on. Therefore there must be sense powers that apprehend material
objects as sensible wholes, that can discern the proper sensible forms from one another, and that
can apprehend material objects in their absence to judge whether to approach or flee material
objects. These modes of apprehension require four internal senses.
First, the faculty responsible for the reception of sensible forms as unified wholes and discern-
ing the proper sense qualities from one another is called the “common” sense, to distinguish
it from the five proper external senses. The sensible form within it Aquinas calls a “phantasm”
(phantasma). This faculty is called “common” because all the proper senses are teleologically
referred to it—they exist to provide it with the information it unifies, and Aquinas calls it their
“common root and principle” (Aquinas 1885: Ia.78.4 ad 1). While in the Summa he does not
speak of it apprehending magnitude, motion, and time, in the commentary he argues that it
462
Thomas Aquinas
does in order to account later for memory and recollection. Second, the power that retains the
sensible forms of material things as such he calls phantasy (phantasia) or imagination (imaginatio).
Third, pursuing goals, a perfect animal makes judgments concerning what is to be avoided
as harmful and what is to be pursued as useful. Employing a famous example taken from Ibn
Sina (Aquinas 1885: Ia.78.4 respondeo), the sheep apprehends a wolf as something to be avoided,
not simply as a wolf. So a faculty of judgment that Aquinas calls the “estimative power” in
animals is posited to account for this judgment of what Aquinas, again following Ibn Sina, calls
“intentions” (Black 2000: 60) that are included within sense experience but not grasped by the
five senses or the common sense. In effect the intentions of harmful or useful are included as
non-sensible features and information conveyed by the five senses and common sense, but not
phenomenologically apprehended by them.
The estimative sense is a faculty for making particular judgments in particular contexts.
This wolf is to be avoided—flee. This gazelle is to be pursued—chase. Where these promote
the good of the animals, Aquinas attributes prudence to these animals, although an imperfect
prudence compared to human intellectual prudence (Aquinas 1985: cap. 1, #34–8). In most
animals, these judgments of the estimative sense grasping the intentions are mostly instinctual
(Aquinas 1885: Ia.78.4 respondeo). However, because human beings possess intellect, the estima-
tive sense in them participates in the judgments of intellect and the desire of the will.
Human intellect in this life, because of its manner of gaining information about the mat-
erial world by abstracting from the particular, makes purely general judgments—broccoli
when eaten is a healthy food; health should be pursued, and so on (Aquinas 1918: II, cap. 74,
b.37–45). The will desires health as understood in general to be good for human beings. So
intellect and the estimative power come together in a synthesis of judgment that this broccoli
before one should be eaten not simply because it is desirable to sense appetite, but because it
is understood as good for one; understood it is desired by the will. Because of this understand-
ing and willing of the broccoli as good for one and so to be eaten, one may with one’s will
overcome the sense aversion one may have for broccoli that would otherwise affect the esti-
mative sense in such a way as to avoid broccoli. Intellectual understanding sheds light upon
what would otherwise be an instinctual estimation, and the power of will informed by that
understanding affects the avoidance or pursuit of some good in that intelligible light. Because
of the role of intellect and will informing the estimative sense, Aquinas calls the estimative
sense in human beings the cogitative sense.5
Fourth, sense memory comes to the fore in Aquinas’ account. Where in other animals
the judgments of the estimative sense are mostly instinctual, Aquinas thinks in human beings
they are acquired through learning. Learning requires memory; it requires a “repository”6 of
sensible forms that can be employed by the cogitative sense in its judgments as images of past
experience—not just sensible forms but sensible forms grasped as images of objects previously
apprehended in sensation. Thus the sensible forms are grasped as temporally indexed (Aquinas
1985: cap. 3, 271–81).7 This temporal indexing of the phantasm can be more or less vague as
between “last Friday at noon,” “last week around noon,” and “a while ago,” for example. Like
imagination, memory conserves phantasms but with the addition of these intentions including
time. The developed habitual state of the faculty is called memory just as the faculty is (Aquinas
1985: cap. 3, 22–44). It, with imagination and the cogitative sense, is a function of a corporeal
organ, the brain (Aquinas 1996: 8 respondeo).
Where Ibn Sina had identified five interior senses, Aquinas only identifies four. As Aquinas
understands him, Ibn Sina had posited a faculty “in between” imagination and estimation, a
faculty for combining sensible forms into new un-experienced sensible forms, like a golden
mountain. This activity is what we would call being imaginative. Aquinas does not deny the
463
John O’Callaghan
reality of this activity. He thinks it more parsimonious to attribute this activity to the imaginative
power itself, rather than a fifth power, so imagination both conserves experienced sensible forms
and engages in this activity of making new sensible forms from what has been conserved. Here
is a reason for distinguishing the two conserving powers, imagination and memory. Imagination
may combine sensible forms in new ways producing sensible forms of objects that have never in
fact been sensed, without the intentions of harmful or useful, and without the intention of time.
For memory to be veridical, it cannot be of its very nature free in this way.8
Recollection (reminiscentia) differs from memory because recollection involves rediscovering
something that was previously experienced but conserved in memory; it involves the active
seeking after of some phantasm as an image of past experience to bring it to mind. Aquinas
writes that we remember “through” recollection, a kind of “movement” of the mind that
terminates in remembering. Recollection begins either with some new sense experience that
prompts the movement toward a memory, or with something already remembered that moves
one on to more recollection (Aquinas 1985: cap. 4, 95–145). Giving it a distinctively rational
valence, Aquinas speaks of recollection as a “quasi-syllogistic inquiring into a memory of past
things according to individual intentions” (Aquinas 1985: cap. 8, 17–34).
Where Aquinas grants an imperfect prudence to other animals, here he grants an imper-
fect memory to other animals as well (Aquinas 1985: cap. 1, 34–8).9 He distinguishes it from
human memory because where other animals have only memory (conservation of phantasms
temporally indexed), human beings have memory and recollection. Presumably this difference
is because while he believes the judgments of the estimative sense in other animals employing
memory is largely instinctual, the judgments of the cogitative sense in humans employs intel-
lect and will in a deliberative fashion, pursuing some understood good. We don’t simply flee
lions, since we may have judged that studying them is good, willed to study them, and then
reasonably judged that while we should be careful about this one before us, still, to study it we
should not flee from it. Here again the will informed by reason can provide an ordering to the
lower powers and passions they might not have in the life of other animals.
There are other significant mentions of memory in Aristotle that Aquinas knows of and that
he comments upon in the relevant commentaries or employs in his own particular discussions.
Significant for the point about memory and learning involved in the judgments of the cogitative
sense is the discussion of memory in Aristotle’s Metaphysics. There, Aquinas again distinguishes
the role of memory in the life of perfect animals from both sensation and imagination. Memory
is involved in the planning an animal engages in to pursue some goal, typically for its survival;
in its teleological function, it has a temporal aspect looking to the future, while its intentional
contents look to the past (Aquinas 1950: I, lectio 1).
Many memories concerning some matter lead to the building up of “experience.” By
experience, Aquinas does not mean a particular case of sense experience, as we might say
someone “had an experience.” He means something like insight into a subject matter or prac-
tice, as we might say someone is “experienced in some matter.” So Aquinas again attributes
an imperfect prudence to animals with imperfect memory. Judgments are made on the basis
of experience. In the Commentary on the Posterior Analytics, he provides much the same account
(Aquinas 1989: II, lectio 20, 112–95).
In both these commentaries, he follows the text of Aristotle, with the exception that in the
Commentary on the Posterior Analytics he distinguishes between “experience” and what Aristotle
called the universal “resting” in the intellect, a distinction between what is taken to be the case
“in certain instances” and what is taken to be the case “in all instances.” The universal that
goes beyond experience loses the temporal indexing within the information held in memory
and built into experience, since judgments involving it will hold for all times. Still, the account
464
Thomas Aquinas
of sense memory is basically the same—its function is teleologically ordered toward the acts of
perfect animals, brute or human, that can plan and move toward a goal.
The important point from the commentaries on the Metaphysics and the Posterior Analytics
is that they provide, however briefly and sketchily, the account of learning involving memory
and the cogitative sense in the transition from sense experience to intellectual understanding by
the grasp of the universal that informs understanding. In that respect, they present a picture of
the distinction between the merely estimative sense in other animals and the cogitative sense in
human animals. Other animals have both memory and experience informing their estimative
judgments about what is pursued and avoided. But the human animal has the cogitative sense
which is the estimative sense informed by intellectual understanding—a learned knowledge
built up from sense memory through experience to understanding, deployed in particular judg-
ments of what to do here and now.
So, sense memory is the internal sense faculty of retention of sensible forms with the inten-
tions of harm or usefulness or pleasure, and so on. But of particular importance as an intention
accompanying sense experience are temporal indices, and that the sensible form employed in
sense memory is understood to be an image of something encountered by sensation in the past.
So memory specifically includes within its object not just the intentions of harmful, pleasing, or
useful, but also the “formality of the past” which formality of its object is the basis for Aquinas
writing that sense memory is memory per se (Aquinas 1985: cap. 2, 198–210).
3. Intellectual memory
While much of what Aquinas says about sense memory reflects with modifications his sources
in Aristotle and Ibn Sina, with the notion of intellectual memory he stakes out a distinctive
position. Ibn Sina identifies memory exclusively with sense memory, and has no account of
intellectual memory. He has the external senses and his catalogue of internal senses. Besides
those faculties, he has the possible intellect and the agent intellect. While each soul has its own
possible intellect, the agent intellect is one universal cause for all. He held for a strong distinc-
tion between soul and body, in which the soul is not understood to be the substantial form of
the body, and though individuated by a particular body it is only accidentally related to it; this
position complicates his account of abstraction.
The process by which the possible intellect acquires intelligible forms in understanding
(acquired intellect) is a complicated interaction of abstraction from sensibles and emanation
from the agent intellect. Material intelligible forms are abstracted from sensibles by the turn-
ing of the possible intellect to the agent intellect, which had emanated the intelligible forms
into matter. The result of being brought to actually understand some intelligible form by
abstraction and emanation is that the possible intellect of an individual soul comes to have
developed within it a greater facility for subsequent abstraction. In effect, its initial bare
potentiality develops into a skill for being united to the agent intellect, which is the repository
of intelligible forms.10
But the point of concern here is that this developed state of the possible intellect does not
become a “repository” of the intelligible forms once understood (Müller 2015b: 499–500). The
soul simply turns to the agent intellect with greater facility upon subsequent occasions to abstract
yet again by emanation the intelligible forms it had considered in act in the past. There is no mid-
dle ground for the possible intellect. It either completely lacks the intelligible forms of things and
has no understanding of them, or it possesses the intelligible forms when it is actively engaged in
understanding those things whose forms it possesses; there is no place for the intelligible forms
within it unless they are being considered in full actuality.
465
John O’Callaghan
Aquinas holds that the soul is the substantial form of a living human body that is the human
being. Soul and body are not two beings related per accidens by some union, but are per
se related as one being (Aquinas 1885: Ia.75–6; Aquinas 1996: 1 respondeo and ad 1). Each
human being has his or her own individual agent intellect that uses sense experience to abstract
the intelligible forms of things from sense experience of those things (Aquinas 1885: Ia.79.5;
Aquinas 1996: 5 respondeo). The agent intellect renders the possible intellect informed regarding
these intelligible forms, and then the possible intellect gives expression to understanding in act.
Take for example oak trees. On Ibn Sina’s view, the body confronted with sense experience
of oak trees or remembering particular oak trees, and the possible intellect of the soul having
developed a greater facility to turn to the agent intellect emanating the intelligible form of oak
tree, then the possible intellect once again abstracts the intelligible form of oak tree, and the
soul actively understands oak trees. Otherwise it has no knowledge of oak trees; and it retains
no knowledge of oak trees. Aquinas thinks that we do not learn again what oak trees are every
time we actively consider them and engage in understanding them. We give expression to a
habitual knowledge that learning has developed in the possible intellect. So even when we are
not engaged in considering oak trees and giving expression to our understanding of them, we
can be said to know them, because knowledge is a habitual state of the possible intellect. There
is no need to acquire the intelligible forms again and again.
Aquinas argues against Ibn Sina that his account of intellectual memory is the best way to
understand what Aristotle means in De anima III.4 that after having learned some point of
knowledge, having “become one” with the object of knowledge, the possible intellect is still
in potentiality to its operation of actually expressing that knowledge, a potentiality different
from that with which it began before “learning and discovering.” It is this developed habit-
ual state of the possible intellect that Aquinas identifies with intellectual memory (Aquinas
1885: Ia.79.6 respondeo; Aquinas 1984: III, cap. 2.1–59; Aquinas 1949: 10.2 respondeo). Like
sense memory, it is a “repository” of forms, but of intelligible forms rather than sensible forms.
These intelligible forms standing habitually in the intellect acquired by abstraction from sense
experience are expressed in understanding.
Like any habit, intellectual memory can be strengthened and lost. And the mere presence of a
habit is not sufficient for its expression. To exercise this intellectual memory in act, the intellect
must return to a phantasm of some sensible particular or particulars (Aquinas 1996: 15 respondeo).
Here Aquinas sees a problem for Ibn Sina, since the latter also stressed the importance of the
phantasm in exercising actual knowledge, but could not in Aquinas’ view adequately explain
that importance (Müller 2015b).
Intellectual memory is only memory per accidens. The claim that it is a kind of memory is
twofold. First, it is habitual like sense memory. Second, there is a temporal aspect to it; not only
is it knowledge learned in the past, but also Aquinas will say that the intellect knows that it is
recalling what it already knew and has retained. But it does not have the essential or per se aspect
of memory, which would be the temporal indexing of the intelligible forms that are its object.
While I learned, retain, and recall the knowledge that the cube root of twenty-seven is three,
time is not included within the actual intentional content of that understanding; it is not part of
its formal object (Aquinas 1985: cap. 2, 147–65, 234–52, cap. 4, 64–94). The process of abstrac-
tion leads to general understanding shorn of the particularities of time and place. “All human
beings are rational animals,” not “that human being yesterday was a rational animal.” To get the
latter, the understanding of intellect must come together with the judgment of the cogitative
power which bears upon particular individuals sensed, imagined, or recalled by sense memory.
Secondary scholarship tends to focus upon this dispute with Ibn Sina about the proper
interpretation of Aristotle to emphasize the importance of intellectual memory in the disputes
466
Thomas Aquinas
between Islamic and Christian thinkers in the history of medieval philosophical thought.
But, it also has important implications for Aquinas’ philosophical anthropology, displaying
his commitment to hylomorphism (Müller 2015b).11 It raises important questions about the
cognitive condition of the incorporeal soul after death, since sense memory as a corporeal
function located in the brain ceases to exist upon the death of the body, while intellectual
memory as an incorporeal function and habit of the soul survives with the soul. Little note,
however, is taken of the great theological importance of intellectual memory for Aquinas.
And yet he shows himself eager to identify this intellectual memory with the memory that
Augustine had said in his De Trinity forms within us part of the image of God, the unity of
mind expressed in the life of the human animal according to memory, intellect, and will
(Aquinas 1996: 19 ad 16; Aquinas 1949: 10:2–3 passim and esp. 2 ad 5; Aquinas 1885: Ia.93.7
ad 3; O’Callaghan 2007).
4. Conclusion
In Aquinas, memory per se is sense memory. It is a corporeal power of the human being that is
a repository of sensible forms temporally indexed, the latter being what Aquinas following Ibn
Sina calls intentions of time with other intentions like harmful or useful. It has a twofold tele-
ological role. First, it provides to the estimative power the sensible forms with these intentions
for the judgments perfect animals make that promote the survival of those animals, by inform-
ing sense appetite to either flee or pursue. Second, in human animals it informs the process of
learning by which sensation becomes experience leading eventually to the grasp of universal
understanding. In the human case, it is called perfect memory because of this secondary role.
Since other animals lack intellect, theirs is imperfect memory. Intellectual memory is memory
per accidens, functioning as a repository of intelligible forms not temporally indexed but that
have been acquired by learning and understanding.
Notes
1 Ibn Sina was the tenth- and eleventh-century Persian Islamic philosopher and theologian known in the
West as Avicenna.
2 For a list of these commentaries as well as a brief description of each of them, see the catalogue of
G. Emery, OP (Emery 1996).
3 This distinction of the two intellectual powers reflects Aquinas’ understanding of the difficult passages
on nous or mind in Aristotle’s De anima III.4–5. Aquinas’ detailed interpretation of these passages is
found in his Commentary on Aristotle’s De anima (Aquinas 1984: III, lectiones 7–10).
4 All animals have the capacities for nutrition, growth, reproduction (sterile hybrids like mules notwith-
standing), and some form of sensation. In his Commentary on Aristotle’s Metaphysics, Aquinas distinguishes
so-called “perfect” animals from “imperfect” animals by the capacity for self-movement; a worm is a
perfect animal because of its capacity to move in pursuit of nutriment, while for Aquinas shellfish are
examples of imperfect animals (Aquinas 1950: I, lectio 1, #10).
5 Strictly speaking, Aquinas attributes prudence to the intellect. However, insofar as prudence bears
practically upon individual things and actions to be pursued or avoided, one can appreciate the inti-
mate relationship in human beings between intellect and the estimative sense that bears upon singular
things in their particularity. No level of intellectual prudence would be effective in pursuing a course
of action, if it were not brought to a synthesis of judgment with the particular things and situations
one is confronted with, judgments concerning the particular carried out by the cogitative sense.
6 “thesaurus” in the Latin.
7 In both the Summa and the Commentary, Aquinas speaks of the memory as storing the non-sensible
intentions, which could be taken to mean that these intentions are stored without the corresponding
phantasms. On this scenario, the intentions would be stored in the memory and the phantasms only
in the imagination. But it is difficult to see how the estimative sense could realign these intentions
467
John O’Callaghan
with the right phantasms stored in imagination. So in the commentary Aquinas speaks of memory
conserving the phantasms not in themselves (conserving them “in themselves” pertains to imagina-
tion), but “as images of something previously apprehended.”
8 Aquinas recognizes that memory is not infallible, since the temporal indexing can come loose from the
phantasm taken as an image. He gives two examples. One involves thinking one is remembering but is
not, because the past time that occurs to one is associated with the wrong image. The other seemingly
more complicated example involves remembering but not thinking one is remembering, because the
time does not occur to one in the activity although it is in fact an activity of the faculty of memory not
imagination (Aquinas 1985: cap. 7, 155–70).
9 This use of “imperfect” is not to be confused with its use as applied to “imperfect” animals. A perfect
animal can have imperfect prudence and imperfect memory.
10 See the complicated interpretative problems posed by Avicenna’s work in Hasse (2013: 109–20). Also
Müller (2015a: 114; 2015b: 497–501).
11 Müller is particularly good on how the difference between Avicenna and Aquinas on intellectual mem-
ory reflects the importance of Aquinas’ hylomorphism, as well as the question of the knowledge of the
incorporeal soul after the death of the body.
Further reading
For discussions of Ibn Sina on memory, see Wolfson (1935); Black (1993, 2000); Müller (2015);
“Quaestio 15” (2015). For more general treatments, see Kemp (1996); Ivry (2012); Hasse (2014).
For the complicated interpretative problems posed by Avicenna’s work, see in Hasse (2013); Müller
(2015a, 2015b). For Aquinas’ engagement with Augustine on the mind in the latter’s De trinitate, see
O’Callaghan (2007).
References
Primary sources
Aquinas, T. (1885). Pars prima summa theologiae. Romae: Ex Typographia Polyglotta S.C. de Propaganda
Fide.
Aquinas, T. (1918). Sancti thomae aquinatis doctoris angelici opera omnia: Summa contra gentiles. Romae: Typis
Riccardi Garroni.
Aquinas, T. (1949). In Spiazzi, R. M. (ed.), Quaestiones disputatae: De veritate. Taurini: Marietti.
Aquinas, T. (1950). In Cathala, M. R., Spiazzi, R. M. (eds.), In duodecim libros metaphysicorum aristotelis
expositio. Taurini: Marietti (Commentary on Aristotle’s Metaphysics).
Aquinas, T. (1984). Sentencia libri “de anima”. Rome and Paris: Commissio Leonina J. Vrin (Commentary
on Aristotle’s De anima).
Aquinas, T. (1985). Sancti thomae de aquino opera omnia: Sententia libri de memoria et reminscentia. Rome and
Paris: Commissio Leonina Vrin (Commentary on Aristotle’s On Memory and Recollection).
Aquinas, T. (1989). Expositio libri posteriorum. Rome: Commissio Leonina (Commentary on the Posterior
Analytics).
Aquinas, T. (1996). Quaestiones disputatae de anima. Rome and Paris: Commissio Leonina Ed. du Cerf
(Disputed Questions on the Soul).
Ibn Sina. (1972). In Van Riet S., Verbeke G. (eds.), Avicenna latinus: Liber de anima, seu sextus de naturalibus.
Leiden: Brill.
Secondary sources
Black, D. (1993). Estimation (“wahm”) in Avicenna: The logical and psychological dimensions. Dialogue,
32: 219–58.
Black, D. (2000). Imagination and estimation: Arabic paradigms and western transformations. Topoi, 19(1):
59–75.
Emery, G., O. P. (1996). Brief catalogue of the works of Saint Thomas Aquinas. Saint Thomas Aquinas
(R. Royal, Trans.). Washington, DC: The Catholic University of American Press, pp. 330–61.
468
Thomas Aquinas
Hasse, D. (2013). Avicenna’s epistemological optimism. In P. Adamson (ed.), Interpreting Avicenna: Critical
Essays. Cambridge and New York: Cambridge University Press, pp. 109–20.
Hasse, D. (2014). Influence of Arabic and Islamic Philosophy on the Latin West. Edward N. Zalta (ed.). http://
plato.stanford.edu/archives/fall2014/entries/arabic-islamic-influence/.
Ivry, A. (2012). Arabic and Islamic Psychology and Philosophy of Mind. http://plato.stanford.edu/archives/
sum2012/entries/arabic-islamic-mind/.
Kemp, S. (1996). Cognitive Psychology in the Middle Ages. Westport, CT: Greenwood Press.
Müller, J. (2015a). Memory in medieval philosophy. In D. V. Nikulin (ed.), Memory: A History. New
York: Oxford University Press, pp. 101–9.
Müller, J. (2015b). Memory as an internal sense: Avicenna and the reception of his psychology by Thomas
Aquinas. Quaestio, 15: 497–506.
O’Callaghan, J. (2007). Imago dei: A test case for St. Thomas’ Augustinianism. In M. Dauphinais, B. David
and M. Levering (eds.), Aquinas the Augustinian. Washington, DC: The Catholic University of America
Press, pp. 100–44.
Torrell, J. (1996). Saint Thomas Aquinas (R. Royal, Trans.). Washington, DC: The Catholic University
of American Press.
Wolfson, H. (1935). The internal senses in Latin, Arabic, and Hebrew philosophical texts. Harvard
Theological Review, 28(1), 69–133.
469
38
JOHN LOCKE AND
THOMAS REID
Rebecca Copenhaver
1. Introduction
As I write this, I remember the first time I saw the statue of David Hume on Edinburgh’s Royal
Mile. But what exactly am I remembering? We commonly think of such memories as about the
past, and in particular, about past experiences. On this view, what I remember now are sensa-
tions, feelings, thoughts, and ideas that I had on the day I saw Hume’s statue. The objects of
memory, on this story, are past experiences. But if we travel three hundred years into the past,
we find two philosophers—John Locke and Thomas Reid—whose theories of memory and
the role it plays in our lives are very different from these common assumptions about memory.
Locke presents remembering as an activity that may be, but need not be, about the past. His
remembering is a roomy notion. It includes recollection, but also contemplation, attention,
reverie, daydreaming, and study. This remembering needs past experience because it needs
ideas that have been already been formed. But remembering need not be directed at the past,
nor need it be mainly about the past. Remembering renews an acquaintance with things already
apprehended. But that acquaintance enables activities not tied to the past: contemplating a
problem, navigating an environment, attending to a story, or planning for the future. What such
activities need is not just present ideas formed in response to present impressions; another thing
they need, according to Locke, is remembering.
Like Locke, Reid holds that remembering renews acquaintance with things already appre-
hended. For both philosophers, then, remembering requires a past which, for Locke, it need
not address. Both Reid and Locke see remembering as releasing us from a persistent specious
present that would keep the world always new, yet leave it always alien. Without remember-
ing, the present would stay incomprehensible and the future would remain unimaginable.
Remembering makes the world less strange, reestablishing connections to things already
experienced, helping make sense of the present, and enabling a vision of the future.
Despite his agreements with Locke, Reid’s account of memory is more restricted: his ver-
sion of remembering is just to conceive of past events that have been witnessed, as having been
witnessed. Moreover, Reid imbeds memory within his direct realism: remembering directs us,
in the first instance, not to experiences of the world—past or present—but to the world itself.
Perception relates directly to present events and things; remembering preserves this relation by
renewing acquaintance with things and events previously apprehended.
470
John Locke and Thomas Reid
2. Locke on memory
Thus the ideas, as well as children, of our youth, often die before us: and our minds
represent to us those tombs, to which we are approaching; where though the brass
and marble remain, yet the inscriptions are effaced by time, and the imagery moulders
away. The pictures drawn in our mind, are laid in fading colors.
(Locke 1975, An Essay Concerning Human Understanding,
Book II, Chapter X, Section 5)
Locke, using a common distinction, divides the human mind’s powers into thinking and voli-
tion. “The power of volition is called the will,” and it acts by willing. “The power of thinking
is called the understanding,” and it acts by perceiving (Locke 1975: 128).1 Locke’s “perception”
is broader than ours. It includes all the powers of thinking—discerning, judging, reasoning,
imagining, and knowing—but also what we now call “perception.” This is Locke’s word for
the power to form ideas in response to sensible objects. Sensible objects make impressions
on our bodies, thereby affecting us through the senses. In response, the understanding forms
sensations, which are ideas of sensible qualities: “yellow, white, heat, cold, soft, hard, bitter,
sweet” (Locke 1975: 104–6, 117, 226–7). The power to form sensations or ideas of sense in
response to sensible objects belongs to the understanding, not the will: we do not exert this
power voluntarily (Locke 1975: 118). Whether an idea is a sensation depends on its origin—on
whether the idea makes its “entrance . . . into the understanding by the senses” or by another
route (Locke 1975: 226–7).
We know about these activities because we have the power to form not just ideas of sense
but also ideas of reflection. When the mind points outside—toward sensible objects, sensations
are formed. But when “the mind turns its view inwards upon itself, and contemplates its own
actions,” ideas of reflection are the result. We reflect by observing our own mental operations
and producing ideas in response to them. By sensing, we have distinct ideas of “white and red,
a square or a circle.” By reflecting, we have distinct ideas of attention, dreaming, reasoning,
judging, willing, and remembering—in short, “the different states of mind in thinking” (Locke
1975: 104–5, 226–8).
Perceiving—the power of thinking—includes what we would now call sensory percep-
tion. It also includes remembering, another activity of the understanding. Locke identifies three
broad types of memory. These are (1) remembrance—which includes recollection, contempla-
tion, reverie, attention, study, dreaming, and ecstasy; (2) retention, and, most controversially,
(3) consciousness of past actions and thoughts.
In Book II of the Essay, Locke presents two accounts of remembering separated from one
another by eight chapters on different topics: ideas of space, duration, extension, number, and
others. Both discussions of remembering—Chapter 10 “Of Retention,” and Chapter 19 “Of
the Modes of Thinking”—present taxonomies, but they do not match each other. Did Locke
write these two chapters to discuss the same topic—remembering—or did he intend distinct
accounts of retention (in Chapter 10) and remembrance (in Chapter 19)? Despite the mismatch,
the two chapters tell a similar story about memory, an account distinct from what Locke says
about consciousness of past thoughts and actions. Chapter 19, the later of the two presentations,
is about remembrance, and its framework is Locke’s distinction between ideas of sense and of
reflection, along with their origins.
Ideas of sense are of sensible qualities that originate in response to impressions made by sen-
sible objects on our bodies. Seeing a cloudless sky gives rise to a distinct idea of blue. The idea’s
origins make it a sensation: its immediate occasion is an impression made by sensible objects
471
Rebecca Copenhaver
on the senses. Ideas of reflection, by contrast, are of mental operations originating in response
to impressions made on our minds by those operations. When someone reflects on how she
solves a puzzle, for example, she may form a distinct idea of reasoning. Again, what makes this
idea one of reflection is its origin—from an impression made on the mind by the mental activ-
ity reflected upon. All of our ideas have their origin either in sensation or in reflection (Locke
1975: 104–6, 118–19).
Once an idea of sense has emerged from an impression—say the sensation of blue while
admiring the afternoon sky—such an idea can be formed again without help from any sensible
objects or impressions. I may have the idea of blue without any blue things affecting my senses.
Likewise with ideas of reflection: ideas about reasoning, once originated, no longer need puzzles
to stimulate them.
This ability—to form an idea of sensation or reflection without the impressions and objects
to which these ideas originally responded—is what Locke calls remembrance (Locke 1975: 226).2
Locke’s taxonomy in Chapter 19 classifies several types of remembrance by degree of activity.
Recollection, contemplation, attention, and study are the more active forms of remembrance.
The more passive forms include dreaming, reverie, and “extasy” (Locke 1975: 227–8).
Recollection is a success term for a voluntary effort to form an idea, like trying to recall the
face of an old friend, and accomplishing it. One type of contemplation (reverie is the other) is
also voluntary. Ideas are elusive—fleeting, successive, and non-continuous. In order to be well
contemplated, an idea must be “held . . . long under attentive consideration”—as when one
considers the idea of a triangle by way of grasping the Pythagorean theorem. Attention is also
a form of remembrance, whereby an idea is noticed “and, as it were, registered.” Registering
ideas enhances remembering: it strengthens the ability to form ideas in the absence of impres-
sions. Particular acts of remembrance—recollection, attention, and contemplation—reinforce
remembrance in general (Locke 1975: 227).
The most active remembrance is study. Locke also calls it “intention,” emphasizing the
mind’s persistent engagement in a voluntary effort that “will not be called off by the ordinary
solicitation of other ideas.” When studying “with great earnestness, and of choice,” the mind
considers an idea “on all sides,” while noticing its “relations and circumstances; and views every
part so nicely that it shuts out all other thoughts and takes no notice of the ordinary impressions
made then on the senses, which at another season would produce very sensible impressions.”
Even for adults, the world is often a blooming, buzzing confusion. But in study, the mind fixes
on an idea, freed for a moment from the inexorable succession of ideas suggested by the senses.
Like attention and contemplation, study focuses on ideas voluntarily. The price is less attention
to and recall for the ideas of sense that we might have registered had we not been in intense
study (Locke 1975: 227–8).
Reverie is another type of contemplation—in one way the opposite of study, though like
it in another. Study is willed and active, reverie is passive and involuntary: we lapse into
reverie. Like study, however, reverie effaces ideas of sense and reflection that might otherwise
command attention. Worries and fantasies, ideas replaying and repeating willy-nilly: these are
distractions, blocking what might be noticed in other circumstances. Ecstasy falls between
reverie and dreaming, and Locke has little to say about, leaving it “to be examined” whether
“dreaming with eyes open” is just a metaphor (Locke 1975: 227).
The contents of dreams are the clearest case of ideas formed without the impressions and
objects from which they originated. Dreams are disconnected from sense and reflection, or
nearly so, and so they are disconnected from the world. In dreams, ideas come and go invol-
untarily, unprovoked by impressions. Dreamers are “retired as it were from the senses,” able
to “sleep out whole stormy nights, without hearing the thunder, or seeing the lightning, or
472
feeling the shaking of the house.” Because dreams are sequences of ideas—confused, indistinct,
dim, and obscure though they may be—dreaming is a form of thinking, of perceiving in Locke’s
vocabulary. The succession of ideas stops only in dreamless sleep, which “closes the scene
quite and, puts an end to all appearances.” While Descartes holds that thinking is the principal
attribute of a mental substance, Locke claims that the mind need not always think. The variety
of different types of remembrance—from earnest study to impassive dreaming—shows that
thinking is an activity or operation, not an essence: activities come in degrees; essences do not
(Locke 1975: 228).
Locke’s taxonomy of remembrance is strange in one way. Except for recollection, no type
of remembrance is actually about the past, at least not explicitly. We may contemplate, attend,
study, ruminate, daydream, and dream about the past, but we need not. Often, we contemplate
problems that need solving, attend to things before us, study absorbing stories, ruminate on what
we need to do, daydream about the future, and dream fantastical worlds. We may turn to the
past to contemplate, attend, study, ruminate, daydream, or dream. But present problems need
contemplation; many things need attention right now; vexed issues can be studied; constant
worries are fodder for rumination; future fantasies are today’s daydreams, nightmares to follow;
in dreams begin tomorrow’s responsibilities. Remembrance is neither directed towards, nor
principally about the past.
Yet remembrance needs past experience. In particular, every act of remembrance, no matter
the type, requires an idea previously had from a sensory impression or from an impression of
reflection. This requirement is a version of the Previous Awareness Condition (PAC), devel-
oped and examined by Sydney Shoemaker among others, particularly in relation to episodic
memory: memories that are constrained by the necessary condition that the person remem-
bering was witness or agent to what is remembered (Shoemaker 1970; Parfit 1985; Malcolm
1977). Episodic memories are often contrasted with semantic memories, such as remembering
that Amelia Earhart was the first woman to fly solo across the Atlantic Ocean. The PAC that
holds for episodic memory does not constrain semantic memory: those who never witnessed
Earhart’s flight, remember (not just know) that Earhart was the first woman to fly solo across
the Atlantic.
Locke’s forms of remembrance also have a PAC: each type of remembrance consists in hav-
ing an idea formed previously in response to impressions of sense or reflection. In other words,
ideas had by contemplating, attending, studying, ruminating, daydreaming, and dreaming must
be ideas had already, in the past, formed from an original impression (Locke 1975: 227–8, 153).
This PAC is necessary but insufficient for remembrance, which needs something else: that in
contemplation, attention, study, and so on, we form ideas unconstrained by present impres-
sions. Recall that when we perceive, we form ideas of sensible qualities in response to present
impressions; when we reflect, we form ideas in response to present impressions made by our
own mental operations. But the ideas we form when we remember need not be in response to
present impressions. The ideas we form when we remember do not require impressions of the
same kind as the impressions of sensation or reflection that first produced those ideas in us. In
fact, the ideas we form when we remember need not be a response to any impressions at all,
even if we remember while various impressions occur all around us. Many objects and events
surround a person deep in study, making impressions on the senses, yet the studious train of
ideas, if it is studious, holds itself apart from those all-too-present impressions. In dreams, there
are no present impressions at all. Remembrance lost in dreams is “retired as it were from the
senses,” in a flood of ideas severed entirely from sense and reflection (Locke 1975: 227–8).
Putting the two conditions together: remembrance is (1) having ideas formed previously
in response to impressions and (2) having them regardless of whether impressions of the same
473
Rebecca Copenhaver
kind as were originally there when forming the ideas, are present. Along with the power to
form ideas of sense or reflection in response to impressions comes the power to form such
ideas again—with or without the originating impressions. Since these impressions precede any
remembrance of the ideas in question, the remembering mind, while receiving various present
impressions, needs—for a remembrance that may not be about the past—past experiences (Locke
1975: 227–8). Remembrance has a past and needs a past, but it need not be about the past.
The earlier discussion of memory in Locke’s Book II, in the chapter “Of Retention,”
presents a similar account of remembering with a different taxonomy. Here too, awareness of
the past is not required for remembering; remembering requires ideas already acquired, but no
longer needs the impressions that originally produced them. In this chapter, Locke studies two
types of retention: contemplation and memory.
Contemplation—also described in Locke’s first taxonomy—is more or less familiar: it sus-
tains an idea of sensation or reflection by voluntary effort, even after the initiating impression
has vanished, sunk in the stream of transient impressions whose flow never stops (Locke 1975:
149–55). The second type of retention—memory—is more elusive.
Locke introduces memory with two comparisons, both strictly metaphorical. First, memory
is a storehouse of ideas, “a repository” where the mind “lay[s] up those ideas, which at another
time it might have use of.” But ideas are impermanent, and their flow is inexorable: except in
dreamless sleep, they keep coming, one after another, though each is transitory; contemplation
may keep them at hand longer, but only briefly. And ideas, as particulars, can originate only
once. Since they “cease to be anything, when there is no perception of them,” they cannot
really be put into storage, and once they “cease to be anything,” they cannot come back into
existence. The impermanence and particularity of ideas puts a new constraint on remembering:
remembering is not having the very same token idea twice or more; remembering is having a
new and distinct idea which is a token of the same type as an old idea: in this way a token idea,
new and in the present, can have a past—the past of its type—without referring to the past
(Locke 1975: 150).
To keep his storehouse metaphorical, Locke claims that
this laying up of our ideas in the repository of the memory, signifies no more but this,
that the mind has a power, in many cases, to revive perceptions, which it once had,
with this additional perception annexed to them, that it has had them before.
(Locke 1975: 150; emphasis added)
Memory is an activity that the mind performs, not a container located in it, spatially or other-
wise. That much is clear. And perhaps Locke is not committed to the position that memory
requires two ideas—a past idea and a present idea of that idea as past. Rather, the contents of this
metaphorical container (1) have been had before and (2) are had now as having been had before.
But following the storehouse metaphor comes a second metaphor, this time about painting:
memory is now “an ability in the mind, when it will, to revive [ideas] again . . . and as it were
paint them anew on itself, though some with more, some with less difficulty; some more lively,
and others more obscurely” (Locke 1975: 150). Memory is an artist, making new ideas out of
familiar materials: old ideas already had. Once again, taken literally, the artist needs two ideas to
make memory: the freshly painted idea and the old idea newly revived.
But like the storehouse, the painter is a metaphor, from which Locke moves to an even more
abstract comparison: remembering is like seeing. The ability to remember brings “into sight”
objects now unseen by any sense. Once again, an idea, had in the past, acts in the present while
severed from its past:
474
And thus it is, by the assistance of this faculty [memory], that we are said to have all
those ideas in our understandings, which though we do not actually contemplate, yet
we can bring into sight, and make appear again, and be the objects of our thoughts,
without the help of those sensible qualities, which first imprinted them there.
(Locke 1975: 150)
Locke’s brief metaphor of memory as a painter suggests that there may be more to memory than
the past possession of one idea and the forming of a new one without help from the origina-
ting impressions. Consider this: walking down the street, I have an idea of a flavor but no present
impressions of taste. Long ago, I had another such idea, though I do not realize it now: the
taste of an unusual brand of toothpaste made a strong impression on me—an impression now
vanished. But the color of a passing car is exactly the color of the long-forgotten toothpaste:
the color revives the idea of its particular flavor. Why I have this idea I do not know, nor do I
recognize it as an idea I have enjoyed in the past. This seems less like remembering than having
an idea out of the blue. Even though I had a related idea long ago, the new idea does not strike
me as familiar.
Locke counters unfamiliarity with familiarity. The painter’s idea is new, but it captures
something recognized as past. Having ideas of the same kind, and having them without their
originating impressions, is not enough. The mind must also take “notice of them, as of a for-
mer impression, and renew its acquaintance with them, as with ideas it had known before”
(Locke 1975: 153). Having two ideas—one idea, and then another idea about having had that
idea before—is not the point. Instead, an original idea formed in response to impressions is an
apprehension. This apprehension makes it possible—by grounding a power or ability—to form
an idea of the same kind as the past idea but with or without its originating impression, thereby
renewing an acquaintance with what was previously apprehended.
Acquaintance comes in degrees. The sensible objects surrounding us and the mental opera-
tions on which we reflect are more or less familiar (Locke 1975: 150–52). We renew and
strengthen acquaintance with things not only through repeated impressions but also by remem-
bering, which gradually makes the world familiar. Without remembering—reacquainting
ourselves with things previously experienced, even in their absence—we would be trapped in
a specious present. Things would press on us again and again, and each idea formed in response
would be new and strange, giving us no purchase on the world. Without memory, endless
impressions would be mere chaos. Memory, therefore
is necessary in the next degree to perception. It is of so great moment, that where

it is wanting, all the rest of our faculties are in great measure useless: and we in our
thoughts, reasonings, and knowledge, could not proceed beyond present objects.
(Locke 1975: 153)
The most familiar ideas are those “oftenest refreshed”—those most important for renewing our
acquaintance with the world. They include “the original qualities of bodies, viz. solidity, exten-
sion, figure, motion, and rest,” along with secondary qualities such “as heat and cold.” Also
crucial are our basic mental operations and the most general “affections of all kinds of being”
such as existence, duration, and number. Remembering makes the world familiar and us at
home in it (Locke 1975: 152).
So what does Locke mean by remembering? He means that remembered ideas are ideas
of the same kind as ideas previously had, but the remembered ideas are had independently of
the impressions from which the original ideas were formed, so that remembered ideas renew
475
Rebecca Copenhaver
acquaintance with what was apprehended when the corresponding ideas were formed originally,
which renewed acquaintance makes what was apprehended, familiar. Remembering navigates
a present world stocked with familiar objects, making us easy with the workings of our own
minds and helping us plan a future.
When Locke’s treatment of memory comes up, the usual problem is personal identity. Locke’s
story inspired the Memory Theory (MT) of personal identity, which equates personal identity
with memory: sameness of episodic memory is necessary and sufficient for sameness of persons.
Here I am not concerned with theories of personal identity—neither Locke’s, nor the MT (see
Chapter 13 of the present volume). I am concerned with the activity on which Locke claims
personal identity depends: “that consciousness, which is inseparable from thinking . . . by [which]
everyone is to himself, that which he calls self ” (Locke 1975: 335). Is this consciousness the same
as remembering or different?
Consciousness—the activity that constitutes identity—can only be about the present,
according to Locke: consciousness is “always as to our present sensations and perceptions”
(Locke 1975: 115, 335–6). This secures the present identity of persons, what makes “self to
itself now” (emphasis added). And this identity may persist, as long as the activity responsible for
it—consciousness—persists. Consciousness persisting over time ensures persistence of persons.
Yet consciousness is not remembering: in the first instance, consciousness is aware of present
thought and actions, present sensations and perceptions. These present ideas need not have been
had before. From a first encounter with a strange thing comes an impression from which a new
idea emerges. Consciousness, for a time, attends the new idea. But remembering works with old
ideas to renew acquaintance with the world. Consciousness and memory both serve purposes
essential to thinking, but the purposes they serve are different.
Locke describes consciousness as “extending backwards,” however, to past thoughts and
actions. Is consciousness not confined to the present? Are we conscious of past thoughts, actions,
sensations, and perceptions? That would surely be remembering. What else could it be? But
the consciousness that “extends backwards” need not be about past sensations, perceptions,
thoughts, or actions. Alternatively, consciousness is confined to the present, as Locke insists,
but a possible object of present consciousness is an idea that renews acquaintance with something
apprehended in the past. Consciousness needs remembering to “unite existences” into the same
person. When ideas are remembered, they become conscious presently, but they are the same in
kind as ideas had previously, ideas present to consciousness at some past moment.
Remembering joins present consciousness with past consciousness. But remembering is not
consciousness of the past. And consciousness as such does not require remembering. Because
we remember, however, acquaintance with things apprehended consciously in the past may be
renewed in present consciousness, thereby joining one consciousness with another and uniting
“remote existences into the same person” (Locke 1975: 340–45). Without remembering, there
could be no reacquaintance with objects of past consciousness, which would put the continued
existence of persons at risk. But this does not make consciousness the same as remembering.
3. Reid on memory
It is by memory that we have an immediate knowledge of things past: The senses give
us information of things only as they exist in the present moment; and this informa-
tion, if it were not preserved by memory, would vanish instantly, and leave us ignorant
as if it had never been.
(Reid 2002, Essays on the Intellectual Powers of Man,
Essay III “Of Memory,” Chapter 1)
476
We move now almost one hundred years to a different account of memory, inspired in
opposition to a picture of the mind its author identified with Locke. Thomas Reid called this
picture he opposed, “the ideal theory,” or the “theory of ideas.” At its core is the Lockean
position that the immediate objects of perception are ideas (Locke 1975: 134). By contrast,
Reid argues that the mind is directed in the first instance not to itself or its own ideas, but to
the world. Accordingly, he presents a direct realist theory of memory: when we remember,
the mind is directed not towards ideas—be they ideas experienced in the past, or ideas of past
experiences—but to the events we experienced in the past.
The interpretation on which Locke’s remarks on personal identity should be read as a kind of
Memory Theory (MT), despite his claim that personal identity consists in consciousness, is not
new. In the Essays on the Intellectual Powers of Man (1785), Reid complains that “it is impossible to
understand the meaning of this, unless by consciousness be meant memory” (Reid 2002: 277).
Reid is famous for his Brave Officer objection to MT (Reid 2002: 276). However, the objec-
tion is not original to Reid. It appears in Berkeley’s Alciphron: or, the Minute Philosopher (1732)
and Reid himself attributes the example to George Campbell.3 Reid’s criticism of MT is that it
confuses truisms about the evidential relationship between memory and personal identity with
metaphysical conclusions about what makes a person identical with herself over time. But even
this insight is not new with Reid; it evokes the discussion in Butler’s influential dissertation “Of
Personal Identity” (1736). What is original to Reid is his own positive account of memory, and
we may investigate this independently of questions concerning personal identity.4
Though Reid constructs his account of memory in opposition to the theory of ideas, and
though he views Locke as a leading proponent of that theory, he shares with Locke the position
that remembering renews our acquaintance with things we apprehended in the past. It is an activity
that renders things familiar by reestablishing our connections to the things previously experienced:
Things remembered must be things formerly perceived or known. I remember the

transit of Venus over the sun in the year 1769. I must therefore have perceived it at
the time it happened, otherwise I could not now remember it. Our first acquaintance
with any object of thought cannot be by remembrance. Memory can only produce a
continuance or renewal of a former acquaintance with the things remembered.
(Reid 2002: 255)
Reid is interested primarily in episodic memory: the sort of memory whose PAC is that a person
(episodically) remembers an event only if she was witness or agent to the event remembered.
Reid could not have remembered the transit of Venus over the sun in the year 1769 unless he
had witnessed it. I cannot remember the transit of Venus in 1769 because I was not witness to
it, not having been born yet. I may have a semantic memory of the transit of Venus; I remember
that Venus crossed the sun in 1769. I learned this fact from reading Reid’s account of the event.
My semantic memory that Venus crossed the sun in 1769 is grounded in my previous aware-
ness of this fact. But I have no previous awareness of the event itself—Venus crossing the sun in
1769—and so I cannot remember the 1769 transit of Venus.
The PAC on episodic memory is a necessary but insufficient condition: if I have an experi-
ence as of having witnessed the transit of Venus in 1769 but I was not in fact witness to the
event (having been born in 1971), then such an experience is not an episodic memory. On the
other hand, I have been agent or witness to many events of which I have no episodic memory. I
witnessed the Perseid meteor shower in 1980, though I do not episodically remember the event.
I remember that I saw the Perseid meteor shower in 1980—my family tells me I was there—but
I do not remember the meteor shower itself.
477
Rebecca Copenhaver
So central for Reid is the kind of remembering that requires previous awareness of the
events remembered that he does not count semantic memory as a form of remembering,
properly speaking (Reid 2002: 264). When I remember that Venus crossed the sun in 1769,
I have a belief or knowledge, rather than a memory. Reid distinguishes such beliefs from the
phenomenon of remembering, because he holds that even when they are true and justified,
such beliefs do not play a role in preserving past apprehension of events. By contrast, Reid
remembers not only the transit of Venus in 1769, but also that Venus crossed the sun in 1769.
Unlike my belief that Venus crossed the sun in 1769, Reid’s belief that Venus crossed the sun
in 1769 preserves his past apprehension of the event itself. Though expressed propositionally,
his belief that Venus crossed the sun in 1769 does not report a semantic memory. It reports
his previous awareness of the transit of Venus and preserves his past apprehension of the event;
it reports an episodic memory.
According to Reid, remembering preserves past apprehension of events to which we were
agent or witness. Remembering is not a current apprehension. It is not a current apprehension
of a past experience, nor is it a current apprehension of a past event. We remember events, not
experiences of events. And we do not remember events by re-apprehending them. Rather, a past
apprehension is itself preserved in remembering. According to Reid, it is impossible to appre-
hend events in the past: apprehension is confined to the present (Reid 2002: 23, 253). Acts
of apprehension establish a direct relation to a present event (or object, or mental operation).
The direct relation established by apprehension is then preserved by acts of remembering the
event apprehended. Remembering is directed not towards past apprehension, but to what was
presented in the past apprehension: namely, the event itself (Reid 1997: 28).
Remembering preserves past apprehension by conceiving of an event previously appre-
hended and believing, of this event, that it happened to me. This belief is unlike the beliefs that
express semantic memories. This belief is a constituent of the act of remembering. According
to Reid, memory consists in a conception of a past event and a belief about that past event, that
it happened to the person who is represented in that memory as agent or witness (Reid 2002:
228, 232, 254, 257). This conception-belief structure mirrors the structure of other complex
mental operations that are directed at objects, such as perception, which is directed at things
presently before me, and consciousness, which is directed at my own mental operations. Each
of these operations consists in a conception and belief; the operations differ from one another
because the objects conceived differ, as do the beliefs about those objects (Reid 1997: 197). The
conception that is a constituent of memory is of an event previously apprehended. The other
constituent of memory is a belief, of the event conceived, that it happened. In particular, it is a
belief that the event happened to me (or was witnessed by me), where the “me” is indexed to
the person who is represented in the memory as agent or witness to the event.
Reid’s direct realist theory of remembering joins his direct realist theory of perception.
Memory, like perception, is directed towards the world rather than our experiences of the
world. When I perceive, I do not perceive my own ideas or experiences; I perceive things in
the world. When I remember, I do not (for the most part) remember my past experiences; I
remember the events I experienced in the past. In remembering those events, I do not currently
apprehend them. Memory is time-travel only metaphorically.5 Rather, because I apprehended
events in the past, I have the ability to preserve a direct relation to these events, by now con-
ceiving of them as having happened to me. In remembering such events, I have access to
more than what I presently perceive. Remembering presupposes past apprehensions but is not
directed towards them. Rather, it is directed towards the world, a world with which I am able
to reconnect by remembering the events that happened there, because I was there. We experi-
ence the world by perceiving it and remembering our path through it.
478
4. Conclusion
Locke and Reid hold that remembering renews acquaintance with the world by preserving
our relation to the things of which we were aware in the past. In remembering, we renew our
acquaintance with what we formerly apprehended. Remembering requires us to have a past, but
its role in our mental economy is firmly rooted in the present and future. By it, we navigate a
world in which we are increasingly at home. Were we unable to remember, we would ever be
to the world as strangers in a strange land.
Acknowledgment
I thank Shaun Nichols and Brian P. Copenhaver for comments and suggestions on earlier drafts.
Notes
1 I have modernized spelling and capitalization but not punctuation in quotations from Locke’s Essay.
2 In Book II, Chapter 19, Locke presents remembrance in terms of ideas of sense only, though it is clear
from the remainder of the chapter that remembrance includes ideas of reflection as well.
3 See Van Cleve (2015: 258, n20).
4 For a similar but more extensive treatment of Reid’s theory of memory, see Copenhaver (2006, 2014).
5 See Chapter 18, “Memory as Mental Time Travel,” by Kourken Michaelian and Denis Perrin, this volume.
Further reading
Folescu, M. (2016) “Remembering Events: A Reidean Account of (Episodic) Memory,” Philosophy and
Phenomenological Research 93(3). doi:10.1111/phpr.12333.
Hamilton, A. (2003) “Scottish Commonsense about Memory: A Defence of Thomas Reid’s Direct
Knowledge Account,” Australasian Journal of Philosophy 81(2): 229–45.
Lännström, A. (2007) “Locke’s Account of Personal Identity: Memory as Fallible Evidence,” History of
Philosophy Quarterly 24(1): 39–56.
Van Woudenberg, R. (2004) “Reid on Memory and the Identity of Persons,” in T. Cuneo and R. Van
Woudenberg (eds.), The Cambridge Companion to Thomas Reid, Cambridge: Cambridge University Press.
Weinberg, S. (2012) “The Metaphysical Fact of Consciousness in Locke’s Theory of Personal Identity,”
Journal of the History of Philosophy 50(3): 387–415.
References
Berkeley, G. (2008) Philosophical Writings, D.M. Clarke (ed.), Cambridge: Cambridge University Press.
Butler, J. (1736/2006) The Works of Bishop Butler, D.E. White (ed.), Rochester, NY: University of
Rochester Press.
Copenhaver, R. (2006) “Thomas Reid’s Theory of Memory,” History of Philosophy Quarterly 23(2): 171–87.
Copenhaver, R. (2014) “Reid on Memory and Personal Identity,” in E.N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy, http://plato.stanford.edu/archives/win2014/entries/reid-memory-identity/.
Locke, J. (1975) An Essay Concerning Human Understanding, P.H. Nidditch (ed.), Oxford: Oxford University
Press.
Malcolm, N. (1977) Memory and Mind, Ithaca, NY: Cornell University Press.
Parfit, D. (1985) Reasons and Persons, Oxford: Clarendon Press.
Reid, T. (1997) Inquiry into the Human Mind on the Principles of Common Sense, D.R. Brookes (ed.),
Edinburgh: Edinburgh University Press.
Reid, T. (2002) Essays on the Intellectual Powers of Man, D.R. Brookes (ed.), University Park, PA: The
Pennsylvania State University Press.
Shoemaker, S. (1970) “Persons and Their Pasts,” American Philosophical Quarterly 7(4): 269–85; reprinted in
Shoemaker, S. (1984), Identity, Cause and Mind, Cambridge: Cambridge University Press, pp. 19–48.
Van Cleve, J. (2015) Problems from Reid, Oxford: Oxford University Press.
479
39
DAVID HUME
Daniel E. Flage
David Hume explicitly discusses memory at two places in his Treatise of Human Nature (THN
1.1.3, 8–10; 1.3.5, 84–6).1 In both cases, he contrasts ideas of the memory with ideas of the
imagination. In this chapter, we examine those discussions within the context of the philosophi-
cal principles he developed in the Treatise. We show that Hume championed a causal theory of
the memory, that an idea of the memory represents the impression that was its original cause,
and that Hume argued that one can never be certain that a putative idea of the memory fulfills
those conditions.
Hume’s first discussion of the memory is in Book I, Part I, Section 3 of the Treatise. At
this point, he had drawn the impressions/ideas distinction on the basis of force and vivacity
(THN 1.1.1.1, 1–2), he had introduced the distinction between simple and complex percep-
tions (THN 1.1.1.2, 2), he had given a preliminary defense of his copy theory of ideas, “that all
our simple ideas in their first appearance, are derived from simple impressions, which are correspondent to
them, and which they exactly represent” (THN 1.1.1.7, 4; original emphasis), and he had introduced
the distinction between perceptions of sensation and perceptions of reflection (THN 1.1.2.1,
7–8). With this sparse conceptual framework, he turned to the distinction between ideas of the
memory and ideas of the imagination. He wrote:
We find, by experience, that when any impression has been present with the mind,
it again makes its appearance there as an idea; and this it may do after two different
ways: either when, in its new appearance, it retains a considerable degree of its first
vivacity, and is somewhat intermediate betwixt an impression and an idea; or when it
entirely loses that vivacity, and is a perfect idea. The faculty by which we repeat our
impressions in the first manner, is called the memory, and the other the imagination. It
is evident, at first sight, that the ideas of the memory are much more lively and strong
than those of the imagination, and that the former faculty paints its objects in more
distinct colours than any which are employed by the latter. When we remember any
past event, the idea of it flows in upon the mind in a forcible manner; whereas, in the
imagination, the perception is faint and languid, and cannot, without difficulty, be
preserved by the mind steady and uniform for any considerable time. Here, then, is a
sensible difference betwixt one species of ideas and another.
(THN 1.1.3.1, 8; original emphasis)
480
David Hume
Given the copy theory, the simple components of ideas of both memory and imagination are
copied from earlier impressions. Hume distinguishes ideas of the memory from ideas of the
imagination in terms of the greater force and vivacity of the former relative to the latter, sug-
gesting that the force and vivacity of an idea of the memory is less than that of an impression and
greater than that of an idea of the imagination. This provides the empirical ground for the dis-
tinction. He implicitly accepted the presumption of pre-philosophical common sense that ideas
of the memory provide information about one’s past, while ideas of the imagination do not. His
allusion to the higher degree of force and vivacity of an idea of the memory relative to an idea
of the imagination anticipates his celebrated remark that “An opinion, therefore, or belief, may
be most accurately defin’d, a lively idea related to or associated with a present impres-
sion” (THN 1.3.7.7, 96). Ideas of the memory are believed; ideas of the imagination are not.
Belief alone is not enough to distinguish an idea of the memory from an idea of the imagi-
nation. Beliefs can be about the past, the present, or the future, and our pre-philosophical
common sense takes memory to concern the past. So, ideas of the memory must represent past
states. Hume’s second characteristic of ideas of the memory takes this into account:
There is another difference betwixt these two kinds of ideas, which is no less evident,
namely, that though neither the ideas of the memory nor imagination, neither the
lively nor faint ideas, can make their appearance in the mind, unless their correspond-
ent impressions have gone before to prepare the way for them, yet the imagination
is not restrained to the same order and form with the original impressions; while the
memory is in a manner tied down in that respect, without any power of variation.
(THN 1.1.3.2, 9; cf. 1.1.3.3, 9)
Memory is constrained to the same “order and form” or “form . . . order and position” (THN
1.1.3.3, 9) as their correspondent impressions. Ideas of the memory are taken to represent earlier
impressions. Here Hume’s discussion can be tied directly to his principles of the association of
ideas.2 Insofar as an idea represents an impression—or anything else—it must resemble it (cf. THN
1.1.1.3, 3; 2.3.10.2, 448), but resemblance alone is not enough. It could happen that the idea I
form of a moment in the Battle of Hastings resembles events that happened, but since I could
not have seen that battle, it could only be an idea of the imagination. What more is needed? We
need something that establishes the temporal position of the impression relative to the idea of the
memory.3 The copy theory of ideas requires that simple ideas are caused by and perfectly resemble
simple impressions. So, there is a sense in which all simple ideas are ideas of the memory, but
Hume seems to be concerned with a more typical case of the memory, for example, remember-
ing the game of billiards I played yesterday.4 Hume, like the rest of us, assumed that there is a
one-to-one correspondence between a determinate effect and a determinate cause. So, an idea of
the memory, as an effect, corresponds to a determinate impression that originally caused it.5 The
original cause occurred at some determinate time in the past. So, causal considerations will pick
out the impression that originally caused the idea of the memory.6 Notice that Hume’s account of
the memory parallels his copy theory of simple ideas: just as there is some impression that was the
original cause of and perfectly resembles any simple idea, so there is some impression that caused
and resembles an idea of the memory. This ties ideas of the memory to the sparse conceptual
framework Hume had developed by this point in the Treatise.
Before examining Hume’s second discussion of the memory, a problem should be noted. Even
if the account above works for cases of remembering impressions of sensation, it is not clear that it
works for remembering all previous psychological states, such as dreams (Traiger 1985: 169–70).
The content of at least some dreams must be ideas of the imagination (cf. EHU 12.11, 153).
481
Daniel E. Flage
The only way one could remember a dream as a dream is if Hume followed Locke and held that
impressions of reflection differentiate occurrent mental states, so that remembering a dream would
involve an idea of reflection that corresponds to a dream state with an associated sensitive or emo-
tive content. Hume’s post-Treatise writings suggest he might have entertained such a position,7 but
it is not found in the Treatise. Since Hume’s sole discussions of the memory are in the Treatise, we
shall not speculate about how Hume might have revised his position.
Hume’s second discussion of the memory occurs within the context of a more general dis-
cussion of causation. After noting that “As to those impressions, which arise from the senses, their
ultimate cause is, in my opinion, perfectly inexplicable by human reason, and it will always
be impossible to decide with certainty” (THN 1.3.5.2, 84; original emphasis), Hume turns to
memory. He wrote:
When we search for the characteristic, which distinguishes the memory from the
imagination, we must immediately perceive, that it cannot lie in the simple ideas it
presents to us; since both these faculties borrow their simple ideas from the impres-
sions, and can never go beyond these original perceptions. These faculties are as
little distinguished from each other by the arrangement of their complex ideas. For,
though it be a peculiar property of the memory to preserve the original order and
position of its ideas, while the imagination transposes and changes them as it pleases;
yet this difference is not sufficient to distinguish them in their operation, or make us
know the one from the other; it being impossible to recall the past impressions, in
order to compare them with our present ideas, and see whether their arrangement
be exactly similar. Since therefore the memory is known, neither by the order of its
complex ideas, nor the nature of its simple ones; it follows, that the difference betwixt
it and the imagination lies in its superior force and vivacity. A man may indulge his
fancy in feigning any past scene of adventures; nor would there be any possibility
of distinguishing this from a remembrance of a like kind, were not the ideas of the
imagination fainter and more obscure.
(THN 1.3.5.3, 85; original emphasis)
The earlier discussion distinguished ideas of the memory from ideas of the imagination empiri-
cally on the basis of degrees of force and vivacity, but also it stressed the theoretical difference
between the two on the basis of resemblance and causation. So, while we might distinguish an
idea of the memory from an idea of the imagination on the basis of its greater degree of force
and vivacity, it was resemblance to an earlier impression that caused it that made an idea an
idea of the memory. This passage focuses on the practice of distinguishing ideas of the memory
from ideas of the imagination. Hume indicates that, in practice, we cannot distinguish an idea of
the memory from an idea of the imagination on the basis of any of its constitutive ideas. Nor
can we establish that a presumptive idea of the memory resembles the impression from which
it was derived, since it is impossible to compare an idea of the memory with any impression.
So, in practice, the sole basis for distinguishing between an idea of the memory and an idea of
the imagination is in terms of the greater force and vivacity of the former idea relative to the
latter. But as Hume continues, he argues that force and vivacity are not a wholly reliable basis
for drawing the distinction in practice:
The more recent this memory is, the clearer is the idea; and when, after a long interval,
he would return to the contemplation of his object, he always finds its idea to be much
decayed, if not wholly obliterated. We are frequently in doubt concerning the ideas
482
David Hume
of the memory, as they become very weak and feeble; and are at a loss to determine
whether any image proceeds from the fancy or the memory, when it is not drawn in
such lively colours as distinguish that latter faculty.
(THN 1.3.5.5, 85)
While we noticed the parallel between the copy theory of ideas and Hume’s account of the
memory, it is not a perfect parallel. While the copy theory requires that there is a perfect resem-
blance between a simple idea and the impression that was its original cause, in the case of ideas
of the memory the resemblance is often imperfect. Hume’s allusion to the clarity of the idea
and the tendency for an idea to “decay” over time—to be less clear, to less perfectly resemble
the impression that caused it—reflects this point. Further, the force and vivacity of the idea
also diminishes as the impression remembered retreats into the past, making it, in practice, an
imperfect basis for distinguishing ideas of the memory from ideas of the imagination. To make
matters worse, there are conditions in which ideas of the imagination can gain force and vivac-
ity and make them indistinguishable from ideas of the memory, for example, “in the case of
liars; who by the frequent repetition of their lies, come at last to believe and remember them,
as realities” (THN 1.3.5.6, 86).
So, it seems that we are left with the following situation. Ideas of the memory represent
(resemble to some degree) the ideas that originally caused them, but we cannot compare an
idea to an impression, so that we cannot know that there is a resemblance. The sole basis for
distinguishing ideas of the memory from ideas of the imagination in practice is in terms of the
relatively greater degree of force and vivacity of the ideas of the memory with respect to ideas
of the imagination. But force and vivacity are an unreliable basis for drawing the distinction in
practice, since ideas of the memory can lose force and vivacity and ideas of the imagination can
gain force and vivacity. So, on the theoretical side, we can never know that an actual idea of the
memory resembles its presumptive cause; on the practical side, we can never know that an idea
is actually an idea of the memory rather than an idea of the imagination.
What can we say? Shall we join the legions of commentators who lament the gross inad-
equacies of Hume’s account of the memory? (See, for example, Holland 1954; Benjamin 1964:
172–4; Passmore 1980: 93–7; Noxon 1976: 271; McDonough 2002.)8 Perhaps not.
When evaluating the philosophical work of a historical figure, we should ask three kinds
of questions. First, is the interpretation we have developed the most philosophically plausible
interpretation that is consistent with the primary texts? The careful historian might claim that
this question is always open. Second, what question is the philosopher addressing? For example,
is it the same question that a contemporary philosopher would ask regarding the philosophical
issue? Third, is the account we have developed consistent with the philosopher’s general prin-
ciples and the foci of the philosopher’s works?
Regarding the second question, we should notice that Hume’s question might not be the
same as that posed by the contemporary epistemologist of the memory. The contemporary phi-
losopher is likely to begin with the assumption that memory provides us with some knowledge of
the past and proceed to develop a theoretical account that justifies that assumption. Such a one
often examines the inadequacies of several theories with an eye to developing a more adequate
theory. It is not clear that this was Hume’s concern. Hume’s question concerned the distinction
between ideas of the memory and ideas of the imagination. If there was a presumption that ideas
of the memory are true, it was no more than a presumption of pre-philosophical common sense.
His concern was not to show that the presumption is true, although his discussion of the form
and order of the ideas of the memory indicate conditions under which an idea of the memory
would be true.
483
Daniel E. Flage
Turning to the third question, we note that our account is consistent with Hume’s attempt
to construct a “science of man” (THN Introduction.4.xv) on the basis of the principles of the
association of ideas: contiguity in space and time, resemblance, and causality. Hume used these
principles to explain why we hold some of our most basic beliefs, such as the belief in the exist-
ence of the external world (THN 1.4.2), the belief in material substance (THN 1.4.3), and the
belief in the identity of the self over time (THN 1.4.6). The explanatory program is juxtaposed
to and usually preceded by Hume’s philosophical examinations of the epistemic warrant of the
beliefs, examinations showing that the belief in question is unwarranted or false. So, examined
within the context of Hume’s general program in the Treatise, the skeptical conclusions in
Hume’s second discussion of ideas of the memory are not surprising. Of course, his conclusions
imply that, in practice, we simply have beliefs, some of which presumably correspond to past
(complex) impressions, and many of which do not. What do we do?
According to Hume, we construct two kinds of realities from our beliefs (THN 1.3.9.3, 108).
One reality is a system of ideas of the memory and present impressions (the self). The other
reality is a causal system beyond the self that Hume calls judgment (THN 1.3.9.3, 108), and
“’Tis this latter principle, which peoples the world, and brings us acquainted with such exist-
ences as, by their removal in time and place, lie beyond the reach of the senses and memory”
(THN 1.3.9.4, 108). Presumably, the sole constraint on these constructions is coherence
(cf. THN 1.3.5.2, 85; 1.1.6.8, 15).
So our account reflects what Hume says in his two explicit discussions of the memory. Ideas
of the memory are forceful and vivacious, that is, believed. If they are true, they represent
an impression that caused and resembles the idea, although we can never be certain that any
ideas of the memory are true. This account ties Hume’s account of the memory to his prin-
ciples of the association of ideas, and it seems to be consistent with the principal elements of
Hume’s philosophy. On our reading, however, Hume’s account of the memory is limited to
past impressions of sensation and, perhaps, past emotional states, but his system in the Treatise
does not seem to account for memories of such psychological states as dreams. It clearly does not
explain memories of how to do things or the role of memory in linguistic meaning.
Not everyone will agree with the details of our account of Hume on memory. Many will
assume that Hume’s principal concern was epistemological rather than psychological. Some will
argue that the discussion in Treatise 1.3.5, that there are no means by which we can know that an
idea of the memory resembles the impression that caused it, is a de facto rejection of the causal-
resemblance condition (Johnson 1987; McDonough 2002: 78–84), but most scholars will grant
that Hume’s account follows at least the general outlines we have sketched.9
Notes
1 We follow the standard referencing form for the Treatise (THN) by book.part.section.paragraph in
Hume (2000) and page in Hume (1978). References to the Appendix to the Treatise are marked App.,
followed by the paragraph in Hume (2000) and the page in Hume (1978). Similarly, references to the
Enquiry Concerning Human Understanding (EHU) will be by section.paragraph in Hume (1999) and
page in Hume (1975).
2 The Hume of the Treatise held that all beliefs can be explained by the principles of the association of
ideas: contiguity in space and time, resemblance, and causality (THN 1.1.4). While he officially retained
the position in his Enquiry Concerning Human Understanding (EHU 3), his associationist psychology played
a far more significant role in the Treatise than in the first Enquiry. In the Treatise, Hume uses the principles
of the association of ideas to explain why one holds such unwarranted beliefs as the belief in necessary
connection (THN 1.3.14), the belief in an external world (THN 1.4.2), the belief in material substance
(THN 1.4.3), and the belief in personal identity (THN 1.4.6).
484
David Hume
3 The reader will notice that I have not discussed “order.” It is because I consider “order” ambiguous
between concerns with the arrangement of simple ideas in a complex idea and concerns with position
in time. THN 1.1.3.2, 9 can be understood as a concern with the arrangement of simple ideas within
a complex idea, what Hume also seems to mean by “form.” THN 1.1.3.3, 9, where Hume writes, “The
chief exercise of the memory is not to preserve the simple ideas, but their order and position,” seems to
understand “order” in terms of position in time, a point supported by Hume’s discussion of a historian’s
ability to discuss historical events outside their temporal order, though the historian “takes notice of this
disorder, if he be exact” (THN 1.1.3.3, 9).
4 We leave open the question whether one can have an idea of the memory that is composed solely of a
simple idea.
5 This original cause is not to be confused with the impressions that trigger the memory, THK 1.3.5.4,
627–8.
6 Daniel Flage has argued that “ideas of the memory [are] relative ideas corresponding to definite descrip-
tions of the following general form, ‘the impression that is the (original) cause of an exactly (or closely)
resembles m,’ where ‘m’ denotes a particular positive idea” (Flage 1984: 172). While the account above
might bear similarities to Flage’s, we avoid calling ideas of the memory “relative ideas” because, (1) as
Leslie Friedman has shown, Hume does not so characterize them (Friedman 1993: 183), and (2) the more
recent use of “relative ideas” by Galen Strawson (Strawson 1989) and others is not consistent with Flage’s
usage (Flage 2007).
7 It is most evident regarding belief, since he seems to have shifted from the force and vivacity as a char-
acteristic of a believed idea of sensation or reflection to a feeling or sentiment that accompanies an idea of
sense or reflection (App. 3, 624; EHU 5.10–13, 47–50). If Hume marked belief by a feeling or sentiment
(an impression of reflection), it would be consistent for Hume to have differentiated other psychological
states in the same way. See Hanson (1988); Flage (1990: 168–86).
8 These commentators will also point to Hume’s verbal infelicities: ideas of the memory are “equivalent to
impressions” (THN 1.3.4.1, 82), his allusions to an “impression of the memory” (THN 1.3.5.1, 84), and
the “repetition of that impression in memory” (THN 1.3.5.7, 86).
9 The most notable exception is Azeri (2013). Azeri argues that there is a distinction between natural
or representative memory—which corresponds to the account we have sketched—and memory via
signs. Since Azeri incorporates elements of the Treatise beyond the discussions of the memory, elements
that Hume relates to history, we do not believe it reflects a proper extension of Hume’s notion of the
memory.
References
Azeri, Siyaves (2013), “Hume’s Social Theory of the Memory,” Journal of Scottish Philosophy 11(1):
53–68.
Benjamin, B. S. (1964), “Remembering,” in Donald F. Gustafson, ed., Essays in Philosophical Psychology,
Garden City, NY: Doubleday Anchor Books, pp. 171–94.
Flage, Daniel E. (1984), “Hume on Memory and Causation,” Hume Studies, 10th Anniversary Issue,
pp. 168–88.
Flage, Daniel E. (1990), David Hume’s Theory of Mind, London: Routledge.
Flage, Daniel E. (2007), “Relative Ideas Re-viewed,” in The New Hume Debate, eds. Rupert Read and
Kenneth A. Richman, revised edn., London: Routledge, pp. 138–53.
Friedman, Leslie (1993), “Another Look at Flage’s Hume,” Hume Studies 14: 177–96.
Hanson, Stacy J. (1988), “Impressions of Belief,” Hume Studies 14: 277–304.
Holland, R. F. (1954), “The Empiricist Theory of Meaning,” Mind 63: 464–86.
Hume, David (1975), Enquiries Concerning Human Understanding and Concerning the Principles of Morals, ed.
L. A. Selby-Bigge, 3rd edn., revised P. H. Nidditch. Oxford: Clarendon Press.
Hume, David (1978), A Treatise of Human Nature, ed. L. A. Selby-Bigge, 2nd edn., revised P. H. Nidditch.
Oxford: Clarendon Press.
Hume, David (1999), An Enquiry Concerning Human Understanding, ed. Tom L. Beauchamp, Oxford
Philosophical Texts, Oxford: Oxford University Press.
Hume, David (2000), A Treatise of Human Nature, eds. David Fate Norton and Mary J. Norton. Oxford
Philosophical Texts, Oxford: Oxford University Press.
Johnson, Oliver (1987), “‘Lively’ Memory and ‘Past’ Memory,” Hume Studies 13: 343–59.
485
Daniel E. Flage
McDonough, Jeffrey K. (2002), “Hume’s Account of the Memory,” British Journal for the History of
Philosophy 10(1): 71–87.
Noxon, James (1976), “Remembering and Imagining the Past,” in Hume: A Re-Evaluation, eds. Donald
W. Livingston and James T. King, New York: Fordham University Press, pp. 270–95.
Passmore, John (1980), Hume’s Intentions, 3rd edn., London: Duckworth.
Strawson, Galen (1989), The Secret Connexion: Causation, Realism, and David Hume, Oxford: Clarendon
Press.
Traiger, Saul (1985), “Flage on Hume’s Account of the Memory,” Hume Studies 11: 166–72.
486
40
G. W. F. HEGEL
Valentina Ricci
Recollection and memory are not among the themes Hegel is mostly famous for. However,
this does not mean that his discussion of recollection and memory cannot be of interest for a
philosophical reflection on these topics. On the contrary, as I aim to show in this chapter, recol-
lection in particular plays an important role within the system, although its significance has not
always been recognized by the scholarship.1 This might be partially explained by the fact that
there is no single systematic locus where Hegel discusses it: rather, recollection appears at several
points within the system itself, usually crucial points of transition between different parts of it.
A careful examination of such places can illuminate the vital role recollection plays for spirit’s
freedom and actualization and for the attainment of objective thought. Recollection, as I will
show, is a key function in that it allows spirit to internalize the spatio-temporal determinateness
of the content of its experience and to attain the stage of thought (in the Psychology) and the
stage of the concept in its purity (in the transition between the Phenomenology and the Logic),
that is, the content of experience that has been purified by its determinateness. In what follows,
I present what I consider the three most important discussions of recollection and memory in
Hegel’s work, namely those we can find in the Psychology, the Phenomenology of Spirit, and the
Logic.2 My aim is to illuminate the function of these concepts in Hegel’s system and to show
how this task can both support a better understanding of Hegel’s philosophy as a whole and offer
some insights to a general philosophical reflection on memory as such.
To begin with, it is worth keeping in mind that Hegel uses the concepts of recollection
(Erinnerung) and memory (Gedächtnis) in different ways, although it can be argued that their
functions are analogous and strictly related to the attainment of spirit’s freedom. If we look
at the systematic place where these concepts are introduced—the Psychology section of the
Encyclopedia3—we can see that their specific cognitive functions are quite distinct, but related.
The part of the Psychology where the two concepts come into play is that concerned with
the “theoretical spirit,” where Hegel discusses the cognitive activity of spirit as intelligence.
Recollection constitutes the second of three stages in which representation (Vorstellung) is
articulated. In what follows, I will say more about the workings of recollection; for now, suf-
fice it to say that it is concerned with the interiorization of the data coming from the senses
and the subsequent formation of images, whereby intelligence becomes active with respect
to its content and starts displaying the capacity to operate in the absence of external stimuli.
Memory (Gedächtnis), on the other hand, can be found only at a later stage. Hegel reserves this
487
Valentina Ricci
term for the process by virtue of which intelligence develops and acquires language. Memory
is the function that enables the subject to fully acquire the connection between the words
(“names”) and their meanings but also, in its final stage (as “mechanical memory”), to let go
of that connection and to become the master and author of meaning itself (EN, § 463). This
is a key function because intelligence, free from the concern with nouns and their meanings,
becomes able to think freely, that is, to produce its own content.
Recollection and memory thus seem to share similar conceptual structures, mainly related
to processes of interiorization of a cognitive content and the ability to recall it. The specific
function of memory, however, seems to be limited to the sphere of language, where its activ-
ity is focused on internalizing the connection between the linguistic sign—characterized by
arbitrariness and thus lacking any intrinsic meaning—and the spiritual content it represents,
thus making that connection a possession of intelligence. Recollection, on the other hand,
resurfaces in many important points of Hegel’s philosophy. I suggest to focus on the latter
for two main reasons: first, Hegel mostly employs this concept, and not that of memory,
when he discusses an important speculative function. Second, and relatedly, the German term
Erinnerung includes a reference both to interiorization and to the act of recalling the content
that has thereby been preserved. Focusing on recollection rather than on memory can thus
provide insight into important parts of the system and into some of the most debated issues
concerning Hegel’s thought.4
1. Recollection in the Psychology

According to Hegel’s discussion of the “theoretical spirit” (EN, §§ 445–68), intelligence at the
first stage of its development “finds itself determined” (EN, § 445; original emphasis). In other
words, it entertains a passive relation with the data it receives from the outside. Hegel describes
the development of intelligence in terms of a “positing what is found as its own” and of a “refu-
tation . . . of the semblance of finding the rational” (EN, § 445). This development takes place
through a dialectical interaction between activity and passivity, whereby what first appeared as
immediate, as a given, is gradually deprived of its immediacy and outwardness, and turned into
the possession of intelligence. The three main stages of this process, which correspond to the
three essential activities of intelligence, are intuition, representation, and thought.
Intuition [Anschauung] is the first stage of the activity of intelligence and, as such, the rela-
tionship with its object is initially characterized by a substantial passivity. This is the stage of
sensation [Gefühl or Empfindung], where the knowing subject is simply affected by the object:
when Hegel writes that the object is “found” or “given” for the intelligence, the reference is
precisely to sensation. However, as soon as the sensation becomes the object of attention, intel-
ligence starts emancipating itself from its initial passivity.
Attention [Aufmerksamkeit], in this sense, is a key moment because it is the first negation of the
immediacy in which the object is for the subject, and in this sense it is the first form of self-
determination of intelligence, which independently resolves to focus on a specific object or
set of data among the many it is presented with. Paying attention to a sensation implies making
a choice among all of the present sensations: for Hegel, this is an act of such great importance
that without it “nothing is for spirit” (EN, § 448).5 Without attention, indeed, the subject would
lose itself in the infinite multiplicity of the sensations it is exposed to. Thanks to attention, spirit
starts recognizing the object as its own, that is, as the object of its cognition and thus as no
longer entirely extraneous to itself. On the other hand, however, the object is still “the other of
itself” (EN, § 448; original emphasis), which has an independent existence and therefore is still
extraneous to spirit in important ways. This sense of extraneousness will accompany spirit, to a
488
G. W. F. Hegel
gradually lesser extent, until the completion of the cognitive process, that is, when the object
will be in intelligence’s full possession, as its own product and as the outcome of its own activity.
The third and last moment of this stage is that of intuition proper. Its outcome is a unity, a
“cohesive fullness of determinations” (EN, § 449, Zusatz; original emphasis), an individual object.
Intuition is the moment mediating the objectivity of sensation with the subjectivity of attention,
in which the first side, however, still prevails. It is only with representation that intelligence
knows the intuition as its own and is conscious of its own activity through its products. In the
Zusatz to § 450, Hegel explains the difference between intuition and representation in terms
of a temporal sublation—an aspect that will be central to the activity of recollection: intuition
becomes something past as it becomes the object of representation, and it is in this sense that
we say “I have seen this.” While the participle in this expression points at the past character of
intuition, the word “have” used in the perfect tense adds the reference to a present in which the
subject owns the product of that intuition, albeit in a different form, that is, the form given to
it by representation. The representational activity of intelligence unfolds in three moments, the
first of which is, significantly, recollection:
Intelligence, in first recollecting the intuition, puts the content of feeling in its inward-
ness, in its own space and its own time. In this way the content is an image, liberated from
its initial immediacy and abstract individuality in contrast to other things, as received
into the universality of the I in general.
(EN, § 452; original emphasis)
When the intuition is recollected, intelligence appropriates it, and this is precisely the moment
in which the transition to representation takes place: this is a key moment for intelligence
because, by recollecting, it severs all bonds to the immediacy of the present intuition. The aspect
of greatest freedom characterizing the cognitive activity at this stage becomes evident in the
relation with space and time: intelligence, through recollection, turns the outward, natural space
and time into inward, ideal space and time, which are now the subject’s space and time.
When the intuition is appropriated by intelligence through the first moment of recollection,
its determinations are sublated, transfigured, turned into determinations posited by the subject,
and that which was an intuition becomes an image [Bild] produced by the “universality of
the I.” What emerges from this activity is a first universalization or generalization of the data
provided by intuition, the formation of “schemes,” as it were, that no longer have the spe-
cific determinateness of intuition and thus the same sensuous “richness,” but make the subject
independent from the immediate presence of the object, which was absolutely necessary in the
stages of sensation and intuition. In addition to ensuring the subject’s freedom from sensuous
immediacy, recollection confers duration to the data of intuition. This is accomplished by the
second moment of recollection, whereby intelligence preserves the image—whose existence is
ephemeral—by storing it in the unconscious. The image can thus resist time, which otherwise
dissolves in its uninterrupted flow everything that intelligence does not capture and preserve in
the “nocturnal pit in which is stored a world of infinitely many images and representations, yet
without being in consciousness” (EN, § 453, Remark).
The famous metaphor of the “nocturnal pit” refers to the supply of images, a “world” that
the subject, through the twofold activity of interiorization and recollection, creates by and for
itself. At first, the subject is unable to evoke these images at will,6 but needs a present intuition
in order to be able to recall them. Recollection proper, in this sense, is the relationship between
the inwardly preserved image and the present intuition, the “subsumption of the immediate
individual intuition under what is universal in form, under the representation which is the same
489
Valentina Ricci
content” (EN, § 454; original emphasis). Intelligence can recognize new intuitions as belong-
ing to a specific “class” identified by the (preserved) image, and verify the image, in turn, by
comparing it with the relevant intuition, in a relationship of mutual verification. This leads to
the third moment of recollection, signaled by the emerging of images from the nocturnal pit of
intelligence and by the capacity of the latter to recall the images even in the absence of the exter-
nal stimulus constituted by the present intuition. In this way, according to Hegel, intelligence
reaches the stage of representation proper, where the content of intelligence (the image) is at the
same time placed back outside the subject itself. This is a crucial achievement, as recollection
enables intelligence to work freely: its activity is no longer bound by a given content, but it can
start operating with its own content. This will take place through the production of symbols,
of signs, and eventually of language, which will lead intelligence to the final stage of thinking,
where the product of intelligence will coincide with, indeed will be, “the thing,” the identity of
subjectivity and objectivity (EN, § 465).
Although Hegel’s characterization of recollection as the relationship between the preserved
image and the present intuition is quite standard, it is important to place it in the context of
intelligence’s development from passivity to activity and thereby to its freedom. In addition, the
conceptual structure I just presented is central for an understanding of the role of recollection
in the system as a whole: in particular, we will see that recollection is an essential component of
various processes whereby spirit becomes free and achieves the identity of the subjective and the
objective side of its thinking. The next context in which recollection plays a key role, and one
that “sets the tone” for the entire development of Hegel’s system, is the Phenomenology of Spirit
and in particular its last chapter. Here, Hegel presents the notion of absolute knowing, that is,
the standpoint of scientific, philosophical knowledge.
2. Recollection in the Phenomenology

The Phenomenology of Spirit may be regarded as the conceptual or philosophical narrative of the
development of consciousness into spirit and the forms of its knowledge and self-knowledge
characterizing the different stages of such development. This path leads consciousness from
a state in which it regards its experiences as extraneous to itself—that is, it cannot recognize
itself in them—to one, absolute knowing, in which this process of recognition and identifica-
tion is completed. The separation of consciousness from its object produces a set of inadequate
forms of knowledge, whose necessary sublation leads from one stage of the narrative to the
next. In the course of the phenomenological path, then, consciousness gradually frees itself
from the extraneousness of its object and, in absolute knowing, finally grasps the entirety of
its experiences (which thus becomes “experience” in the singular, as opposed to a multiplicity
of disconnected experiences) in a unitary, integrated, and coherent form in which all of the
different experiences are included and comprehended as necessary stages within the process of
its own development.7
This, very roughly, is the structure that defines absolute knowing and that can be expressed,
in technical terms, as the identity of consciousness and self-consciousness, being and thought,
subject and object. Recollection plays a key role in the attainment of the absolute standpoint, as
Hegel emphasizes in the last chapter: “As [spirit’s] fulfillment consists in perfectly knowing what it
is, in knowing its substance, this knowing is the withdrawal into itself in which it abandons its outer
existence and gives its existential shape over to recollection” (PhG, § 808; original emphasis).
The completion of spirit’s development coincides with the knowledge spirit has of its
self and its substance, that is, of its own existence and experience. Spirit achieves complete
490
G. W. F. Hegel
knowledge of itself precisely through the interiorization of its existence and experience,
whereby it becomes able to detach itself from the immediacy of these dimensions and to pre-
serve them by placing them in a new dimension: such dimension is one of recollection, upon
which, I submit, the last part of the “Absolute Knowing” chapter, and thereby the phenom-
enological path itself, is built. This is particularly relevant in relation to the position and role of
this work with respect to the system as a whole: precisely at the point where Hegel places the
transition to the developed system, for which the Phenomenology is supposed to lay the founda-
tion, this happens through the recollection and interiorization of spirit’s past experience, insofar
as it grants access to the conceptual dimension and therefore to the dimension of science.
The process at work here is one whereby spirit’s experiential content is preserved in its
interiority, and therefore is sublated (aufgehoben): this means that experience is no longer given
to spirit in the immediacy of the present, but at the same time it is preserved within spirit itself,
and more specifically it is preserved in its knowledge, whose horizon constitutes a new world, a
world mediated by spirit’s cognitive activity:
Recollection, the inwardizing [Er-innerung], of that experience, has preserved it and is

the inner being, and in fact the higher form of the substance. So although this spirit
starts afresh and apparently from its own resources to bring itself to maturity, it is none-
theless on a higher level that it starts.
(PhG, § 808; original emphasis)
Recollection plays the essential role of preserving spirit’s experience and of mediating it by mak-
ing it spirit’s own content. The higher status of this content is due to the fact that it is no longer
something that has been given to spirit, but emerges as the outcome of its own internalization
and reflection—both activities accomplished through recollection. The spirit of absolute know-
ing, which constitutes the standpoint of science, is thus both a new shape and the result of the
process which led consciousness to the comprehension of itself and its own experience. The last
lines of the Phenomenology make this point clear:
The goal, absolute knowing, or spirit that knows itself as spirit, has for its path the
recollection of the spirits as they are in themselves and as they accomplish the organi-
zation of their realm. Their preservation, regarded from the side of their free existence
appearing in the form of contingency, is history; but regarded from the side of their
[philosophically] comprehended organization, is the science of knowing in the sphere
of appearance [phenomenology]: the two together, comprehended history, form alike
the inwardizing and the Calvary of absolute spirit, the actuality, truth, and certainty of
his throne, without which he would be lifeless and alone.
(PhG, § 808; original emphasis)
In absolute knowing, by interiorizing spirit’s past experience, recollection produces the “new
shape of spirit,” that is, the shape of spirit that, building upon the heritage constituted by the his-
tory of the spirits who preceded it, and moving from what is both a starting point and a “higher
level,” can face its new task. This task has as its goal the attainment and development of what
Hegel defines in terms of a “comprehended history,” the outcome of the unification of history,
that is, the existence of spirit that takes place in the form of contingency, and of its conceptual
organization, that is, the conceptual understanding deriving precisely from the recollection and
rationalization of the past experience.
491
Valentina Ricci
3. Recollection as the means of spirit’s emancipation

Hegel’s discussion of recollection in the Psychology and the Phenomenology, as I just sketched
it, can give us some insights into the determinations of this peculiar activity of spirit, which is
significantly placed at two essential transition points within Hegel’s system: in the former case,
recollection constitutes the medium enabling intelligence to move from its initial passivity to
its spontaneity and creativity, which it will fully display in the activity of thought. In the latter
case, analogously, recollection is the activity of interiorization and preservation whereby spirit
relates to its own experience, transforming it—or better, transfiguring it (in the literal sense of
giving it a new shape)—into a conceptual resource that will remain available even after spirit’s
experience, in its immediate presence, has been swallowed by the flowing of time.
Precisely the temporal dimension of recollection constitutes its core feature in the contexts I
have examined: intelligence’s recollection rescinds all ties to external, immediate space and time
in which intuition is first given, and places it in the knowing subject’s own space and time, that
is, in the internal organization of its own sensuous experience. Similarly, the phenomenologi-
cal recollection erases the temporal determination in which spirit’s experience, in the form of
shapes, has been given to it initially, and transfigures time by turning the external connections
which it first established among things into internal, necessary, conceptual, and thus properly
“spiritual” connections. In both cases, what is at stake is the elimination of the temporal dimen-
sion of experience. This aspect is particularly controversial in absolute knowing, where Hegel
mentions both the need to cancel time, and time as the necessary dimension for spirit’s mani-
festation. It is not difficult to observe, however, that the time that is erased, in both cases, is
natural time, that is, the time which organizes events exclusively on the basis of a succession of
“before” and “after.” In this sense, recollection is, with respect to the power of time that erases
everything, the power of spirit over time. What is at stake, thus, in both cases, is spirit’s freedom.8
Recollection, then, seems to be the central function that enables spirit to reach the stage of
absolute knowing. It is the key that enables spirit to abandon the present in which it is immersed
and to open up the dimension of universality, that is, the dimension of the concept. As Hegel
makes clear in the last pages of the Phenomenology, the new shape of spirit that is absolute
knowing gives rise to a collection of “specific concepts” and to their “organic self-grounded
movement” (§ 805). This becomes particularly relevant if we consider that the main task of the
Phenomenology is to lay the foundation for, and to justify the standpoint of, the developed system
presented in the later works, including the Science of Logic and the Encyclopedia.
4. Recollection as method: the logic of Hegel’s philosophy

This context does not allow for an extensive discussion of the role of recollection within Hegel’s
Logic. However, I hope to show the important connection between the conclusion of the
Phenomenology and the beginning of the Logic itself, which plays a key role in determining the
nature of the logical method and of the subsequent development of Hegel’s philosophical system.
For our purposes, the text with which Hegel opens the “Doctrine of Being” in the Science of
Logic, “With What Must the Beginning of Science Be Made?” is the ideal starting point.
The main question Hegel is concerned with is whether the beginning of science should be a
mediated or an immediate one. As one would expect, however, he claims that neither of these
options, understood unilaterally, would be adequate: on the contrary, the overcoming of the
very opposition of immediacy and mediation is precisely what allows us to reach the standpoint
of science, which, as we have seen, is the outcome of the Phenomenology. And it is this connec-
tion, specifically, that Hegel refers to in order to argue that the beginning of the logic is both
492
G. W. F. Hegel
mediated and immediate. Pure knowing, as the “ultimate and absolute truth of consciousness”
(WdL, 46), is what makes the beginning of science a mediated one, that is, a justified one. Pure
knowing, that is, absolute knowing, is the outcome of the unification of subject and object, of
certainty and truth. All of the inadequate forms of knowledge that consciousness has experi-
enced are sublated and at the same time incorporated in pure knowing. Precisely, this process
constitutes the mediation that grounds the Science of Logic, which is thereby placed in the realm
of truth. At the same time, however, the beginning of science must be without presupposition,
and therefore immediate. This seems to pose a problem: how can the beginning of science
avoid being arbitrary and at the same time not presuppose any given content, which would
compromise its very nature as a beginning by pushing back its foundation ad infinitum? The
answer to this question lies in the work of recollection, which—as we have seen—internalizes
the determinateness of the content and gives rise to the concept in its purity, that is, the content
that has been purified from its concrete spatio-temporal determinations. Recollection is the func-
tion that, at the end of the Phenomenology, performs the transition from the concrete richness of
spirit’s experiential content to the purity of the concepts that will form the structure of the Logic.
The abstractness and the immediacy with which science begins, thus, seem to only be possible
insofar as they emerge from a path, that is, the phenomenological path, which constitutes their
mediation and justification.
Recollection, thus, far from being a merely psychological mechanism, seems to perform
an essential methodological and logical function. Scholars like Angelica Nuzzo have pro-
posed, in fact, to understand Erinnerung as responding to a concern that the Psychology, the
Phenomenology and the Logic, although at different systematic loci, have in common, namely
the problem of introducing the notion of objective thinking and of reconciling it with the
“subjective activity,” which for Hegel must be understood as “an essential moment of objec-
tive truth” (WdL, 46).9 Moreover, Nuzzo proposes to understand the logical account of
recollection/memory, that is, “dialectical memory,” as foundational to the psychological
account of recollection in that it “immanently structures the development of theoretical spirit
in the Psychology” (Nuzzo 2012: 61).
Rather than presenting a detailed analysis of the passages in Hegel’s text that further support
this interpretation,10 I believe it is worth focusing on a more general consideration. From the
examination of the three texts I have presented in this chapter, it seems that we can regard rec-
ollection as a key methodological component in the development of the logical determinations
and in their organization in the Logic and throughout Hegel’s system. But if one of the central
conceptual structures of Hegel’s logic, and therefore of his philosophy, is one of recollection11
and as such essentially related to the understanding of a past, how can we understand its claim
to absoluteness and its supposed independence from time and its contingency? Our discussion
of recollection suggests a richer view of the meaning of this claim and one that would seem to
be more appropriate to Hegel’s general project. The absoluteness of the scientific standpoint,
or the objectivity of thought, is contingent on spirit’s capacity to develop the conceptual tools
to understand and make sense of what is starting from its concrete experience—or from the
concrete content of thought, depending on the context we consider. Such absoluteness, thus,
incorporates and is essentially related to time: indeed, philosophy is, for Hegel, “its own time
comprehended in thoughts” (PhR, 21).
This reading of recollection has crucial implications for the understanding of Hegel’s phi-
losophy not as a closed, static, and conservative system, but as a flexible, open, and plastic one,12
which provides both the means to make sense of its past and the conceptual tools to understand
its present. Such present, in turn, will become spirit’s new past, and the material for the develop-
ment of new conceptual structures.
493
Valentina Ricci
Notes
1 This trend, however, has started to change and several new, more comprehensive studies on the topic
appeared in the last decade. For a complete and updated survey of the literature on the topic, see
“Introduction” in Ricci and Sanguinetti (2013).
2 I focus on these contexts because I regard them as foundational with respect to an interpretation of
the function of recollection in Hegel’s philosophy as a whole. There are further places in the system,
however, where recollection plays a key role, like anthropology, the philosophy of history, aesthetics,
the philosophy of religion, and the history of philosophy. For a detailed examination of these specific
instances of recollection, see Ricci and Sanguinetti (2013).
3 References to Hegel’s works are abbreviated as follows: EN, Philosophy of Mind; PhG, Phenomenology of
Spirit; PhR, Philosophy of Right; WdL, Science of Logic.
4 It may be argued that it is not necessary to distinguish the two terms, especially when dealing with a
reading of the general function of recollection within Hegel’s system. Nuzzo’s concept of “dialectical
memory” is an example of such flexible use of these terms. See Nuzzo (2006, 2012).
5 While Wallace and Miller translate “Geist” with “mind,” I believe “spirit” to be a more accurate translation.
I have thus modified the translations in this text accordingly.
6 On the unconscious character of the “nocturnal pit” where the images are preserved, see Masullo (1979)
and Mills (1996).
7 For this reading of the Phenomenology of Spirit, see Bencivenga (2000).
8 Hegel makes this point very clearly in the lectures on the philosophy of spirit, which he held at the
University of Jena at the same time as he was working on the Phenomenology: “[Spirit is] time, which is
for itself, and the freedom of time as well—this pure subject that is free of its content but also master of
it, unlike space and time which are selfless.” See Rauch (1983: 86).
9 See Nuzzo (2006, 2012).
10 For this purpose, the reader can refer to Nuzzo (2006, 2012) and to Ricci and Sanguinetti (2013).
11 See Bencivenga (2000: 56).
12 See Malabou (2004).
Further reading
T. Rossi Leidi, Hegels Begriff der Erinnerung. Subjektivität, Logik, Geschichte (Frankfurt am Main: Peter Lang,
2009) analyzes the role of recollection in the philosophy of subjective spirit, the logic, the Phenomenology,
and the philosophy of history. A. Nuzzo, Memory, History, Justice in Hegel (Basingstoke and New York:
Palgrave Macmillan, 2012) is a stimulating discussion of the role of recollection and memory in these areas
of Hegel’s thought. By focusing on the connection between history and memory, Nuzzo proposes a read-
ing that appeals to contemporary debates as well as to Hegel scholars. V. Ricci and F. Sanguinetti (eds.),
Hegel on Recollection. Essays on the Concept of Erinnerung in Hegel’s System (Newcastle: Cambridge Scholars
Publishers, 2013) provides a comprehensive examination of the role of recollection throughout Hegel’s
system, including areas not covered by other studies, such as aesthetics, the philosophy of religion, and the
history of philosophy.
References
Bencivenga, E. (2000) Hegel’s Dialectical Logic, Oxford: Oxford University Press.
Hegel, G. W. F. (1977) Phenomenology of Spirit, trans. A. V. Miller, Oxford: Oxford University Press.
Hegel, G. W. F. (1991) Elements of the Philosophy of Right, ed. A.W. Wood, trans. H. B. Nisbet, Cambridge
and New York: Cambridge University Press.
Hegel, G. W. F. (2007) Philosophy of Mind, trans. W. Wallace and A. V. Miller, with Revisions and
Commentary by M. J. Inwood, Oxford: Clarendon Press.
Hegel, G. W. F. (2010) Science of Logic, trans. G. di Giovanni, Cambridge and New York: Cambridge
University Press.
Malabou, C. (2004) The Future of Hegel: Plasticity, Temporality, and Dialectic, New York: Routledge.
Masullo, A. (1979) “Das Unbewusste in Hegels Philosophie des subjektiven Geistes,” Hegel-Studien,
Beiheft 19: 27–63.
494
G. W. F. Hegel
Mills, J. (1996) “Hegel on the Unconscious Abyss: Implications for Psychoanalysis,” The Owl of Minerva
28(1): 59–75.
Nuzzo, A. (2006) “Dialectical Memory, Thinking and Recollecting. Logic and Psychology in Hegel,” in
Brancacci, A. and Gigliotti, G. (eds.), Mémoire et souvenir. Six etudes sur Platon, Aristote, Hegel et Husserl,
Naples: Bibliopolis, pp. 89–120.
Nuzzo, A. (2012) Memory, History, Justice in Hegel, Basingstoke and New York: Palgrave Macmillan.
Rauch, L. (1983) Hegel and the Human Spirit: A Translation of the Jena Lectures on the Human Spirit 1805–6,
Detroit, MI: Wayne State University Press.
Ricci, V. and Sanguinetti, F. (eds.) (2013) Hegel on Recollection. Essays on the Concept of Erinnerung in Hegel’s
System, Newcastle: Cambridge Scholars Publishers.
495
41
SIGMUND FREUD AND
JACQUES LACAN
Martin Schwab
1. Introduction
In a volume on theories of memory, psychoanalysis is of interest for its idea of unconscious memory.
Unconscious memory is different from many other familiar forms in five important respects:
(1) A sharp line separates structural or systemic memory from historical memory.
(2) What is remembered is inaccessible to ordinary remembering consciousness, accessible only
after an unfamiliar reconstruction.
(3) Remembrances are unrecognizable as to that they recollect, and what they recollect.
(4) Unconscious memory is an agential power of its own, recalling itself to the subject after
equally unconscious formative processes (repression, displacement).1
(5) The psychic forces and agencies participating in remembering articulate the remembrances
in figural representation (metaphor, metonymy, staging, iconicity).
Psychoanalysis provides the theory and the practice of transforming unconscious memory into
conscious memory. The ability to decipher and bring to consciousness what was out of reach for
remembrance, and the curative power of this transformation are integral parts in the ascription
of unconscious memory.
2. Sigmund Freud
Unconscious memory is part of Freud’s dynamic model of the soul, a topology of psychic
regions and a system of psychic agencies, separated by boundaries and allowing for transac-
tions between the regions, both cooperative and antagonistic in the interactions between the
agencies. The Id, Ego, and Superego, are functional and dynamic units, formed in child-
hood, further transformed in the course of a life. Each of the agencies consists in a bundle of
functions and processes held together and distinguished from the other agencies by guiding
principles. The Ego operates under the reality principle, the Id under the pleasure principle.
The Superego has the task of holding the ego to demands on its self-realization (ego ideal) and
represents internalized societal norms (e.g., conscience).
496
Sigmund Freud and Jacques Lacan
Freud’s “Project”: sketch of a “scientific” theory of memory

Freud sets the stage for unconscious memory even before he develops the core of his doctrine
in his draft “Project for a Scientific Psychology” from 1895.2 The “Project” introduces key
concepts and sets the stage for Freud’s later models of memory.3
The “Project” conceives of the psychic apparatus in neurological terms. Psychic activity is
neuronal activity. The neuronal basis of the soul consists of three kinds of specialized neurons
and neuronal activity in the brain: those in charge of perception of inner and outer stimuli,
those in charge of memory, and those in charge of our everyday consciousness. The psyche
is organized as interconnected networks of neurons transmitting psychic energy within and
between the networks. The “Project” makes two basic assumptions. First: trajectories chan-
neling the energy flows are not preprogramed but the result of tracings [Bahnungen] left by
discharges of energy traversing the neuronal systems. Second: neurons do not act upon each
other unless a threshold of excitation has been reached. Energy can enter the apparatus from the
outside (the outer senses) or from the inside (e.g., corporeal needs). The system of neurons as
a whole has a general tendency to stabilize itself at a certain overall, internal energy level, seek-
ing homoeostatic equilibrium, for Freud a “principle of constancy.” The system will therefore
discharge excess, and perhaps also seek recharge when this level is suboptimal.
Perceptional neurons respond to stimulation but return to their previous receptivity after
excitation and discharge. They are not permanently transformed by the impact of sensuous
stimuli.4 The neurons in charge of memory, however, “have the ability to be permanently
altered by the course of an excitatory action.”5 That transformation takes the form of a path
or a pathway [Bahnung], traced by an excitation, channel for later excitations of a sufficiently
similar kind. The differential totality of those paths is “memory” [Gedächtnis] as a function-
ally and structurally differentiated system. The most basic form of memory is thus a pathway
created and reinforced by the trajectories of energy flows. The mnemic system “remembers”
previous discharges that have traced the pathway through its tendency to repeat. Memory is
not “of events” but the trace left by events. It is structural or systemic memory and recall.
The first remembering consists in the path, trace, or facilitation [Bahnung], a trajectory, which
creates, transforms, or enhances an itinerary for future energy flows. Events of second remem-
bering are repetitions, energy running along a pathway earlier discharges have also followed. The
mnemic system thus combines a retentive and a transformational aspect in its neuronal model.
I insist on this first type of memory because it will be the blueprint for later conceptions of
memory, both in Freud and Lacan.
The main features of the model in the “Project” are:
(a) The order of the pathways is the memory system. Discharge along those paths is systemic
remembering. So far, all differences are differences in pathways and difference of points for
input and output (the sense organs and the motor system).
(b) So far neither that order nor its processes involve consciousness either in the tracing or in
the processing of recollection. The Memory neurons in charge of recalling external and
internal perceptions act in purely quantitative terms, and according to the topological
differences of the pathways. Consciousness and qualitative differences require a setting
different from the systems of perception and structural memory.
(c) Consciousness is an inner perception of a special feature of energy flux. It is tied to a feature
of discharge that varies in a continuous way (amplitude of energy flow).
497
Martin Schwab
(d) No depositing occurs; the tracing [Bahnung] forms a matter that is there. “Memory” is
a “trace” and a system of thresholds and directions for discharging and retaining energy.
Memory is neuronal system invested [besetzt] with energy, offering “facilitations” [Bahnungen]
and hindrance for the discharge of energy. Resistances of the neuronal “matter” are removed
or created along connected series of traversed knots (neurons). Freud will talk of a record,
something like a written record of an event.6
So far, episodes are recorded by tracings they leave. There are no record of discharges that sim-
ply follow a path without transforming it. Remembrance, here understood as episodic memory,
a nexus between earlier and later experience such that the later experience is “about” the earlier
experience in its meaning, does not find an adequate place in the “Project.” This holds for both
conscious and unconscious remembering. “To reproduce or remember is,” says Freud, “devoid
of quality.”7 The specific reflexive character of everyday remembrance, the fact that the recall-
ing refers back to what it recalls in its meaning is unable to be accounted for in the neurological
terms of the “Project.”
The “Project” nevertheless discusses reflexive remembering. It addresses it as a special
case of consciousness by introducing the concept of “memory-image” [Erinnerungsbild].8
Interestingly, the “Project” here lays the theoretical foundation for an understanding of
reflexive remembering that is peculiar. An image that is able to function as reflexive remem-
brance owes this capacity to Freud’s antagonistic understanding of the psychic dynamism.
Let a memory-image exist in whatever form, remainder of an experience. Let that image
include markers for its episodic origin. Now assume a remembrance of the kind “I remember
that I experienced . . .” Then Freud’s structural account in the “Project” implies, without
spelling it out for memory, that conscious remembering of the episodic variety is possible
only through the fact that the soul inhibits an alternative to it. That alternative is recurrence of
the original experience—not as reflexive remembrance, but as “the same experience again.”
Hallucinatory revival, perhaps also in the form of déjà vu, i.e., of the thought “I have had
this experience before” needs to be excluded for reflexive remembrance to be possible. The
phenomenal difference between recurrence and reflexive remembrance is the result of an
inhibitory and transformative action performed by the ego, against the basic tendency of rep-
etition or recurrence in the Id. The “Project” starts Freud’s theoretical attitude to interpret
even everyday mental occurrences as resulting from the antagonism of the psychic agencies,
and to conceive of psychic activity as interaction of antagonistic forces and process differences
(primary and secondary processes).9
The “Project” thus begins to distinguish structural and episodic memory. Structural memory
is formed and transformed due to episodes, but it is in turn active only by conveying its rep-
etitious form to episodic memory. Reflexive memory (remembrance) is addressed, but not
accounted for. And it is conceived as the result of an interaction of antagonistic forces.
Memory, conscious and unconscious, in dreams and neuroses10

Dream consciousness—the dreaming mind—is sui generis when compared to diurnal wake
consciousness, but is also able to be reflexively remembered. Dream memory of this kind is sys-
temically distortive of the dream experience, unable to be recalled in a manner that mimics the
experience of dreaming. Psychoanalysis makes a major contribution to the exploration of that
distortion, understanding dreaming as a rational activity. Dream memory is also a paradigm case
for epistemic problems of wake memory. I will not pursue these issues here because my focus
here is on unconscious memory.
498
A dream as paradigm for unconscious remembrance

I turn to “dream-as-unconscious-memory.” Freud discusses it in the context of neuroses.11 I
will use a case history, in which Freud treats a traumatic experience of a neurotic patient as an
event, and the dream as its recollection. To dream is here an example of episodic memory, of
reflexive unconscious remembering. The case is usually called “The Wolf Man” and presents
parts of the treatment of a young Russian adult, Sergei Pankejeff. Freud published it under the
title “From the History of an Infantile Neurosis” in 1918,12 diagnosing obsessional and phobic
symptoms in Pankejeff’s childhood.
When Sergei Pankejeff, aka the “Wolf Man,” becomes Freud’s patient as a young adult he
is “entirely incapacitated and completely dependent on other people.”13 The troubles of the
adult continue a childhood era that “was dominated by severe neurotic disturbance . . . in
the shape of anxiety hysteria (animal phobia), then changed into an obsessional neurosis with a
religious content.”14 His childhood phobia manifested itself as predominantly a fear of wolves,
but more generally of animals, accompanied by sadistic behavior towards animals. The subse-
quent obsession is characterized by blasphemous thoughts (“God-shit,” “God-swine”), which
require attunement, for example through signing the cross while breathing deeply. When the
boy meets handicapped people or beggars in the street, he needs to inhale.
Two events (including the boy’s immediate responses to them) recall themselves in the
symptoms just sketched. The first event is the “Primal Scene,”15 which occurred at age 1½. The
boy sleeps in his cot in his parents’ bedroom. He wakes up and observes his parents perform-
ing intercourse a tergo, the father upright behind his wife, the mother bent down. The child
sees their genitals in the act of making love; he also hears the noises they make, and sees the
expressions on their faces. His responses, as reconstructed in the analysis: his father appears as an
aggressor, the mother as a victim, but visibly and perhaps also audibly enjoying what she suffers.
The second formative event is the “Seduction Scene.”16 It takes place when the patient is age
3¼. This time the active role is played by Pankejeff’s sister, his superior in age, initiative, and intel-
ligence. The siblings have an antagonistic relationship. In the seduction scene, the sister proposes
to her brother “bottom showing” and performs it on him. Later, she plays with his penis, and tells
him that his beloved nanny does the same thing with men. Freud interprets the consequences of
the seduction scene as a disposition to passivity, an erotic predilection for buttocks, and anal-sadistic
character traits in Pankejeff. Both scenes have been recorded in the unconscious, together with the
affective responses, but have been removed and withheld from consciousness by repression.
The dream allows Freud to reconstruct the scenes that are at the root of the neurotic symp-
toms.17 But it is also a paradigmatic example for the role of memory in the context of neuroses.
The dream and the neurotic symptoms both are remindings of the “Primal Scene” and its
impact. The dream Pankejeff remembers during the cure18 occurred just before his fourth
birthday, two-and-a-half years after the “Primal Scene,” and three-quarters of a year after the
“Seduction Scenes”: young Pankejeff is lying in his bed. Suddenly the window opens, and the
boy sees six or seven white wolves sitting on the branches of the big leafless [wintertime] wal-
nut tree in front of the window. The wolves have bushy tails, untypical for wolves, and have
their ears pricked up. The child screams and wakes up, terrified by the threat of being eaten up
by the wolves. Freud’s thesis: the dream is staging the episode of the “Primal Scene” in dream
representation. The dream follows the principles of dream formation Freud has explored in his
Interpretation of Dreams. It is activity of the Id, but also formed by the forces of repression and the
censorship of the Ego, which call for strategies of displacement and condensation of the dream
thoughts, and articulation of them in figural language.19
This is how Freud sees the “Primal Scene” taken up in the dream:20
499
Martin Schwab
Table 41.1 The Wolf Man’s dream as “Primal Scene”
Dream, as described by the Wolf Man “Primal Scene,” as reconstructed
It was night, I was lying in my bed I had been asleep [in my cot]
Suddenly the window opens of its Suddenly I woke up.
own accord
The big walnut tree Tree as lookout spot. Cot or father as lookout spots,
father looking at boy?
[Also: reference to the place of the “Primal Scene”].
An intermediate link: A tale in which a tailor Accounts for the tree and the danger that is located
takes refuge on a tree in order to escape from a at the lookout place.
pack of wolves. In that story a tailless wolf seeks Castration anxiety from seeing father making love to
revenge for having lost its tail to the tailor. penisless mother.21
Tailless wolf proposes to other wolves to climb Climbing—on each other—father upright
on each other in order to reach the tailor. “jumping” mother. Father as aggressor.
Dream wolves have climbed up on the tree.
The wolves: six or seven. Going from a single “wolf”/ father to several—
Fairy tale trace: “The Wolf and the Seven Goats.”22 interpreted as distortion due to resistance.
Sitting on the tree, ears pricked. Father looking at him, anger of being disturbed in
Looking at him with strained attention. lovemaking.
Perhaps also: the boy looking intently at the parental
scene.
They were quite white. Bedclothes and, possibly, parents’ underwear.
[Winter fur of arctic wolf?]
They sat there motionless Opposite to lovemaking motions of parents.
[Reversal into opposite?] Also: father stops his action
and looks at him [perhaps annoyed].
Foxtails instead of wolftails. Castration fears in seeing violent action from father
Combines with tailless wolf in other story, in towards penisless mother.
which a tailless wolf seeks revenge for his loss Later passivity as identification with mother.
to his enemy who has climbed up a tree.
The “Primal Scene” is a historical event, a lived experience recorded in unconscious memory,
withheld from conscious memory by repression and lack of conscious understanding. The
dream is a return, a “recall” emanating from the experience of the “Primal Scene” stored in
the Unconscious. Now, Freud seems to explicitly deny that the Wolf Man’s dream is a “recol-
lection” [Erinnerung].23 But he says that merely because he wants to deny that the dream is a
remembrance in the reflexive understanding of conscious remembering: The dreamer neither
dreams “I am recalling the ‘Primal Scene’,” nor “this dream is about an earlier experience.” The
recalling dream is not a reminiscence. It is, however, a recall of something recalling itself to the
dreamer in dreamlike terms and manner, and, at the same time, reminds the dreamer of the
repressed experience. What recalls itself is the memory of the “Primal Scene” and its repressed
status, both in factual and affective terms. The experience of the “Primal Scene” is a dynamic
item in unconscious memory that brings itself into a renewed presence and effectiveness in the
dream: “The effects of the scene were deferred, but meanwhile it had lost none of its freshness
in the interval between the ages of one and a half and four years.”24
What returns in this return is not recognizable to the consciousness of Pankejeff in his
way of remembering the dream. It is only in the course of the analysis that Pankejeff’s
consciousness is led to recognize the dream as recalling the “Primal Scene.” But the dream
500
itself is already reaching back to the “Primal Scene,” both through its meaning, and its func-
tion internal to Pankejeff’s neurosis. That forward-feeding force and its backward-feeding
correlate, and the appearance of that relation in “Gestalt” unrecognized for what that force
and that meaning are, is Freudian unconscious memory, here in the form of unconscious
reflexive remembering.
Where, in the dream experience, is the successor to the indexical elements that are constitutive
of conscious remembering? In ordinary remembrance, these are integral parts of remembering,
from vague déjà vu to memory of detail and circumstances of an original experience. There are
two events—the “Primal Scene” and the wolf dream—separate in the way the recollected item
and the recollecting item need to be for remembering. In the case of the Wolf Man, the object
that figures as the remembered item (the traumatic experience) (a) is the bearer of the forward-
feeding force that brings about the return, (b) that to which the remembering event relates back
in meaning and motivation, and (c) the recollecting item is a demand for dealing with the recol-
lected item. The trauma reminds the subject of something that it needs to be aware of, namely
with the trauma that has not healed. The dream and young Pankejeff’s neurotic symptoms point
back to the “Primal Scene” in this triply meaningful way. The “dream-work” [Traumarbeit] is
the activity of remembering. It reaches back to the “Primal Scene” and subjects it to its own
thinking and wish-fulfillment. The dream is thus not just a recurrence of the same, the self-same
resurfacing at another place in time, repetition. The forward-feeding pressure of what resides in
the memory and the backward-feeding grasp of the dream-work—primary process and second-
ary reworking—participate in the creation of the dream and join forces in the dreaming activity.
In sum, the constellation is sufficiently close to conscious reflexive remembering to form a case
of “unconscious remembering,” and sufficiently different from the original experience to not
just be “recurrence”: unconscious episodic memory.
Memory in the psychoanalytic cure

Here, I will need to be brief. For the role and function of memory in the psychoanalytic cure,
we need to leap from the boy to the adult patient. The purpose of the cure is, for Freud, the
removal of the neurotic deformation from the psychic organization, which is, negatively, the
disappearance of the neurotic symptoms, and positively, the improvement of the patient’s abil-
ity to live his life in a self-chosen way. Freud laconically introduces the state of his patient as
“entirely incapacitated and completely dependent on other people.”25
Psychoanalytic treatment aims at finding the unconscious roots for the neurotic symptoms,
bringing them to consciousness, and, by working through them in therapy, break their power
over the subject.26 One of the means to achieve that result is to transform unconscious memory
into conscious memory. The reconstruction of the “Primal Scene,” and the interpretation of
the dream as being unconscious memory are part of that “working through.” They need to be
brought to the patient’s consciousness. Memory in the discursive sense plays an important role
in this process, also as evidence for the existence of the remembered event.
3. Jacques Lacan
Lacan claims to continue and to bring into its own Freud’s psychoanalysis, allegedly mis-
represented by some of Freud’s followers, while proposing his own theory under the guise of
a “return to Freud.” Three fundamental changes mark Lacan’s originality. First, Lacan draws
on the existential philosophies of Heidegger and Sartre and on Kojève’s interpretation of
Hegel for his idea of who we are, who and what we ought to become, and how we aspire to
501
Martin Schwab
realize ourselves: Lacking being, we need to take care of our being by becoming. We aspire
to attain full being, but our condition lets us pursue that goal and at the same time fail to
attain it. Recognition by others is our most basic demand. Second, Lacan uses the theories
of the linguistic structuralists to conceive of the psychoanalytic subject in linguistic terms and
its processing modes as a form of speech, and of the unconscious as a form of language. The
subject constitutes itself in terms of symbolic orders, which the subject neither initiates nor
controls. Identifying self and others by way of images, the subject and its others will be system-
atically taken for what they are not. The consequence is misrepresentation of self and others.
This condition constitutes the order of the imaginary. Necessarily inserting themselves into
symbolic orders subjects are “spoken” by something other than themselves—the ‘Other’—an
anonymous place the subject cannot reach as his own place. Together, the two orders of the
Imaginary and the Symbolic prevent us from becoming unified, self-empowered, and self-
empowering subjects, leaving us ontologically split, acting and expressing ourselves in terms
that alienate our being. A Sisyphus model of man!
Lacan articulates the complex order of our decentered condition in the famous Schema L
(see Figure 41.1):27
“A”/(O) stands for an anonymous other, the non-selfness condition of the symbolic order.
“S” is the subject of psychoanalysis, the subject whose conduct is in terms of the symbolic order.
The vector from A→ S expresses the fact that the subject is spoken by an order that is not where
that subject is. The relation A-S thus represents the decentered condition of the subject being
“spoken” in a language from where the subject is not: the unconscious. The relation a′-a tells us
that the images we have of ourselves and others do not reflect the reality of the represented item,
but determine what we take that item to be, and bar the subject’s access to A/O.
The joint action of those two orders determines our destiny; determination by events is
secondary; events are more markers than causes of turning points in the trajectories of our
lives. The Lacanian subject is moving along a life-path of self-historicization,28 both consciously
and unconsciously, driven by existential demand and oriented by projects of self-realization.
Realization falls short because of the alienating force of the imaginary and the symbolic orders.
I first try to illustrate Lacan’s version of structurally unconscious memory at a place that
exhibits its form. (Below I will turn to the manner in which this form is executed.) Memory and
remembering are explicitly discussed in a thought experiment published as an addendum29 to
the “Seminar on the Purloined Letter.”30 Negatively, Lacan wants to illustrate “that the remem-
bering [mémoration] at stake in the unconscious . . . is not related to the register that is assumed to
be that of memory, insofar as memory is taken to be the property of a living being.”31 Lacan here
seems to reject that the psyche is durably formed and transformed by the impact of events that
(Es) S a ′ other
n
atio
rel
y
nar
a gi un
co
im ns
cio
us
(ego) a A Other
Figure 41.1 Lacan’s Schema L.
502
leave a trace that causally or teleologically determine subsequent behavior. The “Bahnungen”
of Freud’s “Project” and the remembering of trauma in the Wolf Man would be examples of
what Lacanian unconscious “mémoration” is not. As a consequence, the model and mechanism
of memorization need to be conceived in a way that steers clear of the venerable analogy with
writing as a model for memory.32 Lacan’s model is to show “how a formal language determines
the subject.”33
Lacan devises the following thought experiment: let reality be a random sequence of events
at the level of reality. This reality would not exhibit a causal or teleological order. Lacan’s basic
reality consists of a random distribution of a binary distinction, for example “head” and “tail”
upon tossing a coin, one or the other, but neither both nor none. Randomness is the decisive
feature of this reality assumption. Let the two events be symbolized by + and -. The sequence
could be: - + - - + - - - + + - + + ++ - - + - - . . .
Next Lacan introduces a language. Its terms are (1), (2), and (3). Meaning assignments will
relate those terms to groupings of the tossing events, concretely to groups in the sequence con-
sisting in three consecutive events. At certain places, there will be a group of three times ‘heads’
(+++) or three times absence of heads i.e., ‘tails’ (---). There will also be groups of three that
contain the same kind of event twice next to each other, and one of the other kinds (++-. -+,
+-, -++). And there will be groups that alternate the two outcomes in a symmetrical manner:
“heads,” “tails,” “heads,” and “tails,” “heads,” “tails” (+-+. -+-). The groups can be ordered
according to three formal features: homogeneity, dissymmetry, and symmetry.
The assignments: The number (1) designates homogeneous series of three (+++ or ---);
(2) stands for dissymmetrical groups (++-, --+, +--, -++), and (3) for symmetrical alterations
(+-+, -+-). Note that not only is the basic sequence of “real” events a random series. It is also
the case that the referents for the (1), (2), and 3) are not independent: each upcoming referential
group of three events shares with the referent of its predecessor term two events, retaining them
as part of its referent.
The meanings assigned to the terms introduce a determined nexus between the possibilities
for the terms to appear in a sequence. This is the structural analog to the ideas of trace and causal
or teleological nexus between the terms. The possibilities for the terms (1), (2), and (3) to follow
each other are limited—a limitation that will be decisive for Lacanian structural memory. Thus,
after a (1) standing for a homogeneous group no dissymmetrical alteration is possible (after +++
or ---, no +- . . . or -+ . . . is possible). So, the appearance of one group opens the possibility for
two kinds of groups as successors to appear, and renders impossible the appearance of another
kind. Succession of the terms (1), (2), or (3) is thus regulated. Also note that the actual refer-
ences of the terms are not univocally determined by their meanings. For each of the terms, their
referent can be realized in more ways than one. Reality matters!
With this we have reached Lacan’s claim “to have found in the ordered chains of a formal
language the entire appearance of remembering [mémoration] and quite especially of the kind
required by Freud’s discovery.”34 (In the light of other writings, I would like to attribute
to the thought experiment a somewhat weaker thesis, concerning only formal features of
unconscious memory.)
What kind of memory does Lacan find in the formal system? The series itself anticipates
and remembers: “La série se souviendra . . . .”35 This anticipation and that remembering are
purely internal to the order, even though the places for their actual occurrence depend on the
random events and their environment in the sequence. No order in the reality of the events
determines what can and what cannot occur at the level of the chain. It is only within sets of
possibilities predetermined by structure that reality has the power to select and to decide on the
signs that occur. Reality functions as trigger, not as cause or end. So much for the structure.
503
Martin Schwab
According to Lacan, the semantics of that system will constitute remembering in occurring
chains through co-occurrences and lack of co-occurrence, through possibilities enabled, and,
most importantly for the Lacanian unconscious, through absences—falsely attributed to the
order of reality when the formal laws are not known. Those regularities are necessary and
rigidly determined. As the remembering series is moving along, certain earlier constellations
exclude the appearance of other constellations at later points. I have pointed to two examples
from the formalism of Lacan’s paper.
Now the relation between an earlier event Et and a later event Ft+i, a relation that consists in
“Et makes possible/impossible the appearance of Ft+i” at place t+i, is not normally considered
to be memory in the sense of remembrance. We do not normally think of the occurrence of
Ft+I being memory-wise related to Et, just because the two co-occur in a necessary order. In
our everyday understanding, whatever may occur and then find its place in memory does not
depend on an, in principle, necessary, absence of other experiences to be able to be remem-
bered. We posit an agent as that which brings about the recalling activity. That agent is either
the subject wanting to recall or receiving the remembrance without calling it up. Or we ascribe
to forces, known or unknown, the event that drives the remembering. There may have been
a reminder somehow associated with the memory, or there may be a power at the side of the
memory driving the recall. In the ordinary understanding, forgetting is a contingent event, per-
haps also driven by a force. Freudian repression is an example for inaccessibility to consciousness
due to a contingent reason. Inaccessibility and forgetting are not characters that inhere in the
experience that undergoes forgetting. In Lacan’s thought experiment, however, inaccessibility
is necessary due to the formative rules of the signifiers (1), (2), and (3). In the structural under-
standing, the occurrence or elision of a specific succession is not an empirical fact caused by
other facts. It is certainly true that an occurrence or absence at a specific point in an occurring series
depend on their being triggered by events attributed to the real. But for Lacan, the nexus between
the earlier and the later signifiers is independent from reality and depends on the meaning of
the occurring term. The absence or presence at a specific place can of course be predicted and
retrodicted in a hypothetical judgment a priori for every series of signifiers that is ordered by
the rules that govern the sequence of the signifiers. But this does not make the dependent terms
say that they are so dependent, something the dream of the Wolf Man does in Freud’s reading.
How about Lacan’s “the series will remember . . . ”?36 The series is of course the place where the
exclusion/elision appears. The series is also the sequence of signifiers in which elisions manifest
themselves. But do series that display an order qualify as subjects of remembering? Remembering
in the remembrance sense is, in the most basic understanding, an activity of the mind, and
requires a “minding” agency. Of course, we also say that the tree remembers the drought of an
earlier time of its growth through the tree-rings. The rings are enduring material transformations
of the growing tree, and can be read indexically towards the climate at the time of a specific year
in the life of the tree. But the rings are not subject to forgetting, do not suffer from recalling
inaccuracy, do not operate in the tree in an active way, with consequences for the life of the tree
at a later time—all of which are features of, as it were “mind-memory.” At this place, Lacan’s
memory does not require a mind in the traditional understanding of mind and psyche.
What model of memory would Lacan’s formal model propose to us if we took it as an account
of unconscious memory? The formal structure exhibits what Lacan takes symbolic systems to
be. Their formalism determines (and limits) possibilities and impossibilities. “Memorization”
comes in tiers: the first recollection is of the symbolic order itself, acquired and internalized,
determining the patterns of occurring series. Series consist of conduct, a chain of signifiers, a
structurally formed sequence of meaningful behavior enacting a symbolic order. The adoption
and internalization of the order—let us think of gender identity—is memorized and becomes
504
personality structure. To “remember” the structure is simply to acquire the structure as a device
that confers form and guides. The fact that a sequence enacts a structure is not per se a remem-
brance of the structure. It just is the repeated action of the symbolic order in accordance with
its constitutive rules.
The second tier consists in the occurring series. If we want to think of the nexus of events as
a case of “remembering” that “remembrance” would be something that happens to the subject,
not something done by the subject. The occurring series itself is not a doer, for it is not the series
that acts to bring about the nexus that ends in an appearance or an elision in a specific environ-
ment. The series is, as I said the place where the event is necessary. Reality merely functions as
a trigger that activates the rule of the signifier. The Lacanian formal system is, if we take it as
memory, a “repetition automatism.”37
For another model of memory, we need to move beyond the rigid formal thought experi-
ment. The framework changes and turns into a basis for unconscious remembering when we
now turn to the existentialist strand of Lacan’s theory. First, we let the subject not just passively
figure as suffering a determination it receives from the “Other,” but as an agent or agency
that actively internalizes the Schema L, that is: an agent or agency that self-alienates in his/her
self-realization. The symbolic order and the elision that is part of its logic would then appear
as self-chosen, albeit unconsciously. This would be a subject desiring to be its own by pursuing
its possibility of being as its own (Heidegger). Trying to become his/her own possibility, that
subject would cut itself off from the path toward that goal through the very medium it uses in
the pursuit. In Lacanian terms, the subject would use, needs to use, the frame of the Symbolic
Order and the Imaginary for the understanding he or she has of self and of others, and would
thereby miss out on living his or her possibility as his/her own. Even where this authenticity
happens in a cure, the split condition remains in place.
The decisive part, goal, and impediment in a neurotic condition enters the subject as a piece
of its history that places a burden on it: “The unconscious is the chapter of my history that is
marked by a blank or occupied by a lie.”38 That hidden piece demands recognition and con-
frontation, which can come from the unconscious in the form of neuroses. But it can also occur
when the subject gets to that hidden place, and takes position in it and towards it, in the cure in
interaction with the analyst. What is at stake is, as it were, that the subject “owns up” to itself,
its history, and its desire as something unable to be fulfilled in a definitive way. I repeat: even
after a “successful” cure, the Lacanian subject remains a split subject, necessarily.
Unconscious memory will be a part of Lacan’s understanding of neuroses, symptoms, and
curative speech. He distinguishes two main types of neuroses: obsessional and hysterical.39
Neuroses are problem-cases of a complex, forward-feeding self-relation of the subject: “Neurosis
is a question that is being posed for the subject from where it [i.e., the subject] was before the
subject came into the world.”40 The existential question for the hysteric is supposed to concern
gender identity, and to raise issues around acceptance, rejection, and concrete form of the
subject’s gender role:41 “Who are you, as a man/woman?” The question for the obsessional neu-
rotic concerns his or her being: “Am I or am I not?”—against a background of a lack of being
(Sartre). Neurotic symptoms confront the subject with that question, of course in a way that
remains hidden to the subject and answers are for the most part a refusal to answer the question
in an authentic self-asserting way (“I am gendered . . . -”; “I am . . .”—and the negations “I am
gendered otherwise, I am deficiently gendered”; “I am a fake . . .”). The specific question will
have arisen at a certain place in the development of the subject, even of subjectivity as a condi-
tion, and will have remained unanswered or will have been answered in the manner of empty
speech. The question and its neurosis are thus historical events. This qualifies them as candidates
for a memory relation.
505
Martin Schwab
An application of this model can be found in a paper from 1953, entitled “The Neurotic’s
Individual Myth.”42 It is a secondary analysis of a Freudian case history called “The Rat Man.”43
The Rat Man suffers from an obsession, which is a response to a social constellation, both his
own and of his environment. A family history is its central piece. The patient’s father had given
up a poor woman, apparently the love of his life, to marry the Rat Man’s mother, who was
above him socially, thus elevating his social standing. The Rat Man’s father had also at one point
gambled away money that belonged to his regiment, and was saved from the shame of being
found out by a friend who jumped in and paid his debt—a dishonoring experience for the Rat
Man’s father.44
At the patient’s side: the Rat Man’s obsessional symptoms form an order that is a continua-
tion (with variations) of the earlier order, at the same time a response to it. These are some of
the elements that match the earlier constellation: on the one hand, a debt he, the Rat Man, had
incurred has been paid by somebody else. A superior enjoins the Rat Man to pay back what
he owes, but to a person who is not the Rat Man’s creditor. The patient develops an obsession
to execute this order to the letter, in spite of the fact that he knows the real creditor is a young
woman. Strange discrepancy between an apparently non-sensical obsession to settle a debt to
someone who is not the creditor, and his awareness that he owes a debt to a young woman!
Also in the background: pressure by the father to marry a rich woman. The patient’s obsessional
comportment matches two features of his father’s history: on the one hand, choice between a
rich and a poor woman, on the other hand repayment of a military debt. Lacan: “a schema [of
the patient] that, complementary in certain points and supplementary in others, parallel in one
way and inverted in another, is the equivalent of the original situation,”45 and
The scenario reflects, in a mode that is no doubt incomprehensible to the subject—

but not absolutely so, far from it—the inaugural relationship between the father, the
mother, and the friend . . . . it re-enacts a ceremony which reproduces almost exactly
that inaugural relationship.46
We thus find a dynamism between two orders, one order being that of the obsessional symp-
toms in the Rat Man (the obsessional order), the other being the order of the family history
(the familial order). The facts of the family history are organized into an order of blemish and
debt. The obsessional order has the form of a “repetition”—to be understood in the sense of
“bringing back”—a response to the familial order, but also of the Rat Man’s compensatory
action for felt deficiencies of the familial order.
These two orders are largely congruous. It would not be difficult to draw a table comparing
the two orders for relevant similarities and affinities as I did for the Wolf Man. The obsessional
order continues, emulates, stages, uses for its own purposes—the familial order. The obses-
sional order recalls the familial order, calling it up and reproducing it, also “making good”
what was “bad” in the familial order. Lacan’s “Neurotic’s Individual Myth” is thus close to
Freud’s recollection model: there exists an earlier constellation—the familial formation—and
a later constellation—the obsessional successor—and the later order is tied to the earlier order
as a recollection. The recollective bond is manifest in the similarities between the two constel-
lations and the nexus of reparation. These links have reminding function and transcend mere
structural repetition of the formal variety discussed above. The obsession ties together the two
items through meaning and the nexus of “problem” and “solution to that problem,” all this of
course unconsciously.
In the Rat Man case (Lacan), the family history holds the place of the Wolf Man’s traumatic
experience (Freud). The witnessing of the parental intercourse is an experience, and so are the
506
facts stored in the Rat Man’s memory of the family history. But those facts are not what the Rat
Man unconsciously recollects. What the obsession “re-minds” is the constellation, the organiza-
tion of those facts into an order, including the evaluative idea of an unresolved debt, a continuing
blemish and the injunction to take care of his father’s unfinished business. This constellation has
never been consciously experienced. Its unity, form and affective valence are in fact composed
by the obsession. In other words, the recollecting agent brings about what he recollects, at least
as regards organization and evaluation. The neurosis builds and uses a construct.
The Rat Man will have had a disposition to do just that, that is: to fall into an obsessional
neurosis. It is the existential question that needs to be in place in order to account for the deci-
sive organization. The reader should remember: the obsessional neurosis is an answer to the
existential question “Am I or am I not?” “Ontological uncertainty of being” is the motivation
for choosing and constructing the obsession. The patient tries to overcome that uncertainty by
appropriating the family history “as if it were his own,” as if he had failed in his life in the way
his neurosis feels his father did. His motivation? In order to “be,” the subject needs to
see himself in another, [a being] more advanced, more perfect than he . . . The subject
always has an anticipatory relationship to his own realization which in turn throws him
back onto the level of a profound insufficiency and betokens a rift in him, a primal
sundering, a thrownness, to use the Heideggerian term.47
From a memory point of view, the subject uses remembered facts to draw up an unconscious
memory he “makes up”! He is the unconscious author of both what he recollects and his recol-
lective obsessional activity, of both the a quo and the ad quem of the recollective relation.
To sum up the two legs of my discussion of Lacan: we have found two differently original
forms of memory in Lacan. The first was the “memorization” ascribed to structure, memory
as a repetition automatism, the remembering item is the structure, and the recurrences do not
possess recollective meaning. The second was the insistence of an unresolved existential ques-
tion that took the form of an unconscious, constructed history remembered in the obsessional
symptoms, a way of pursuing self-realization in the misguided manner of the obsession.
Notes
1 We will see below that this is not quite true for Lacanian unconscious memory.
2 Freud (1895).
3 For example, Freud (1925).
4 Later clearly articulated in Freud (1925).
5 Freud (1895: 300).
6 Compare with Freud (1925).
7 Freud (1895: 370); i.e., conscious remembering. Compare the identical observation at p. 375. The
sections “The Quality Problem” and “Consciousness” show that Freud is aware of the problems.
8 Freud (1895: 381). I see that move as a change in conceptual apparatus. Freud moves from purely neu-
ronal concepts, and the assumption of merely providing a neuronal substructure to mental phenomena
to the use of semiotic concepts (image) and mental concepts, as if he had the substructure for the images
and the mental events, too. But this is not the case. At this place, also his first use in the “Project,” the
memory-image would still be a structural residue, and not yet an image that carries the indicators of the
event of its tracing.
9 The “Project” (Freud 1895) implies this understanding in its presentation of the experience of satis-
faction (p. 379), of pain (p. 381), of distinguishing primary and secondary processes (p. 386) ascribed
to an ego organization (p. 384). Also Part III of the “Project,” pp. 417 ff. “Attempt at an Account of
Normal . . . -Processes.”
10 For reasons of space I omit discussion of the role of memory in Freud’s cultural writings, from the
Psychopathology of Everyday Life (1901) to Moses and Monotheism (1939).
507
Martin Schwab
11 Neuroses are “psychogenic affections in which the symptoms are the symbolic expression of a psychical
conflict whose origins lie in the subject’s childhood history.” Freud distinguishes them from psychoses
where the affection “lies in a primary disturbance of the libidinal relation to reality exemplified in the
delusions of paranoia or schizophrenia” (Laplanche and Pontalis 1973: 266).
12 Freud (1918: 7–122).
13 Freud (1918: 7).
14 Freud (1918: 8).
15 Freud (1918: Section IV entitled “The Dream and the Primal Scene,” pp. 29–47). For the term “Primal
Scene,” see p. 39 and the editor’s footnote.
16 Freud (1918: Section III, “The Seduction and Its Immediate Consequences,” pp. 175–85).
17 For a discussion of the epistemic problems of psychoanalytic reconstructions, compare Freud (1937:
256–69).
18 Freud (1918: 29, 30).
19 Compare Interpretation of Dreams, Chapter VII, “Psychology of the Dream Process,” Section E, “The
Primary and Secondary Process-Repression.”
20 Freud (1918: 42, 43).
21 Castration is of course a central part of the Oedipus complex, itself a decisive stage in the development
of a child’s gender identity.
22 For fairy-tale memories as “screen memories” (Deckerinnerung) compare The Psychopathology of Everyday
Life, Chapter IV.
23 Freud (1918: 44).
24 Freud (1918: 44, fn1).
25 Freud (1918: 7) (cited earlier).
26 Compare Freud (1914: 145–56).
27 In Lacan, “Seminar” (1956: 40), and “Seminar” (1954–55: 243).
28 Lacan (1953: 216 ff).
29 Lacan, “Introduction” (1956: 33–41).
30 In “Presentation of the Suite” (1956) in Ecrits, French edn., p. 41, English edn., p. 30. Lacan tells his
readers that he has written this text after delivering the seminar, but that it elaborates in writing
something that served as an introduction to the seminar in 1955. “Remembering” is not altogether
a happy translation of “mémoration,” a term that articulates the idea of placing or receiving something
into memory.
31 Lacan, Ecrits (2006: 31).
32 Plato, Theaetetus, 191c–196c.
33 Lacan, Ecrits (2006: 31).
34 Lacan, Ecrits (2006: 31).
35 Lacan, Ecrits (1966: 48; 2006: 36).
36 Lacan, Ecrits (1966: 48; 2006: 36, emphasis added).
37 Compare Lacan’s observations on “repetition automatism” in “Seminar” (1954–55: 86–90). The affinity
with Freud’s ‘Bahnungen’ will be obvious.
38 Lacan (1953: 215).
39 “Seminar” (1955: Chapters XII and XIII).
40 Lacan (1957: 43). “World” here needs to be understood in the Heideggerian sense that “world”
presupposes “Dasein.”
41 “Seminar” (1955: 168–72). Lacan’s take on hysteria is succinctly articulated in an interpretation of the
Dora case. For Dora, her neurotic problem is her own bodily femininity, which she is unable to rec-
ognize. Here, the hysteric problem is also against a foil of family history, albeit a history in which the
patient is a player. I think the case is less a case of unconscious memory than a case of unconscious staging.
Compare Lacan (1951: 180–81).
42 Lacan (1953). Also Lacan’s hesitant footnote in Lacan (1966: 72).
43 Freud (1909: 151–249). Also in Three Case Histories (1963: 1–81).
44 Lacan (1953: 410–11).
45 Lacan (1953: 413).
46 Lacan (1953: 414).
47 Lacan (1953: 423–4).
508
Bibliography
Laplanche J., and Pontalis, J.B. (1973) The Language of Psychoanalysis. London: Hogarth Press, 1973.
Freud
Sigmund Freud, The Standard Edition of the Complete Psychological Works of Sigmund Freud. Volumes 1–24.
(Cited as S.E. and number of volume.) Translator James Strachey. London, Hogarth Press.
—— (1895) Project for a Scientific Psychology, S.E., 1, pp. 281–397.
—— (1900) The Interpretation of Dreams, S.E., 4 and 5.
—— (1901) The Psychopathology of Everyday Life, S.E., 6.
—— (1909) “Notes on a Case of Obsessional Neurosis,” aka “The Rat Man,” S.E., 19, pp. 151–249. Also
in Freud, Three Case Histories. New York: Touchstone, 1963, pp. 1–82.
—— (1914) “Remembering, Repeating and Working Through,” S.E., 12, pp. 145–56.
—— (1915) “The Unconscious,” S.E., 14, pp. 159–216.
—— (1918) “From the History of an Infantile Neurosis,” aka “The Wolf Man,” S.E., 17, pp. 1–122. Also
in Freud, Three Case Histories. New York: Touchstone, 1963, pp. 161–280.
—— (1925) “Note on the Mystic Writing Pad,” in Freud, General Psychological Theory, Chapter XIII.
—— (1937) “Constructions in Psychoanalysis,” S.E., 23, pp. 256–69.
Lacan
Lacan, Jacques, Ecrits. Editions du Seuil, 1966. English translation: Lacan, Jacques, Ecrits, first complete
edition in English, trans. Bruce Fink. London and New York: Norton, 2006.
—— (1953) “Function and Field of Speech and Language in Psychoanalysis,” Ecrits, pp. 197–268.
—— (1945) “Logical Time and the Assertion of Anticipated Certainty,” Ecrits, pp. 161–75.
—— (1946) “Presentation of Psychic Causality,” Ecrits, pp. 123–58.
—— (1951) “Presentation of Transference,” Ecrits, pp. 176–85.
—— (1953) “The Neurotic’s Individual Myth,” Lecture at the Philosophical College; text ed. Jacques-
Alain Miller, approved by Lacan. The French original published in Ornicar 17 (1979); trans. Martha
Noel Evans, in Psychoanalytic Quarterly 48(8) (1979): 405–25, from which I cite.
—— (1956) “Introduction,” Addendum to the “Seminar on the Purloined Letter,” Ecrits, pp. 33–41.
—— (1956) “Presentation of the Suite,” Addendum to the “Seminar on the Purloined Letter,” Ecrits,
pp. 30–33.
—— (1956) “Seminar on the Purloined Letter,” Ecrits, pp. 6–48.
—— (1957) “The Instance of the Letter in the Unconscious, or Reason Since Freud,” Ecrits,
pp. 412–41.
—— (1958) “The Direction of the Treatment and the Principles of Its Power,” Ecrits, pp. 489–542.
—— (1966) “On the Subject Who Is Finally in Question,” Ecrits, pp. 189–96.
The Seminar of Jacques Lacan (1953–54), Book I Freud’s Papers on Technique.
—— (1954–55) Book II The Ego in Freud’s Theory and in the Technique of Psychoanalysis.
—— (1955–56) Book III The Psychoses.
—— (1964) Book XI The Four Fundamental Concepts of Psychoanalysis.
Other
Plato, Theaetetus.
509
42
HENRI BERGSON1
Trevor Perri
In 1896, at the tail end of what has been referred to as a “golden age” for the study of memory
(e.g., Schacter 2001: 88), Henri Bergson published his groundbreaking Matter and Memory: An
Essay on the Relation of the Body to the Mind. In this book, Bergson proposes an integrated theory
of multiple forms of memory that ultimately entails nothing less than a radical reconception of
consciousness, the material world, and the terms of their relation. During Bergson’s lifetime, this
book was praised by such notable figures as William James (1920: 179), who described it as a
work of “exquisite genius” that effects a Copernican revolution in philosophy akin to the criti-
cal turn made in Kant’s Critique of Pure Reason, and Walter Benjamin (1969: 157), who described
it as a “monumental” work that towers above other efforts that can be classified under the head-
ing of the philosophy of life. In more recent years, the originality and significance of Bergson’s
philosophy of memory in particular has been recognized in fields as diverse as cognitive neuro-
science,2 psychology,3 literary studies,4 and the interdisciplinary field of memory studies.5
However, when one turns to Matter and Memory itself with the aim of becoming acquainted
with Bergson’s philosophy of memory, one immediately encounters certain difficulties.6 For
one thing, Bergson employs the terms “memory” (mémoire) and “recollection” (souvenir)—
terms that he often uses more or less synonymously—in various, not obviously consistent,
ways. For example, unlike authors who stress the need to maintain a terminological distinc-
tion between memory as intention and memory as what is intended (e.g., Ricoeur 2004: 22),
Bergson sometimes uses the terms “memory” and “recollection” to refer to the act of repre-
senting some past event, and sometimes he uses these terms to refer to the past event that is
represented. Further, Bergson also sometimes uses the term “memory” in a more colloquial
sense to refer to a capacity for remembering the past, like when one says one has a good or a
bad memory. And finally, in what Arnaud François (2008: 30) has recently described as “one of
the most astounding and important theses in his work,” Bergson also sometimes uses the term
“memory” to refer to the past that is preserved. That is, rather than conceiving of memory
as a way of relating to the past from the perspective of the present, Bergson regularly equates
memory with the totality of one’s past as it is preserved in itself.
To further complicate matters, Bergson distinguishes multiple forms of memory throughout
his writings—often without indicating how these forms of memory relate to one another. For
example, in the first chapter of Matter and Memory, aiming to highlight the fundamental role of
memory in subjective life, Bergson identifies two forms of memory. On the one hand, Bergson
510
Henri Bergson
suggests that a contraction memory gathers together a plurality of independent moments to con-
stitute our enduring lived present.7 And, on the other hand, Bergson also suggests that another
form of memory, which we might refer to as perception memory, informs the sense of and provides
the content for every conscious perception (2001: 34).
Then, without referring back to this distinction between contraction memory and percep-
tion memory, in the second chapter of Matter and Memory Bergson describes another distinction
between two forms of memory. Deleuze remarks in Bergsonism that “this is a completely dif-
ferent principle of distinction” (1991: 125, n1), and he warns that we should avoid conflating
the distinction between contraction memory and perception memory with the new distinction
introduced in the second chapter between habit memory and what we might refer to as recollection
memory (1991: 127, n35). According to Bergson, whereas habit memory is an implicit, non-
representational “motor memory” (1991: 86) of the body that manifests itself as a disposition
to react in a more or less fixed way to one’s surroundings, recollection memory is the explicit
representation of some event or episode from one’s past life.
In addition to these four forms of memory, Bergson also describes another form or aspect
of memory in Matter and Memory that he terms pure memory. Pure memory is conceived as the
totality of one’s past experience preserved as an integral whole in an unconscious, virtual state.
And while this pure, virtual memory is precisely what is actualized in perception memory and
in the explicit recollection of memory-images, Bergson stresses that pure memory must not
be confused with these forms of memory since “memory that has been actualized in an image
differs profoundly from pure memory” (1991: 140).
Finally, in his subsequent works, Bergson discusses still other forms or aspects of memory
that are not obviously equivalent to any of the aforementioned forms of memory. Specifically,
in 1907 in Creative Evolution, developing an idea that is implicit in both Time and Free Will and
Matter and Memory, Bergson identifies a non-intellectual memory of the will that constitutes one’s
character and personal style. And in a 1908 article dealing with the phenomena of false recog-
nition and what is commonly referred to as the feeling of déjà vu, Bergson directly contradicts
Aristotle’s claim that all memory is “of the past” by arguing that there is a memory of the present
formed simultaneously with every present perception.
With the aim of providing an overview of Bergson’s philosophy of memory that can also
serve as a point of entry to his philosophy as a whole, this chapter explores the different senses
and forms of memory that Bergson describes, paying special attention to how they are distinct
from one another and how they are unified. It is my intention to show that, although these vari-
ous senses and forms of memory are different from one another (sometimes essentially so), they
are also continuous and unified insofar as they are equivalent to different tones of one mental
life and to different tensions of one duration.
1. From a metaphysics of memory to forgetting

As a first step towards explicating how these different senses and forms of memory relate to one
another, note that throughout his writings Bergson considers the mind from the divergent but
complementary perspectives of psychology and metaphysics. According to Bergson’s idiosyn-
cratic conception of these disciplines, psychology considers the mind just insofar as it facilitates
practical action, and metaphysics strives to achieve an intuition of this mind that is immediate,
which is to say as it is in itself and not as it is mediated by action and interest.8
Proceeding in these two ways at different points in his work, Bergson arrives at two dis-
tinct but ultimately complementary conceptions of memory. On the one hand, investigating
what he describes as “the practical and thus ordinary operation of memory” (1991: 78), Bergson
511
Trevor Perri
develops what we can refer to as a psychological account of the various ways that the past that is
preserved is manifested in the present to guide and inform one’s practical action. And on the
other hand, Bergson also consistently strives to consider memory as it is in itself and not just
insofar as it functions in the service of practical human action. Proceeding in this way, Bergson
develops what we can refer to as a metaphysical conception of memory that he refers to at one
point as memory “in the special sense that we give this word” (1991: 222).9 Unlike the ordi-
nary, common-sense conception of memory as the preservation and reproduction of some past
experience in everyday life, memory in this special sense is the ongoing, automatic synthesis of
the past and the ever-changing present into an undivided whole.
Although he does not explicitly refer to it there in this way, Bergson already develops such a
metaphysical conception of memory when describing the form and structure of a consciousness
that endures in his early work Time and Free Will. In this book, aiming to establish that there is
a difference in kind between the time or duration (durée) of consciousness and space, Bergson
argues that, as it is most immediately given in itself, and not as it is commonly conceived by the
practically directed human mind, consciousness does not comprise a collection of discrete states
or independent moments that can be counted as things spread out in space can be counted.
Rather, according to Bergson, duration and consciousness itself are given in immediate intui-
tion as a whole of continuous and interdependent but heterogeneous and successively given
qualities; they are given as a multiplicity, but a continuous multiplicity that is different in kind
from the more familiar multiplicity of number. While more could be said about this duration
and multiplicity, the crucial point for the purpose of this chapter is that the continuity of the
multiplicity of duration depends on an ongoing synthesis that is effected by what I have referred
to as contraction memory. Specifically, according to Bergson, duration and consciousness itself
are constituted as independent instants which are continuously and automatically contracted
together in memory and synthesized with the ever-changing present to form an undivided
whole. In the absence of this effort of contraction memory, Bergson suggests that there would
only be instantaneity and an “eternal present” (2001: 153).
To be clear, this duration constituted by contraction memory is not just a continually passing
present. Rather, according to Bergson, in principle, contraction memory synthesizes the totality
of one’s past with every new present. Hence, instead of appealing to the more traditional image
of a flowing river to illustrate duration, Bergson writes that the entire past continuously gathers
in the ever-growing present as snow accumulates on a snowball as it rolls down a hill (1998: 2).
What this image of the snowball is meant to suggest is that one’s past in principle does not pass
away; rather, according to Bergson, the totality of one’s past and one’s memory accumulates in
the present such that, in principle at least, consciousness includes “in an undivided present the
entire past history of the conscious person” (1946: 152).
Thus, according to Bergson’s metaphysical consideration of memory, a contraction memory
synthesizes independent moments to constitute a continuous multiplicity of successively given
parts, which suggests for Bergson that the entirety of the continually accumulating past is in
principle present at every moment.10 However, as Bergson is well aware, this metaphysical
account of memory as it is in itself cannot be the whole story. For one thing, the idea that
memory continually and automatically contracts together or synthesizes a plurality of moments
to constitute a continuous, extended present that includes the entire past seems to leave no room
for a crucial fact of experience—namely, discontinuity and forgetting.
However, as Jean Hyppolite explains in his masterful 1949 essay “Various Aspects of Memory
in Bergson,” despite the charge made by a later generation of philosophers that Bergson
overemphasizes the continuity and cohesion of consciousness and duration while neglecting
512
Henri Bergson
discontinuity and forgetting, Bergson confronts these issues head on in Matter and Memory.11
Specifically, Bergson argues that ordinarily we are conscious and live through the continuity
of our duration and the totality of our memory just to the extent that it is useful for action and
practical life. So, while consciousness might in principle encompass the totality of memory, and
while the entire past might in principle inhere within the present, in fact one’s consciousness
and the totality of one’s actual memory is ordinarily “narrowed down by action” (1991: 14),
and in fact the entirety of the preserved past only becomes present insofar as it can “make itself
useful” (1991: 140). This is to say that while it may be the case that in principle “our whole
personality, with the totality of our memories, is included, undivided in our present perception”
(1991: 165), this whole is never actually given as such.
Nevertheless, the duration and memory that are forgotten in this way are also never completely
lost. Rather, the totality of one’s past is preserved in and by itself in an unconscious state that
Bergson refers to as “pure memory” (1991: 156).12 Although he most often emphasizes the “pow-
erlessness” (1991: 141) of this unconscious, pure memory, Bergson also suggests that “the totality
of our memories pushes from the depth of the unconscious” to become conscious and has an at
least implicit effect on our present experience.13 As he writes in Matter and Memory: “Our whole
past mental life conditions our present state, without determining it in a necessary way; whole
also it reveals itself in our character, although none of its past states are manifested in character
explicitly” (1991: 148). Bergson comes to characterize this implicit effect of the whole of one’s
past on the present as a non-intellectual “memory of the will” (1998: 6), and he reiterates that
this memory of the will manifests itself in the “push” or “tendency” of one’s character (1998: 5).
In addition to manifesting itself implicitly in character, this ever-changing, unconscious,
virtual whole of the past is also what comes to be actualized in practically directed life in the
form of independent memories and recollections. In turning from considering memory as it is
in itself to considering how the unconscious, pure memory is actualized in the service of action,
we turn from what Bergson conceives of as a metaphysics to a psychology of memory.
2. A psychology of multiple manifestations of memory

Accounting for how the entire past that is preserved in memory manifests itself in one’s practi-
cally directed present is a central preoccupation of Matter and Memory. As Bergson indicates in
both the 1898 and 1910 introductions, the guiding idea of the book is that one’s consciousness
or mental life always comprises the integral totality of one’s memory or one’s “whole personality”
(1991: 14), but this totality is given in different ways. Sometimes, tending more toward present
action, one’s memory is given only insofar as it most generally resembles and can help one
navigate some present circumstances. And, at other times, in a way that tends more towards
contemplating or dreaming the past, one’s memory is manifested in its particularity and with all
the nuances that make one specific past experience differ from another. As Bergson describes it,
there are “different planes of consciousness” or “different tones of mental life,” and “our psy-
chological life can be played (se jouer) at different heights, sometimes closer to and sometimes
further from action, according to the degree of our attention to life” (1991: 14).
Bergson introduces this idea of different planes of consciousness or tones of mental life in
the second chapter of Matter and Memory by pointing to a radical difference between two ways
that past experience can be utilized for present action—that is, between two forms of memory
in the ordinary, psychological sense of the term. Anticipating contemporary discussions of
procedural memory and episodic memory, Bergson describes how, on the one hand, past expe-
rience can be actualized in the present in the form of “motor apparatuses” (dispositifs moteurs) or
513
Trevor Perri
“motor mechanisms” that have been set up in our bodies subsequent to some repeated action
or experience (1991: 77–8). And, on the other hand, he describes how one’s past can also be
actualized in the present in the form of independent recollections or representations of some
specific past event. In contradistinction to the first form of memory, which he claims is, in fact,
just “habit interpreted by memory” or “habit more than memory,” Bergson characterizes the
explicit recollection or representation of specific past events as “memory par excellence” and as
the “true memory” (1991: 84, 151).
Bergson explains the difference between these two ways that the action of the past can be
said to be manifested in the present—between what I have referred to as habit memory and
recollection memory—by considering the example of studying a lesson in order to learn it
“by heart” (1991: 79). Specifically, reflecting on two different experiences of memorizing a
series of words, Bergson distinguishes two essentially different ways that one can be said to
remember the words. On the one hand, I can say that I remember the words of a poem or a
song that I have memorized when, after repeating them a number of times, I am eventually
able to recite or repeat the words accurately without reading them. In this case, the lesson has
been committed to habit memory. And, on the other hand, I can also say that I remember,
but this time in a very different sense, when I think back to the experience of studying the
lesson and recollect a specific instance of reading the words that I wanted to memorize. That
is, after I have committed the lines of a poem or a song to memory in the sense that I am able
to recite this poem or song easily and accurately without reading it, I can also call to mind a
particular episode in which I read these lines. In this latter case, rather than remembering how
to recite the lines, I remember that I have read the lines at some specific, individual time in
my personal life history.
After considering how these memories are acquired and how they are manifested, Bergson
argues that these two forms of memory differ in nature rather than in a degree of more or less.
As he writes, “consciousness reveals to us a profound difference, a difference of nature, between
the two sorts of recollection” (1991: 80). And further, Bergson argues that, although they over-
lap and cooperate, the two forms of memory are “theoretically independent” from one another
(1991: 81). Specifically, Bergson argues that habit memory can “do without” (se passer de) the
recollection or representation of the history of one’s past life and that one’s representations of
past events are “independent of” the habit memory of the body (1991: 81).
Nevertheless, although Bergson argues that these two forms of memory are theoretically
independent, and while Bergson sometimes suggests that habit is not really memory, he also sug-
gests that, considered from another perspective, habit is just a highly contracted tension of our
duration and a form of memory that still deserves to be classified as such. That is, while Bergson
sometimes characterizes the sensorimotor habits of the body as independent from recollection
memory, he also characterizes these same habits as a contracted tone of mental life. And while
he sometimes claims that motor habits are simply present material mechanisms of the body that
are not really memory since they have no obvious relation to the past, Bergson also character-
izes habit as a unique and genuine form of memory that he designates a “memory of the body”
(1991: 152).
Moreover, even if the two forms of memory might exist independently in principle, we do
not encounter habit or recollection in isolation from one another in fact in ordinary life. These
two forms of memory are just two limits of our experience or just two extreme ways that one’s
past tends to be manifested rather than two fixed, neatly delineated forms of memory. The
implicit, non-representational memory of habit and the explicit, representational memory of
recollection are, to use Bergson’s own language, just “two extreme limits, at which the psy-
chologist must place himself alternately for convenience of study, and which are never really
514
Henri Bergson
A B
A′ B′
A′′ B′′
Figure 42.1 Bergson’s cone of memory
reached in practice” (1991: 168). And between the two extremes of habit memory and recol-
lection memory, there is a continuity and even a unity insofar as each manifestation of memory
is a different tone of one mental life. Thus, although they are perhaps distinct and independent
at the limit of our experience, pure habit and pure memory are also joined since, according to
Bergson, they are just two extreme tensions of one memory and duration.
This idea that habit memory and recollection memory are simultaneously distinct, continu-
ous, and unified is precisely what is meant to be depicted by the famous image of the inverted
cone presented in the third chapter of the book (1991: 162) (shown here in Figure 42.1).
Specifically, on the one hand, the base of the cone marked as AB represents the integral total-
ity of memory given in all its particularity and nuance. On the other hand, the apex of the
cone marked as point S represents the sum of the present and actual sensorimotor mechanisms
that have been acquired in my material body. And the plane P that point S intersects with rep-
resents the totality of the parts of the material universe that are in my field of perception. So,
disregarding for a moment everything but the apex of the cone S that intersects with plane P,
this image depicts a system of habits or motor mechanisms that inhere in the material world and
that are independent of all memory—simply stated, the point S is on plane P and not on plane
AB. However, although pure habit memory and pure recollection memory are thus repre-
sented as different in nature (insofar as one is represented by a point and the other by a plane),
and although they are represented as distinct and independent of one another (at least if we only
focus on point S on plane P and plane AB), the two forms of memory are also represented as
unified and as fundamentally continuous with one another insofar as they are both part of the
totality of the memory that is represented by the cone SAB (since the apex of the cone at point
S is an integral part of, so to speak, the cone SAB as one of its limits). And these extremes are
represented as continuous with one another in the sense that between the plane of the mate-
rial universe that includes the totality of my sensorimotor habits, on the one hand, and the
plane of pure memory, on the other, there are an infinite number of planes that correspond
to all different tensions of memory or to all different tones of mental life—two of which are
represented in the image by sections AB and AB . Bergson describes the distinction of and
515
Trevor Perri
gradual progression between these two united extremes in the following important passage
from Matter and Memory:
Between the plane of action—the plane in which our body has contracted its past into
motor habits—and the plane of pure memory, in which our mind conserves in all its
details the picture of our past life, we believe we can discover thousands and thousands
of different planes of consciousness, a thousand integral and yet different repetitions of
the whole of our lived experience.
(Bergson 1991: 241)
Without further considering here the mode of being of these virtual repetitions of the whole
of one’s past life, allow me to emphasize that, according to Bergson, each of these thousands of
different planes of consciousness or tones of mental life repeats, in a more or less contracted way,
the integral totality of one’s ever-growing past experience.
This is not to say, however, that the whole of one’s memory is always (or ever) given as a
plurality of completely distinct memories. Although the whole of memory is always included in
our present perception, according to Bergson, the continuous multiplicity of our consciousness,
duration, and memory is more or less distinguished into several memories that can be considered
independently depending on one’s circumstances. Bergson characterizes the way that the whole
of memory is experienced and more or less contracted as a “translation,” and he compares it to
the way that “a nebulous mass, seen through more and more powerful telescopes, resolves itself
into an ever greater number of stars” (1991: 166). And, in addition to this movement of “transla-
tion,” by which one’s past is more or less contracted and one’s memories more or less detailed and
distinct, Bergson suggests that by a process of “rotation” the past presents the aspect of the whole
of memory that is most useful (1991: 168).14 That is, just as we can focus on an individual note in
a melody while keeping the whole in mind, individual memories can be manifested in perception
without thereby being separated from the whole of the past that is preserved.
3. Conclusion: the movement of memory

Bergson’s philosophy of memory entails a rejection of traditional associationist theories of mind
(most clearly advocated by thinkers such as James Mill, Herbert Spencer, Alexander Bain, and
John Stuart Mill) according to which the mind contains a number of more or less independent,
atomistic sensations, perceptions, and ideas that are then associated with one another. Against
this view, Bergson argues that there is nothing stored “in” the mind (as things are stored in a
cabinet) and mental states are not primarily distinguished from one another. Rather, according
to Bergson, the entirety of one’s present and past experience forms a continuous whole that is
manifested differently depending on the demands imposed on us by virtue of the fact that we
are alive.
And despite his own occasional suggestions to the contrary, according to Bergson there are
not two, three, or even four types of memory. To suggest that this is the case is still to conceive
of the different forms of memory and of individual recollections as fixed, static, and something
ready-made (tout fait). As Bergson suggests with his reference to different tensions and tones of
one duration and one integral memory, memory is a single dynamic process in and through
which the past that is preserved in itself can be manifested in an infinite number of different
ways.15 Highlighting this dynamism and movement of the fundamental memory of duration as
it differentiates itself, Bergson states in his 1904 course on the history of theories of memory that
memory “is not a thing; it is a process; it is a movement” (2004: 108).
516
Henri Bergson
Notes
1 This is an abridged and slightly revised version of an article of the same title that was first published in
Philosophy Compass 9, 12 (2014): 837–47.
2 E.g., Eichenbaum (2011: 16) credits Bergson with being one of the first to recognize the difference
between habit and memory.
3 E.g., Schacter (1987: 504–5; 1996: 165) claims that Bergson recognized implicit memory and was an
early proponent of multiple memory system theory.
4 Scholars have pointed to the influence of Bergson’s philosophy of memory on literature, especially the
movement of modernism. See, e.g., Gontarski et al. (2013).
5 See, e.g., Whitehead (2009).
6 For especially illuminating recent discussions concerning Bergson’s philosophy of memory in particular,
see: Al-Saji (2004); Ansell-Pearson (2010); Bernet (2005); de Warren (2015); François (2008); Lawlor
(2003); Riquier (2009);Worms (2004). For his influential treatment of Bergson’s philosophy of memory,
see especially Deleuze (1991) and (1994).
7 Deleuze refers to this memory as “memory-contraction” (1991: 127), and Worms refers to it as
“immediate memory” (2000: 43).
8 See Bergson (1991: 15).
9 Vieillard-Baron (1991: 50–51) and Lawlor (2003: 131) point out that discussions of Bergson’s philoso-
phy of memory tend to focus either on the psychological or on the metaphysical aspects of this account.
This chapter aims to present both and to show their relation.
10 Deleuze (1991: 61) has described this in principle coexistence of the whole past with every present as
one of the paradoxes of memory. For a discussion of the paradoxes of memory, see Al-Saji (2004).
11 Hyppolite (2003: 113).
12 This pure memory of the totality of one’s past is unconscious, as Bernet (2005: 62–3) explains, in the
sense that it is conserved in consciousness yet independent of any present awareness.
13 Bergson (1920: 177).
14 For a helpful discussion of this rotation and contraction of memory, see Lawlor (2003: 51).
15 As Worms has elucidated: “Thus, one can say if one wishes: there are not two memories but three, or ultimately
there is only one” (2004: 164; original emphasis).
References
Al-Saji, A. (2004) “The Memory of Another Past: Bergson, Deleuze and a New Theory of Time.”
Continental Philosophy Review 37(2): 203–39.
Ansell-Pearson, K. (2010) “Bergson on Memory.” In Memory: Histories, Theories, Debates, eds. Susannah
Radstone and Bill Schwarz, pp. 61–76. New York: Fordham University Press.
Benjamin, W. (1969) Illuminations: Essays and Reflections, ed. Hannah Arendt. New York: Schocken.
Bergson, H. (1920) Mind Energy, trans. H. Wildon Carr. New York: Henry Holt.
—— (1946) The Creative Mind, trans. M. L. Andison. New York: Citadel.
—— (1991) Matter and Memory, trans. N. M. Paul and W. S. Palmer. New York: Zone.
—— (1998/1907) Creative Evolution, trans. Arthur Mitchell. Mineola, New York: Dover.
—— (2001) “Time and Free Will: An Essay on the Immediate Data of Consciousness,” trans F. L. Pogson.
Mineola, NY: Dover.
—— (2004) “Histoire des théories de la mémoire.” In Annales bergsoniennes II: Bergson, Deleuze, la
phénoménologie, ed. Frédéric Worms, pp. 41–149. Paris: PUF.
Bernet, R. (2005) “A Present Folded Back on the Past (Bergson).” Research in Phenomenology 35:
55–76.
Deleuze, G. (1991) Bergsonism, trans. Hugh Tomlinson and Barbara Habberjam. New York: Zone Books,
—— (1994) Difference and Repetition, trans. Paul Patton. New York: Columbia University Press.
de Warren, N. (2015) “Memory in Continental Philosophy: Metaphor, Concept, Thinking.” In Memory:
The History of a Concept, ed. Dmitri Nikulin. Oxford: Oxford University Press.
Eichenbaum, H. (2011) The Cognitive Neuroscience of Memory: An Introduction, 2nd edn. Oxford: Oxford
University Press.
François, A. (2008) Bergson. Paris: Ellipses.
Gontarski, S. E., Ardoin, P., and Mattinson, L. (eds.) (2013) Understanding Bergson, Understanding Modernism.
London: Bloomsbury.
517
Trevor Perri
Hyppolite, J. (2003/1949) “Various Aspects of Memory in Bergson,” trans. Anthea V. Colman. In Leonard
Lawlor (ed.), The Challenge of Bergsonism. New York: Continuum.
James, W. (1920) The Letters of William James. London: Longmans, Green.
Lawlor, L. (2003) The Challenge of Bergsonism. London: Continuum.
Ricoeur, P. (2004) Memory, History, Forgetting, trans. Kathleen Blamey and David Pellauer. Chicago, IL:
Riquier, C. (2009) Archéologie de Bergson: Temps et métaphysique. Paris: PUF.
Schacter, Daniel L. (1987) “Implicit Memory: History and Current Status.” Journal of Experimental
Psychology: Learning, Memory, and Cognition 13(3): 501–18.
—— (2001) The Seven Sins of Memory: How the Mind Forgets and Remembers. Boston, MA: Houghton
Mifflin.
Vieillard-Baron, Jean-Louis. (1991) Bergson. Paris: PUF.
Whitehead, A. (2009) Memory. London: Routledge.
Worms, Frédéric (2000) Le vocabulaire de Bergson. Paris: Ellipses.
—— (2004) Bergson ou les deux sens de la vie. Paris: PUF.
518
43
BERTRAND RUSSELL
Paulo Faria
A man who had not seen Mr. K. for a long time greeted him with the words:
“You haven’t changed a bit.” “Oh!” said Mr. K. and turned pale.
Bertolt Brecht, Stories of Mr. Keuner
1. An unsteady mind?
Bertrand Russell is widely known for his willingness to change his philosophical mind. He is
also now and again saddled with a reputation for carelessness about providing readers with clear
and convincing explanations of his turnabouts. More often than not, he would seem to have just
come up with a wholly fresh start and go happily on working in the new framework.
Russell’s ideas about memory are no exception to that perceived pattern of progress. He
delivered to the world, in the space of a few years (between 1912 and 1921), two starkly dif-
ferent theories of memory, themselves fragments of much wider and conspicuously contrasting
philosophical outlooks, with little explicit argument to account for the change.1
From 1912 to 1918, indeed, Russell consistently held a very radical form of direct realism
in the philosophy of memory, uncompromisingly eschewing the almost universally held rep-
resentational view of memory—remembering conceived as the present representation of a past
occurrence.2 Memory, according to Russell’s earlier account, is a wholly unmediated cognitive
relation with the past—the kind of relation Russell would call, following Grote and James, and
consistently since The Principles of Mathematics (1903), “acquaintance.” Then, from the summer
of 1918 onwards, most notably in the Beijing lectures of 1920–21 which were published as The
Analysis of Mind (1921), he embraced, in a momentous change of heart, an ostensibly represen-
tational view, on which memory is a relation between a mental image and a past occurrence,
accompanied by the belief that the occurrence is past—after which the picture would seem to
have remained in place, albeit with the occasional embellishment, in his later work.
The shift was, on the face of it, a dramatic step back from the radical realist agenda which,
taking his cue from Moore’s 1898 lecture “On the Nature of Judgment” (Moore 1899), Russell
had managed to consistently hold for two decades: the rejection of every form of a repre-
sentational approach to intentionality in general, that is, of every form of the view that the
mind relates to its objects through the interface of representations—call them ideas, impressions,
images, concepts, or what have you.
519
Paulo Faria
In what follows I present and briefly assess the development of Russell’s thinking about
memory, emphasizing, against the grain of the received view, the continuity in his thought.
2. Setting the stage: 1898–1911

The core notion of the early theory is that of acquaintance, by which Russell means a mode
of cognition which is both immediate and purely receptive. Taken together, these features bring
to the fore the most important feature, for our purposes, of Russell’s notion of acquaintance:
its wholly nonconceptual character. To be acquainted with an object is not to entertain the sup-
position, let alone to judge, that it is so-and-so. It is to be the subject of a dual relation with
the object such that being presented with the latter is a sufficient condition of its obtaining.
It is also the most basic mode of cognition, on which all others are grounded: “All cognitive
relations—attention, sensation, memory, imagination, believing, disbelieving, etc.—presuppose
acquaintance” (Russell 1914a: 1).
The notion had been playing a pivotal rule in Russell’s thought since the heyday of the
“revolt against idealism” initiated by Moore in 1898. The idea of a nonconceptual mode of
cognition, and of its logical priority over judgment, was indeed part and parcel of the Moore-
Russell picture of judgment as grounded on receptivity, rather than being an exercise of what
Kant and the Idealists had described as “the spontaneity of the understanding.” The picture
dictates a conception of philosophical inquiry on which philosophy proceeds through the analy-
sis of complexes (judgments and concepts) to the ultimate unanalyzable constituents we find
ourselves to be acquainted with.3
Immediate awareness in acquaintance is from the beginning opposed to the more rounda-
bout mode of cognition which is achieved through what Russell called “denoting.” In The
Principles of Mathematics (1903), that was the task of denoting concepts. These abstract constituents
of propositions, themselves objects of acquaintance, allow us to think of objects, such as infinite
totalities, which cannot be given to us in acquaintance: it is by virtue of being acquainted with
the denoting concept every odd number, for instance, that I succeed in thinking about the set of all
odd numbers. Denoting is thus “the inmost secret of our power to deal with infinity” (Russell
1903: 73).
Denoting concepts are dispensed with in the theory of descriptions introduced in “On
Denoting” (Russell 1905). Denoting is now the task of a class of linguistic expressions (“denot-
ing phrases”) to be sharply distinguished from singular terms, or what Russell calls “proper
names.” The gist of the theory is that denoting phrases, in spite of occurring as grammatical
subjects and seeming to behave as singular terms, are actually predicates—indeed, higher-order
predicates, specifically quantifiers, to be dealt with in parity with other expressions of logical
generality. No propositional constituent (a fortiori, no denoting concept) stands to them as their
meaning, unlike what happens with genuine singular terms.
The logical distinction between singular terms and denoting phrases has an epistemolog-
ical counterpart in the distinction between knowledge by acquaintance and knowledge by
description, which is the topic of Russell’s “Knowledge by Acquaintance and Knowledge by
Description” (1911). In that important paper, Russell’s first extended discussion of epistemo-
logical issues since the inception of the theory of descriptions,4 acquaintance is defined as the
converse of the relation of presentation of an object to a subject: “the word acquaintance is
designed to emphasize, more than the word presentation, the relational character of the fact
with which we are concerned” (Russell 1911: 109). Objects of acquaintance include sense
data, introspectible objects (among which, most likely, the self), universals, and relations. By
contrast, physical objects and other people’s minds are deemed inaccessible to acquaintance
520
Bertrand Russell
and hence only knowable by description. Humean doubts about the availability of the self as
an object of acquaintance are mentioned but, for the time being, dismissed on the ground that
self-acquaintance must be presupposed in the attempt to frame a suitable identifying description
of the self (1911: 110). As we will see, such assurance was short-lived.
3. Memory as acquaintance with the past: 1911–17

Soon after publishing “Knowledge by Acquaintance,” Russell undertook to rewrite the paper
for inclusion as a chapter in his introductory book The Problems of Philosophy, which he com-
pleted in the summer of 1911. The revised version contains a number of departures from its
predecessor, most remarkable among which is a reconception of acquaintance which makes it
possible for Russell to articulate, for the first time, a theory of memory.
The main novelty lies in the introduction of a dynamic approach to cognition, which
necessitates a revision of the previous definition of acquaintance as the converse of the rela-
tion of presentation. The most primitive form of cognition, approached for the first time
diachronically, now gets defined simply as direct awareness “without the intermediary of
any process of inference or any knowledge of truths” (Russell 1912: 46). Presentation, while
being a sufficient condition of acquaintance, is no longer a necessary condition: if we knew
only what is presented to us “we could not know anything about the past” (Russell 1912: 48),
not even by inference, as “we should never know that there was anything past to be inferred”
(1912: 49). Memory (episodic memory, the only kind of memory Russell ever dealt with at
any length—actually, the only kind of knowledge of the past he was ever prepared to count
as memory proper) is accordingly defined as retained acquaintance: acquaintance with what
has once been presented to us and is no more present. It is “the source of our knowledge
concerning the past,” including whatever inferential knowledge we may have of it, as “we
should never know that there was anything past to be inferred” (1912: 49) should we not
enjoy acquaintance with the past.
The main ostensible problem for the theory is that of accounting for retention—specifically,
making sense of the idea of being now acquainted with what has been and is no more. The prob-
lem has been generally deemed insoluble by Russell’s commentators, who often seem to see it
as a main motivation for Russell’s abandonment of the theory.5
By Russell’s lights, however, there simply was no such problem. Although he was careful to
acknowledge the role of images in recalling, he was adamant that such images cannot possibly
constitute the memory itself. The sheer fact that “the image is in the present whereas what is
remembered is known to be in the past” (1912: 115) should suffice to show that; but Russell
goes on to appeal to the fact that we often know, with some degree of certainty, how accurate
our present image is of the remembered object—a psychological fact ostensibly involving a
comparison between both which would be impossible unless the remembered object, as distinct
from the image, were somehow “before the mind”: “Thus the essence of memory is not consti-
tuted by the image, but by having immediately before the mind an object which is recognized
as past” (1912: 115).6
By contrast, accounting for fallibility and error in memory is more of a problem for Russell,
as these clash with the requirements that knowledge by acquaintance be (what knowledge by
description could not possibly ever be) both “perfect” and “complete” (1912: 47).7 The solu-
tion is to introduce degrees of self-evidence in what is remembered, according to the nearness
in time of what is remembered (1912: 115–16).
That sounds dangerously reminiscent of a representational theory of memory (on which
degrees of vividness of memory-images would stand for degrees of reliability of recollection),
521
Paulo Faria
so it is understandable that Russell should go for a more careful account. Such a new account
is to be found in the 1913 unfinished manuscript Theory of Knowledge and, for the most part, in
the six chapters of that book that Russell published in The Monist in 1914 and 1915, particu-
larly “Sensation and Imagination” (1915a) and “On the Experience of Time” (1915b). Its main
novelty is the distinction between what Russell now calls “immediate memory” and a kind of
memory which relates to the past in a more roundabout way—namely, that which pertains to
knowledge by description. Since dual cognitive relations do not admit error (Russell 1984: 169),
error in memory must involve judgment, and is only possible when past occurrences are singled
out by description, rather than by acquaintance.8
Unlike such knowledge by description of the past, immediate memory is “a two-term rela-
tion of subject and object, involving acquaintance, and such as to give rise to the knowledge that
the object is in the past” (1915b: 222). The definition is not intended to settle the issue whether
there is such a thing as immediate memory. However, Russell argues, it seems highly unlikely
that there should be no such thing: “It is indubitable that we have knowledge of the past, and
it would seem, although this is not logically demonstrable, that such knowledge arises from
acquaintance with past objects in a way enabling us to know that they are past” (1915b: 222). The
point is that we would not be able to infer from our knowledge of the present any proposition
concerning the past (not even as a hypothesis) unless we were currently conscious of something
past as past. Since knowledge by description of the past is liable to error, it cannot involve percep-
tion of a fact of which (whatever is meant by) “past” is a constituent: “Since, however, the word
‘past’ has significance for us, there must be perception of facts in which it occurs, and in such
cases memory must not be liable to error” (1915b: 224). Such memory must, therefore, be
acquaintance with the past.
4. Acquaintance on the rocks: 1913–19

As I remarked, Russell’s dismissal of Humean doubts about the self was short-lived. Already in
1913, in the unfinished Theory of Knowledge, we find him willing to grant that there is, after all,
no such thing as acquaintance with oneself, with the consequence that the word “I,” previously
treated as a proper name, “must stand for a description” (1984: 36; cf. 1914b: 441). The a priori
claim that self-acquaintance was presupposed in every attempt to frame a description of the
subject of experience now gives way to the negative a posteriori verdict from introspection: the
subject, as Hume claimed, is nowhere to be found in experience.
The novelty threatens to be seriously disruptive. If the subject cannot be known by acquaint-
ance, the theory mandates that it be treated as an inferred entity. Acquaintance must then be
conceived as a relation one of whose terms must be inferred rather than being directly experi-
enced. For the time being, that’s the best that Russell manages to do in order to keep the relational
account of experience rather than embracing a form of neutral monism (1914b: 439)—the
paradigm, by his lights, of a non-relational theory of experience (see the discussion in 1914a).
Early in 1914, though, the methodological principle which Russell called “the supreme
maxim in scientific philosophizing” was safely in place, and a revision of that interim account
became unavoidable. Introduced in “The Relation of Sense-Data to Physics” (Russell 1914c),
the principle—a form of Ockham’s razor by Russell’s lights—mandates that, whenever
possible, logical constructions be substituted for inferred entities. The principle will apply, among
such diverse entities as cardinal numbers or physical objects, to the self, and that makes for
increased pressure on the relational account of experience. It should accordingly come as no
522
Bertrand Russell
surprise that in the Harvard Lectures of 1914 Our Knowledge of the External World, Russell’s
first sustained application of the method of logical construction to empirical knowledge, the
scanty occurrences of the word “acquaintance” are to be taken, with one single clear exception
(1914d: 151), in the looser ordinary sense rather than in the technical, specifically Russellian,
one (see, e.g., 1914d: 73, 101, 123).9
From the summer of 1914 to that of 1917 Russell busied himself with non-philosophical
work, concentrating on his opposition to Great Britain’s involvement in the world war. He
resumed work in philosophy late in the summer of 1917, doing work that would eventuate
in the London lectures on The Philosophy of Logical Atomism, delivered early in 1918. There is
plenty of novelty in these lectures—not, however, with respect to acquaintance and the self.10
It was only in the summer of 1918 that, while in prison under the wartime Defence of
Realm Act, Russell resumed thinking about the consequences of the vanishing of the self for,
inter alia, the theory of memory. In a set of notes taken in August 1918, we find him at last
positively connecting the rejection of the subject to that of acquaintance itself, and to the need
for a reconception of memory:11
“Imperative to get rid of ‘Subject’.”

“Involves abandonment of distinction between sense-data and sensation.”
“Involves different theory of imagination and memory.”
“Tends to make the actual object in memory (e.g.) more remote from the present
mental occurrence than on the old theory.”
“Having extruded ‘acquaintance’ as an ultimate relation, we shall now say that,

in memory, the image is accompanied by a judgment that there was a past object
resembling it. It is in this way the image acquires ‘meaning’.”
(Russell 1986: 261–2)
In January 1919, Russell wrote a preliminary draft of what would eventually become The
Analysis of Mind, and in February he started writing the lengthy and important paper “On
Propositions: What They Are and How They Mean,” to be presented in the Aristotelian Society
meeting later in the year.
In that paper Russell is categorical that the unavailability of the self as an object of acquaint-
ance requires the abandonment of the relational conception of experience:
the theory which analyses a presentation into act and object no longer satisfies me. The
act, or subject, is schematically convenient, but not empirically discoverable. It seems
to serve the same sort of purpose as is served by points and instants, by numbers and
particles and the rest of the apparatus of mathematics. All these things have to be con-
structed, not postulated: they are not of the stuff of the world, but assemblages which
it is convenient to be able to designate as if they were single things.
(Russell 1919a: 25)12
The first effect of the rejection of the subject is thus to render necessary a “less relational” theory
of mental occurrences (1919a: 26). The articulation of such a theory will be the main burden
of Russell’s philosophy of mind for the ensuing couple of years, and it is in this new framework
that his later theory of memory acquires its distinguishing features.
523
Paulo Faria
5. Starting over: 1919–21

“If we had retained the ‘subject‘ or ‘act‘ in knowledge,” writes Russell in The Analysis of Mind
the whole problem of memory would have been comparatively simple. We could
then have said that remembering is a direct relation between the present act or sub-
ject and the past occurrence remembered: the act of remembering is present, though
its object is past. But the rejection of the subject renders some more complicated
theory necessary.
(Russell 1921: 163)
The more complicated theory was first adumbrated, as we saw, in the groundbreaking paper
“On Propositions” (Russell 1919a), whose novelty is announced by Russell to lie in the fact that
it chiefly results from “an attempt to define what constitutes ‘meaning’ and to dispense with the
‘subject’ except as a logical construction” (1919a: 1).13
Fulfillment of the first of these desiderata is secured by what Russell presents as the “impor-
tant discovery” at the heart of the new theory, namely, the acknowledgment that, contrary to
his earlier view, images lie at the heart of the theory of meaning.14
That is not to say, though, that everything was wrong with the relational account of expe-
rience. On the new theory, images are said to have meaning in virtue of being, as Hume
contended, “copies” of impressions (1919a: 22). That does not make them relational in nature;
nevertheless, “at least in the case of memory-images, they are felt to point beyond themselves, to
something which they ‘mean’” (1919a: 26). In accordance with the 1918 notes, such “pointing
beyond themselves” is now deemed to be achieved through judgment. Memory belief is accord-
ingly introduced as the belief that the memory-image is a copy of something which existed
previously (1919a: 27). In contrast with the early theory, then, images now become constitutive
of memory; and, given the role of memory belief in securing their reference to past images, all
memory turns out to involve judgment.
The Humean principle that images are “copies” of past impressions now gives pride of place
to the theory of memory in the analysis of mind. It is incumbent on the theory of memory, that
is, to account for the fact that images come to be taken as “copies” of past sensible experience
(1921: 158). But Hume’s principle is also responsible for the introduction of considerations
pertaining to causality in the account of the “meaning” of images, such that will give the theory
its most distinctive shape (the “copy” principle cashed out in terms of causation), and clinch the
case for the introduction of judgment as constitutive of memory. That is the role of a famous
thought experiment, introduced in The Analysis of Mind, about false memory.
The abandonment of direct realism should predictably bring about a further increase in
Russell’s concern over fallibility and error in memory, and that was indeed the case, with the
further twist added by the adoption of a causal account of meaning, which invites the kind of
Humean skepticism about causation displayed by the thought experiment: “There is no logical
impossibility in the hypothesis that the world sprang into being five minutes ago, exactly as it
then was, with a population that ‘remembered’ a wholly unreal past” (1921: 159).15
As the main ostensible challenge for the early theory was to account for retention of acquaint-
ance, the corresponding new challenge is to explain how we manage to compare present image
and past event. After all, the image is happening now while the event is in the past—which, as
we saw, is precisely why on the early theory images were deemed inessential to remembering.
The fact that the connection between image and past event is supposed to be causal now raises
524
Bertrand Russell
the Humean problem brought to the fore in the skeptical scenario: the memory-image which is
now happening may be wholly independent of any happening in the past.
Now Russell is very clear about what he cannot at this point embrace as a solution to the
problem. It is no use to appeal to the fact that we seem to know (at least sometimes) that images
are copies and (sometimes) that they are good or bad copies. That, admittedly, seems to suggest
that we have “a way of knowing the past, by means of which we can criticize the memory-
image. But I do not think that such inference is warranted” (1921: 161). Indeed, the extrusion
of acquaintance made that inference unavailable. There simply was no candidate left to being
that direct way of knowing the past.
Russell’s solution is that, given the logical independence between image and past occurrence,
belief is required as well: the mere occurrence of images “would not suggest any connection
with anything that had happened before” (1921: 160). It is the memory-belief, then, which
endows the memory-image with what Russell calls its “meaning”: “The mere occurrence of
images, without this feeling of belief, constitutes imagination; it is the element of belief that is
the distinctive thing in memory” (1921: 176). So all genuine memory involves judgment, and
is accordingly inherently fallible, contrary to the earlier theory.
For all that, Russell’s account of the kind of judgment which, on his new theory, constitutes
memory bears witness to an important, if widely overlooked, element of continuity in his thought.
On the earlier theory, all knowledge of the past was ultimately derivable from judgments of the
form “This is past,” where the reference of “this” was an object of (present) acquaintance (1915b:
224–5). Now the simplest kind of memory judgment is said to be of the form “This occurred,”
where “this” refers ambiguously to the present image and the past occurrence:
In the simplest kind of memory image we are not aware of the difference between an
image and the sensation which it copies, which may be called its “prototype”. When
the image is before us, we judge rather “this occurred”. The image is not distin-
guished from the object which existed in the past: the word “this” covers both, and
enables us to have a memory-belief which does not involve the complicated notion
“something like this”.
(1921: 179)
Such ambiguity, it should be emphasized, is essential to the memory judgment: if the word
referred only to the image, the judgment would be false (1921: 179–80).
My suggestion is that we should think of such reference to the past, even though it can
only be achieved in tandem with reference to present images, as a vestigial form of the
retained acquaintance of the earlier theory. All in all, as Judson remarks, “the closest analogue
to acquaintance-memory which does not employ acquaintance is MI [memory-image]-based
memory” (1988: 81). Seen in that light, Russell’s progress, rather than illustrating the per-
ceived discontinuity in his thought, brings to the fore the extent to which he managed to
abide, appearances notwithstanding, by what Quine called the “maxim of minimal mutila-
tion”: rather than being prompted by a wholesale rejection of the earlier account of memory,
presumably prompted by some insurmountable internal inadequacy of the latter, the new
approach is embraced as the one which best accommodates the view of memory as a distinc-
tive relation to the past once the relational account of experience, centered on the concept of
acquaintance, is given up. Nestroy’s remark, famously used by Wittgenstein as an epigraph for
Philosophical Investigations, is apposite here: “The trouble about progress is that it always looks
much greater than it really is.”
525
Paulo Faria
Notes
1 See, for the best available statement of the received view, Pears (1974). See also, antedating that paper,
Pears (1967), and the discussion in Urmson (1969) and Perkins (1973). In my departure from the
received view, I side with Judson (1988) and Martin (2015), trying to improve upon Faria (2010).
2 On the contrast between direct and indirect realism see Jordi Fernández’s Chapter 7, “The Intentional
Objects of Memory,” in this volume.
3 The conception guides the inquiry into the foundations of ethics and mathematics, respectively, in
Moore’s Principia Ethica and in Russell’s The Principles of Mathematics, both published in 1903.
4 Knowledge by acquaintance was first called by its name and cursorily discussed in the 1903 manuscript
“Points about Denoting”: cf. Russell (1994: 306).
5 See the excellent discussion in Martin (2015).
6 I don’t mean that this should be taken as the last word on the topic. See, for an enlightened defense
of the view of memory as retained acquaintance, which is immune to Russell’s later strictures on
(Russellian) acquaintance, Martin (2001).
7 See, for a careful examination of these requirements, Proops (2014).
8 On “descriptive memory,” see Russell (1984: 171–4).
9 By contrast, Russellian acquaintance will be emphatically reinstated, almost four years later, in the 1918
lectures on logical atomism (see Russell 1919b: 21, 32–3). Not for long, though.
10 It is worth noting, anyway, that late in 1917, on the occasion of a reprint of “Knowledge by Acquaintance
and Knowledge by Description” in the collection of essays Mysticism and Logic, Russell cared to add a
footnote stating that he was no longer prepared to hold the first-person pronoun “I” a proper name
(cf. Russell 1918: 214). He abides by that decision in the 1918 lectures, but goes no further.
11 My attention was drawn to these notes by the invaluable Martin (2015).
12 Like other logically constructed entities, the subject may exist (1919a: 26). What has to be excluded is
any theory which, like the earlier one, is incompatible with its non-existence.
13 The theory was further articulated in a series of lectures delivered in London in the spring of 1919.
A revised draft of these lectures was later incorporated in the Beijing lectures of 1920–21 which were
eventually published as The Analysis of Mind.
14 That is presented as the main outcome of Russell’s assessment of the claims of behaviorism, cf. 1919a:
10–15. The assessment is restated in The Analysis of Mind (Russell 1921: 151–4).
15 I don’t mean to suggest that Russell intends to be promoting skepticism about memory. He stresses
that the skeptical scenario in itself is uninteresting, its role being purely heuristic: “All I am doing is to
use its logical tenability as a help in the analysis of what occurs when we remember” (1921: 160). See
Baldwin (2001).The skeptical scenario does not appear in the more popular exposition of An Outline of
Philosophy (1927); yet it will recur as late as in the more technical Human Knowledge: Its Scope and Limits
(1948: 205, 228).
References
Baldwin, Thomas (2001) “Russell on Memory,” Principia 5: 187–208.
Faria, Paulo (2010) “Memory as Acquaintance with the Past: Some Lessons from Russell, 1912–1914,”
Kriterion 121: 149–172.
Judson, Lindsay (1988) “Russell on Memory,” Proceedings of the Aristotelian Society, New Series 88: 65–82.
Martin, M. G. F. (2001) “Out of the Past: Episodic Recall as Retained Acquaintance,” in C. Hoerl and
T. McCormack (eds.), Time and Memory, Oxford: Clarendon Press, pp. 257–84.
—— (2015) “Old Acquaintance: Russell, Memory and Problems with Acquaintance,” Analytic Philosophy
56: 1–44.
Moore, G. E. (1899) “The Nature of Judgment,” in G. E. Moore, Selected Writings, ed. Thomas Baldwin,
London: Routledge, pp. 1–19.
Pears, David (1967) Bertrand Russell and the British Tradition in Philosophy, London, Collins.
Pears, David (1974) “Russell’s Theories of Memory,” in George Nakhnikian (ed.), Bertrand Russell’s
Philosophy, reprinted in D. Pears, Questions in the Philosophy of Mind, London: Duckworth, 1975,
pp. 224–50.
Perkins, R. (1973) “Russell on Memory,” Mind 82: 600–601.
Proops, I. (2014) “Russellian Acquaintance Revisited,” Journal of the History of Philosophy 52: 779–811.
Russell, Bertrand (1903) The Principles of Mathematics, London: George Allen & Unwin.
526
Bertrand Russell
—— (1905) “On Denoting,” Mind 14: 479–93.

—— (1911) “Knowledge by Acquaintance and Knowledge by Description,” Proceedings of the Aristotelian
Society 11: 108–28.
—— (1912) The Problems of Philosophy, Oxford: Oxford University Press.
—— (1914a) “On the Nature of Acquaintance: Neutral Monism,” The Monist 24: 161–87.
—— (1914b) “On the Nature of Acquaintance: Analysis of Experience,” The Monist 24: 435–53.
—— (1914c) “The Relation of Sense-Data to Physics,” Scientia 16: 1–27.
—— (1914d) Our Knowledge of the External World as a Field for Scientific Method in Philosophy, London,
Routledge, 1993.
—— (1915a) “Sensation and Imagination,” The Monist 25: 28–44.
—— (1915b) “On the Experience of Time,” The Monist 25: 212–33.
—— (1918) Mysticism and Logic and Other Essays, London: Routledge, 2004.
—— (1919a) “On Propositions: What They Are and How They Mean,” Proceedings of the Aristotelian
Society, Supplementary Volume 2: 1–43.
—— (1919b) The Philosophy of Logical Atomism, London: Routledge, 2010.
—— (1921) The Analysis of Mind, London: Routledge, 1992.
—— (1927) An Outline of Philosophy, London: Routledge, 2009.
—— (1948) Human Knowledge: Its Scope and Limits, London: Routledge, 1992.
—— (1984) Theory of Knowledge: The 1913 Manuscript, ed. Elizabeth Ramsden Eames and Kenneth
Blackwell, London: Routledge.
—— (1986) “Three Notes on Memory [1918],” in John G. Slater (ed.), The Collected Papers of Bertrand
Russell, Vol. 8: The Philosophy of Logical Atomism and Other Essays, 1914–19, London: George Allen &
Unwin, pp. 261–3.
—— (1994) “Points about Denoting [1903],” in Alasdair Urquhart (ed.), The Collected Papers of Bertrand
Russell, Vol. 4: Foundations of Logic 1903–05, London and New York: Routledge, pp. 305–13.
Urmson, J. O. (1969) “Russell on Acquaintance with the Past,” The Philosophical Review 78: 510–15.
527
44
MAURICE HALBWACHS
Dmitri Nikulin
Collective memory is at the center of the debate about the relationship between individual
memory and the memory as defined by one’s participation in a group.1 The term was intro-
duced by Maurice Halbwachs (Reims, 1877–Buchenwald concentration camp, 1945),2 who
was a prominent French sociologist and wrote on a wide variety of topics, including statis-
tics, Leibniz’s philosophy, probability theory, religion, suicide, urbanism, the working class,
and social morphology and psychology. Yet Halbwachs became mostly known—more often
referred to than read—for his work on memory, to which he returned often. Halbwachs set
down the foundations for his work on memory with the groundbreaking study The Social
Frameworks of Memory (Les cadres sociaux de la mémoire, 1925), which introduces the notion of
collective memory within different social groups. His thinking on memory then continues in
numerous works, including a piece on collective memory among musicians, “The Collective
Memory of Musicians” (“La mémoire collective chez les musiciens,” 1939), which argues that
the memory of musicians is social in nature, the book The Legendary Topography of the Gospels
in the Holy Land (1941), which traces the changes in the ways the social memory of places is
transformed over time due to their new imaginary representation and interpretation, and the
posthumously published Collective Memory (1950), which discusses the relationship between
collective, individual, and historical memory, and collective memory’s temporal and spatial
determinations. Throughout these works, providing references to introspection, literary exam-
ples, historical observations, and empirical data, Halbwachs touches upon different aspects of
memory as related to language, history, and various social groups (family, class, nation, reli-
gious, and professional community). Yet the main insight of Halbwachs remains the same:
memory is primarily a social collective phenomenon, which means that one can only remember
within a group.
Although a thinker cannot be reduced to a set of intellectual influences, such influences can
nevertheless provide a useful insight into one’s work. The two major thinkers who made a strong
impact on Halbwachs, and who present two opposite approaches to memory, are Bergson and
Durkheim. While Bergson stressed the importance of introspection and individual reflection as
the source of his theoretical method, Durkheim was interested in collective representations and
argued for the social nature of our thought, pointing out the necessity of understanding any
individual phenomenon within the society or larger collectivity.3 Critical reflection on the work
of both thinkers sets two poles in Halbwachs’s account of memory: individual and collective.
528
Maurice Halbwachs
Another important intellectual connection that made a profound impact on Halbwachs was the
Annales group in Strasbourg, where he spent the years of 1922–33. Halbwachs stayed in close
association and debate with its founders Marc Bloch and Lucien Febvre. He actively participated
and also edited the group’s mouthpiece, Annales d’histoire économique et sociale. The Annales aimed
at nothing less than a revolution in historiography and thinking about the past by changing the
very idea of the subject of history, from single individuals portrayed as heroes or great personali-
ties to entire groups of people (where the subject of a historical study can be a small village or
even a sea), defined by their occupations and activities.
1. Collective memory
Halbwachs’s famous thesis concerning memory is that personal memory is always inscribed
into, and understood from, collective or group memory. Collective memory, then, becomes
the condition of the possibility of individual memory.4 According to Halbwachs (1992: 53),
“We can understand each memory as it occurs in individual thought only if we locate each
within the thought of the corresponding group.”5 On this interpretation, we always need oth-
ers in order to remember: in Ricoeur’s succinct formulation, “no one ever remembers alone”
(Ricoeur 2004: 122). That we remember with others and with the help of others, and that
it is possible to do so, seems uncontestable. Yet Halbwachs’s claim is much stronger: we only
remember with others, and it is impossible to remember otherwise. In this way, his position
presupposes a strong reduction of the individual memory to the collective one. The reason for
this is that we always act within a group, commonality, or community of which “we are but
an echo” (Halbwachs 1980: 45). Once one realizes the inevitable inscription of the individual
into social groups, the very unity of an individual and her memory “dissolve rapidly into a
multiplicity” (Halbwachs 1980: 49). On this interpretation, one—the individual—turns out to
be a split many and becomes one only when considered the one of a collective that embraces
and incorporates the many as one-and-many. Not only memory but knowledge too is a collec-
tive project for Halbwachs (1992: 176): it is only because we share thinking and remembering
with others that we are capable of thinking and remembering individually. Siding thus with
Durkheim against Bergson, in line with the critique of accepting an immutable human essence
or nature outside of social interaction, Halbwachs takes collective memory as “the totality of
traditions” preserved and passed on in the exercise of various social functions. Since any group
participates in many different social functions and activities at the same time, there are “at least
as many collective memories as there are functions” (Halbwachs 1992: 141). Memory, there-
fore, is itself a “collective function” (Halbwachs 1992: 183; cf. Halbwachs 1980: 23).6
In support of this rather strong position, Halbwachs provides two arguments, which, how-
ever, are not fully articulated and are stated rather casually. The first is the argument from
dreams, to which Halbwachs dedicates the entire first chapter of The Social Frameworks of Memory,
where he takes memory and dreaming as two opposite functions of the recalling mind.7 Since in
dreams our mind “is most removed from the society” (Halbwachs 1992: 42) and “dreams can-
not evoke the recollection of events or of complex pictures” (Halbwachs 1992: 174), memory
has to be social and collective. This argument does not seem to work, since one cannot exclude
that, even if our dreams are haphazard and lacking in the same kind of coherence as our waking
experience, their origin quite often lies in a social event or interaction.
The second argument in favor of collective memory comes from psychological self-
observation or “introspection.” Halbwachs justifies the relevance of introspection, which
for him provides verifiable results that are universally applicable to everyone, by suggesting
that the assumed interior/exterior distinction as implied in the act of introspection is a false
529
Dmitri Nikulin
one, since it presupposes an isolated individual, whereas we in fact are always part of a group.
Therefore, if “we consider not an isolated individual but a group of people who live together
in society, this distinction makes no sense” (Halbwachs 1992: 168). Moreover, on this inter-
pretation, a purely exterior act of observing an individual is then neither meaningful nor
possible. For this reason, the introspective act of recollection should be collective and thus
an individual act of self-observation should be universally valid. Yet, this argument appears
to beg the question in that it presupposes the collectivity of (memory) experience in order to
justify introspection as the method of establishing collective memory.
2. Framework and group

Memory for Halbwachs is thus always collective. In order to describe its constitution and pres-
ervation, he introduces the concept of social frames or contexts (cadres sociaux), which are social
mechanisms and practices that allow for the transmission of the collectively shared memory:
“Collective frameworks are . . . the instruments used by the collective memory to reconstruct
an image of the past which is in accord, in each epoch, with the predominant thoughts of the
society” (Halbwachs 1992: 40). Memory, then, is not just a combination or sum total of indi-
vidual memories but a social form that allows each individual to remember and constitute her
memorable experiences, outside of which one simply cannot remember.8 The framework, then,
“is made of notions—notions of persons and facts” (Halbwachs 1992: 40). These notions are
mutually bound and interlinked, and thus constitute “a concatenation of ideas and judgements”
(Halbwachs 1992: 176, cf. p. 53).
Since no one can remember by oneself, one does it only within a group, which is the main and
defining social unit for Halbwachs. Each society consists of many different groups (Halbwachs
1980: 24) and every group tends to preserve and maintain itself through its customs, institutions,
and traditions (“It resists with all the force of its traditions” (Halbwachs 1980: 134)), as well
as through collectively exercised and rehearsed memories. As such, a social group is a kind of
affective community in which we share, conserve, and renew our memories (Halbwachs 1980:
30–33). However, an individual is not reduced to a group, because one participates simulta-
neously in many groups (“constellations,” Halbwachs 1980: 41) of various sizes and forms of
activity (family, professional, leisurely, class, religious, national, etc.). Moreover, this participa-
tion can change over one’s life (Halbwachs 1980: 72, 76), due to circumstances and because
traditions that define social groups can change and be replaced by other traditions (Halbwachs
1992: 185). Again, one is not an isolated and “undivided” individual but is one-and-many in a
group, which sets the frame, form, and limit of what one can perceive about the past, i.e., what
and how one can remember and recall it. Collective memory is thus a group memory defined
by a particular social framework.
Because we belong to many groups at the same time, we share many different memories
(Halbwachs 1980: 104), and we lose them when we change groups. The social frame as the form
for collective memory established by each group is not a priori. It is itself constituted by the past
experiences that are preserved in a group and follow a group’s particular “mentality” (Halbwachs
1992: 59), which the group builds and follows based on its memories and experiences. Each
group, therefore, has its own different tradition or non-a priori “logic” (Halbwachs 1992: 52–63,
83; Halbwachs 1980: 63–68), or its own “original collective memory” (Halbwachs 1980: 78).
But if this is the case, are group memories mutually incommensurable or can they be understood
or remembered within another group? Besides, do collective memories depend on the size of a
group? Halbwachs does not provide answers to these questions, yet for him the groups are not to
be mutually sealed off, because we all participate in the same society and share the same language
530
Maurice Halbwachs
(Halbwachs 1992: 43). As Aristotle has famously argued, one cannot have too many friends, but
just as many as one needs for a shared life.9 If one follows this line of reasoning and if one takes
collective memory to constitute the life of a collective, then the memory in and of a small group
might be different from that in and of a large group, at least in its media (oral vs. written), lasting
effect, and scope.
This means that we can communicate with members from other groups and thus learn about
and from their collective memory. An objection to Halbwachs here might be that once a col-
lective framework is irretrievably lost (e.g., of a Paleolithic community), it is hard to see how its
memories can be reconstituted based solely on the preserved artifacts in the absence of a collec-
tive practice within a framework. Or, if a wormwood smell brings back a childhood memory
related to a community long gone with social practices that do not exist anymore, do I still
remember within and under the spell of the framework of a group? Yet this group does not exist
anymore and is now utterly imaginary, persisting only within my memory, which, according
to Halbwachs, is supported by the extant collective memory of other currently existing groups.
But how and if such an imaginary group is possible and how it can communicate its memories
to other groups remains ultimately unexplained. One could say that once a concatenation of
collective memories is broken and the members of its group are gone, its memory becomes lost
forever to a history.
3. Individual, collective, and historical memory

Halbwachs, however, does not reject individual memory or dissolve it altogether in collective
memory. Since collective memory is a totality of traditions, each one being a function of a
collectively shared activity, each individual memory is itself only a functional aspect of the col-
lective group memory—a “viewpoint on the collective memory” (Halbwachs 1980: 48). Every
act of remembrance is that of an individual, yet is always exercised and inscribed within the
frame of collective memory. This frame constitutes the limit beyond which individual memory
cannot go, because collective memory is the totality of the interwoven remembrances medi-
ated by social practices that make individual memory both possible and meaningful: “While the
collective memory endures and draws strength from its base in a coherent body of people, it is
individuals as group members who remember” (Halbwachs 1980: 48).
Memory is about the past, as it is reproduced under the conditions and from the point of
view of the present (Halbwachs 1992: 47–49).10 Yet the past is present through both memory
and history, even if it comes alive differently in the two. Halbwachs distinguishes between his-
torical and autobiographical memory (Halbwachs 1980: 50–87): historical memory is external
and transmitted mostly through written records, whereas autobiographical (which we now call
episodic) memory is internal and always collective, kept within by a group.11 Historical memory
is “the sequence of events remembered in national history” (Halbwachs 1980: 77), i.e., it is the
memory of the largest group, that of a nation.
However, since the nation is a modern construction, itself historical, history is equally a con-
struction, which artfully and skillfully arranges the past according to current social and political
categories. Halbwachs thus excludes the possibility of historical memory of smaller social groups,
making history strongly historiographic, built according to the dominant ideal of what a nation
is meant to achieve in a long perspective. This makes history an inevitably long, seemingly
universal, forcibly unified, and glaringly systematic narrative about the past told by the rules and
with the means of the present, split into definitive periods, in order to bring order, structure,
purpose, normative thrust and meaning (of progress, decline, etc.) to its story. History is thus
what distinguishes one period from another in an orderly way (Halbwachs 1980: 57). Collective
531
Dmitri Nikulin
memory, on the contrary, is passed on by and within the existing perspicuous or tacit traditions
and customs, so that nobody can single-handedly own or control it.
Collective memory is relatively short and is the depository of a tradition; its boundaries are
irregular and uncertain, and it is continuous in its scope and reach. History, on the contrary,
extends over long periods of time—much longer than a collective memory can reach back—
and is a record of changes (Halbwachs 1980: 86). History is a reconstruction of the past, which
it divides into artificially constructed epochs.12 But in order to tell the story about the entire
past, history inevitably has to make abridgements (Halbwachs 1980: 106), while keeping most
notable facts (Halbwachs 1980: 78) that are selected from the contemporary perspective, which,
however, being itself historical, might change over time.
History is therefore opposed to collective memory.13 History for Halbwachs begins where
memory ends and the tradition declines: “General history starts only when tradition ends and
the social memory is fading or breaking up” (Halbwachs 1980: 78). History is preserved through
a coherent written narrative, when the spoken word dies (Halbwachs 1980: 79). With the
advent of modernity, social bourgeois memory is much enlarged (Halbwachs 1992: 133), yet
becomes shallow and superficial. It is modern (re)constructed, universal history that comple-
ments collective memory and often substitutes for it.
Furthermore, there is one unified diachronic history, which is construed as national yet can
be extended into universal—yet many synchronic collective memories, often dissociated and
partial.14 Collective memory stays within the limits of a group and is always fragmented, whereas
history places itself outside and beyond particular groups and hence is construed as national and
universal. Moreover, unlike historical memory, collective memory is present not only in texts
but much more in oral transmission and communal practices. Most importantly, collective
memory for Halbwachs pays attention to similarities, whereas history is interested in differences
and tends to neglect resemblances.15
To sum up, history and collective memory for Halbwachs are constituted differently and in
opposition to each other. History is one (“unitary”), tends to be universal, dead (a “crowded
cemetery,” Halbwachs 1980: 52), is for few specialists only, divided into (artificial) periods,
based on a written record of events, interested in differences, and extends over long periods of
time. Collective memory, on the contrary, is plural, so that “there is ‘no universal memory’”
(Halbwachs 1980: 84), but always “several collective memories” (Halbwachs 1980: 83); it is
alive and often oral, open and accessible to everyone, continuous, a depository of traditions,
lives off resemblances, and is relatively short (Halbwachs 1980: 78–84).
4. Recollection and forgetting

But if we remember with others, do we also forget with others? In other words, is forgetting
a natural mechanism of memory dying off when it is no longer rehearsed and needed—or is it
also a collective enterprise? Halbwachs offers noteworthy insights on oblivion, which, however,
remain without much supporting argumentation and which differ in the Social Frameworks and
in the Collective Memory. In the Social Frameworks, he suggests that a recollection is richer when
it arises out of the interaction and intersection of a number of frameworks. Forgetting, then, is
the “deformation” of recollections, when social frameworks within which we recollect “change
from one period to another” (Halbwachs 1992: 172). When the social frame or the form of
remembering is gone, we no longer remember. But Halbwachs makes an even stronger claim:
society unites people, overcoming the atomic unitary dispersion of individuals and making them
one out of many. For this reason, “society tends to erase from its memory everything that might
separate individuals, or that might distance groups from each other” (Halbwachs 1992: 182–183).
532
Maurice Halbwachs
Oblivion as deformation might be considered a passive process that results from the transforma-
tion of social groups that cancels one’s participation in them, but the social erasure of everything
that separates people suggests a much more active process. Unfortunately, Halbwachs does not
describe its mechanism. Clearly, the group erasure should be unwitting and remain concealed
from the members of a group. If people start paying attention to the fact that they are made to
forget something, they will never forget it for the reason that the attempt to force people to forget
without their will or consent will make the mechanism of oblivion explicit and conspicuous and
thus remembered in a group.
In the Collective Memory, we also find a different account of oblivion, which suggests that
Halbwachs was returning to the topic for many years. Indeed, if collective memory is that which
unites people in groups and eventually in society, its dissolution suggests the end and the impos-
sibility of the social. This means that we forget when we are separated from a group, which
provides us with the “body of remembrances that we share in common” (Halbwachs 1980: 30).
Forgetting, then, is the separation from a group and hence is equally collective, as is remem-
bering. Halbwachs makes a rather astonishing claim that in fact “there is no such thing as an
absolute void in memory . . . we forget nothing” (Halbwachs 1980: 75). That the past remains
in its entirety in our (individual) memory is Bergson’s thesis, but Halbwachs’s understanding
of memory as collective leads him to radically reinterpret the assertion that nothing is ever
lost from memory, for we can reconstruct the past based on the data available in social memory.
Apparently, this should always be possible, because of the richness and saturation of collective
memory that has the form of a system, i.e., is somehow (well) organized (Halbwachs 1992: 53).
This leads us to the question of recollection as reconstruction.16
Halbwachs does not distinguish between different kinds of memory, although he mentions
habitual memory (which is “la mémoire habitude” in Bergson’s Matter and Memory (Halbwachs
1992: 47)) and autobiographical memory, which is mostly called episodic today. Most impor-
tantly, Halbwachs does not differentiate between memory and recollection, in contrast to Aristotle,
who in his short treatise On Memory and Recollection (De memoria et reminiscentia) makes an impor-
tant distinction between memory (mn m ) and recollection (anamn sis). Aristotle’s distinction
has been differently reproduced and interpreted by many thinkers throughout history. Memory
for Aristotle is a state of having an image of a past event in a non-discursive act that retrieves the
memory without following any consecutive steps in order to extract it (De mem. 449b24–5).
These images can be kept in and retrieved from imaginary memory places or loci memoriae,
which is the ancient mnemonic method ascribed by Cicero and Quintilian to Simonides.
Recollection, on the contrary, is a discursive process, inquiry, or search similar to a logical
argument or syllogism that goes from premises to conclusion, which is the sought-after recol-
lection that is recovered in and by this process (De mem. 453a10).17 Memory one can have; to
recollection one should come.
For Halbwachs, every fact and act of the past can be recovered by recollection that extracts
the past from the collective memory. This means that remembrance or recollection is not a copy
of things past, but the active reconstruction of the past “achieved with data borrowed from the
present” (Halbwachs 1980: 69; cf. pp. 51, 68–71). Collective memory, then, is not established
forever as an immutable and authoritative set of remembrances—rather, the memories of the past
are constantly revised and often replaced by new ones, according to the current views and concerns
of a group (Halbwachs 1980: 54). Collective remembering as reconstructive recollection is thus
always a creative process of retrieving the meaning of the shared past.
But how do we make this reconstruction? For Halbwachs, “our recollections, each taken
in itself, belong to everybody; but the coherence or arrangement of our recollections belong
only to ourselves” (Halbwachs 1992: 171). In other words, the necessary data for a recollection
533
Dmitri Nikulin
as reconstruction is kept and supplied by the group, but the actual remembrance is an act of an
individual, which is her own but only possible within a collective group memory as the pre-
condition and the limit for such remembering. In order to (collectively) remember, we need to
locate our recollections as memories with the help of a kind of loci memoriae, the “landmarks that
we always carry within ourselves” (Halbwachs 1992: 175; cf. Halbwachs 1980: 134).18 Yet an
act of individual recollection within a group memory is the act of arrangement and reasoning: “in
order to remember, one must be capable of reasoning and comparing and of feeling in contact
with a human society that can guarantee the integrity of our memory” (Halbwachs 1992: 41).
Therefore, to retrieve the past, one needs to reason, thus moving from one recovered memory
to another, until one comes to the sought-after recollection. In this sense, Halbwachs’s recollec-
tion as a reasoned arrangement of the past is very similar to Aristotle’s anamnetic recollection,
except for being a process supported by the memory of others.
Although Halbwachs does not discern between memory and recollection, oftentimes he in
fact describes (collective) recollection without being aware of its distinction from (collective)
memory. On the very last pages of his Social Frameworks of Memory, Halbwachs unwittingly
comes to distinguish between memory and recollection:
Recollections which have not been thought about for a long time are reproduced
without change [as memories]. But when reflection begins to operate, when instead
of letting the past recur, we reconstruct it through an effort of reasoning [as recollec-
tions], what happens is that we distort that past, because we wish to introduce a greater
coherence.
(Halbwachs 1992: 183)19
The duality and complementarity of memory and recollection also appears under the guise of
the distinctions between image and name (Halbwachs 1992: 71), image and notion (p. 61),
image and idea (pp. 174–175, with reference to Bergson, Plato, and Spinoza), tradition and idea
(p. 188), tradition and reason (p. 184). Yet Halbwachs never recognizes his rediscovery and the
difference between memory and recollection.
I’d like to conclude by suggesting a new concept, that of collective recollection.20 Unlike collec-
tive memory that allows us to have remembrances of the past retrieved in the form of images that
we might have activated when needed, collective recollection is a process of careful reconstruc-
tion of the past in a reasoned process with others, by which we come to retrieve, understand,
and appreciate the past. Collective recollection can be characterized by the following features:
(1) While collective memory remembers what it remembers, often biased by a contemporary
political agenda, collective recollection is (should be) critical, insofar as it allows for, and pre-
supposes, the discussion of its sources and starting points, which should be put to scrutiny
according to the standards of public, reciprocal, and transparent justification.21
(2) Being collective, such recollection is participatory, open to everyone and thus potentially
democratic.
(3) Collective recollection presupposes commonly shared reasoning, which provides reasons for
recollection. Such public reasoning in principle can always be rationally contested, and thus
be the subject to rational (counter)argument and criticism.
(4) Collective recollection is thus (re)constructive and constructivist, insofar it is a process, and
not an act, coming to a publicly shared and recognized recollection as achieved, and not as
a given.
534
Maurice Halbwachs
Defined in this way, collective recollection might be a useful concept in memory studies, as well
as in contemporary social and political debate.
Acknowledgment
I want to thank Rainer Forst, Jeremy Gauger, Joseph Lemelin, Kourken Michaelian, and the
members of the Forschungskolleg in Bad Homburg for most helpful remarks on the chapter.
Notes
1 See Nora (1996), Olick (1999), Sutton (2009), and Gensburger (2011).
2 Cf. Semprun (1994: 51–53).
3 See Durkheim (1898). See also Mary Douglas’s introduction to Halbwachs (1980: 6–9), and Keightley
and Pickering (2012: 93). In Halbwachs, collective memory is related to collective consciousness as
linked to collective beliefs and class-consciousness. See Halbwachs (1992: 187).
4 Cf. Le Goff (1992: 94–95); Keightley and Pickering (2012: 93–94).
5 “When I remember, it is others who spur me on; their memory comes to the aid of mine and mine
relies on theirs (si je me souviens, c’est que les autres m’incitent à me souvenir, que leur mémoire vient
au secours de la mienne, que la mienne s’appuie sur la leur) . . . [Memories] are recalled to me exter-
nally, and the groups of which I am a part at any time give me the means to reconstruct them (ils me
sont rappelés du dehors, et . . . les groups dont je fais partie m’offrent à chaque instant les moyens de les
reconstruire)” (Halbwachs 1992: 38).
6 Cf. Cassirer (1953).
7 Cf. Nietzsche (1996), aph. 12.
8 “No memory is possible outside frameworks used by people living in society to determine and retrieve
their recollections” (Halbwachs, 1992: 43). Cf. “The individual calls recollections to mind by relying on
the frameworks of social memory” (Halbwachs, 1992: 182).
9 Aristotle (2009), 1170b20–1171a20.
10 Cf. Coser (1992).
11 See Lewis A. Coser’s introduction to Halbwachs (1992: 23–24).
12 “History is neither the whole nor even all that remains of the past. In addition to written history, there
is a living history that perpetuates itself through time and permits the recovery of many old currents
that have seemingly disappeared” (Halbwachs 1980: 64).
13 See Nikulin (2008).
14 According to Geary, the distinction between collective and historical memory consists in that “the former
is the fluid, transformative and enveloping lived tradition of a social group. The latter is analytic, critical,
and rational, the product of the application of specialized scientific methodology” (Geary 1984: 11). Cf.
Hutton (1993: 147–153).
15 “[L]’histoire s’intéresse surtout aux différences, et fait abstraction des ressemblances sans lesquelles
cependant il n’y aurait pas de mémoire” (Halbwachs 1980: 75).
16 Cf. “[T]he various groups that compose the society are capable at every moment of reconstructing their
past” (Halbwachs 1992: 182). “[W]e can find in society all the necessary information for reconstructing
certain parts of our past represented in an incomplete and indefinite manner, or even considered com-
pletely gone from memory” (Halbwachs 1980: 75).
17 See Nikulin (2015b: 36–37, 60–66).
18 Most groups “engrave their form in some way upon the soil and retrieve their collective remembrances
within the spatial framework thus defined” (Halbwachs 1980: 156).
19 Cf. “We might perhaps be led to distinguish two kinds of activities within social thought: on the one
hand a memory, that is, a framework made out of notions that serve as landmarks for us and that refer
exclusively to the past [memory]; on the other hand a rational activity that takes its point of depar-
ture in the conditions in which the society at the moment finds itself, in other words, in the present.
This memory functions only when under the control of reason [recollection]” (Halbwachs 1992: 183;
cf. p. 189). This latter kind of memory Halbwachs also calls “a collective reflection” (1992: 183).
20 See Nikulin (2015a: 21).
21 Cf. Forst (2012).
535
Dmitri Nikulin
Further reading
Bergson, H. (2004 [1896]) Matter and Memory [Matière et mémoire], trans. N. M. Paul and W. S. Palmer,
Mineola, NY: Dover.
Bloch, M. (1953 [1949]) The Historian’s Craft: Reflections on the Nature and Uses of History and Those Who
Write It [Apologie pour l’histoire, ou métier d’historien], trans. Peter Putnam, New York: Alfred A. Knopf.
Febvre, L. (1925 [1922]) A Geographical Introduction to History [La terre et l’évolution humaine: introduction
geeographique à l’histoire], trans. E. G. Mountford and J. H. Paxton, New York: Alfred A. Knopf.
Halbwachs, M. (1941) La topographie légendaire des Évangiles en Terre sainte. Étude de mémoire collective, Paris:
Presses Universitaires de France. The conclusion of the book is translated and published in Halbwachs
(1992), pp. 191–235.
Halbwachs, M. (1950 [1939]) “The Collective Memory of Musicians [La mémoire collective chez les
musiciens],” in The Collective Memory, trans. Francis J. Ditter and Vida Yazdi Ditter, New York: Harper
& Row, pp. 158–186.
Michaelian, K., and Sutton, J. (forthcoming) “Collective Memory,” in Ludwig and Jankovich (eds.), The
Routledge Handbook of Collective Intentionality, London: Routledge.
Olick, J. K., Vinitzky-Seroussi, V., and Levy, D. (eds.) (2011) The Collective Memory Reader, Oxford:
References
Aristotle (1957) “On Memory and Recollection,” in On the Soul, Parva Naturalia, On Breath, trans. W. S. Hett,
Cambridge, MA: Harvard University Press, pp. 288–315.
Aristotle (2009) The Nicomachean Ethics, trans. David Ross, Oxford: Oxford University Press.
Cassirer, E. (1953) Substance and Function, trans. W. C. Swabey and M. C. Swabey, New York: Dover.
Coser, L. A. (1992) “The Revival of the Sociology of Culture: The Case of Collective Memory,”
Sociological Forum 7: 365–373.
Durkheim, E. (1898) “Representations individuelles et representations collectives,” Revue de métaphysique
et de morale 6: 273–302.
Forst, R. (2012) The Right to Justification: Elements of a Constructivist Theory of Justice, New York: Columbia
University Press.
Geary, P. J. (1994) Phantoms of Remembrance: Memory and Oblivion at the End of the First Millenium, Princeton,
NJ: Princeton University Press.
Gensburger, S. (2011) “Réflexion sur l’institutionnalisation récente des memory studies,” Revue de synthèse
132: 411–433.
Halbwachs, M. (1980 [1950]) The Collective Memory [La mémoire collective], trans. Francis J. Ditter and Vida
Yazdi Ditter, New York: Harper & Row.
—— (1992 [1925]) The Social Frameworks of Memory [Les cadres sociaux de la mémoire], in On Collective
Memory, trans. Lewis A. Coser, Chicago, IL: Chicago University Press, pp. 35–189.
Hutton, P. H. (1993) History as an Art of Memory, Hanover, NH: University Press of New England.
Keightley, E., and Pickering, M. (2012) The Mnemonic Imagination, New York: Palgrave.
Le Goff, J. (1992) History and Memory, trans. Steven Rendall and Elizabeth Claman, New York: Columbia
University Press.
Nietzsche, F. (1996) Human, All Too Human: A Book for Free Spirits, trans. R. J. Hollingdale, Cambridge:
Nikulin, D. (2008) “Memory and History,” Idealistic Studies 38: 75–90.
—— (2015a) “Memory in Ancient Philosophy,” in D. Nikulin (ed.), Memory: A History, Oxford: Oxford
—— (2015b) “Memory in Recollection of Itself,” in D. Nikulin (ed.), Memory: A History, Oxford: Oxford
Nora, P. (ed.) (1996 [1992]) Realms of Memory [Les lieux de mémoire], 3 vols., trans. Arthur Goldhammer,
New York: Columbia University Press.
Olick, J. K. (1999) “Collective Memory: The Two Cultures,” Sociological Theory 17: 333–348.
Ricoeur, P. (2004 [2000]) Memory, History, Forgetting [La mémoire, l’histoire, l’oubli], trans. Kathleen Blaney
and David Pellauer, Chicago, IL: University of Chicago Press.
Semprun, J. (1994) L’écriture ou la vie, Paris: Gallimard.
Sutton, J. (2009) “Looking Beyond Memory Studies: Comparisons and Integrations,” Memory Studies 2:
299–301.
536
45
FREDERIC BARTLETT
Brady Wagoner
Frederic Bartlett is widely credited with the insight that remembering is a reconstructive
process, and that social factors play a principle role in it. These arguments were forcefully
advanced in his classic book Remembering: A Study in Experimental and Social Psychology
(1932/1995). In this work, he set his approach in contrast to that of Hermann Ebbinghaus’s
(1885/1913) Memory: A Contribution to Experimental Psychology, the other major classic in the
psychology of memory. Whereas Ebbinghaus treated memory as a self-contained faculty for
retaining bits of information, Bartlett saw it as a creative and situated activity that was closely
tied to imagination (see also De Brigard, Chapter 10, this volume). These differences led
Bartlett to adopt an opposing experimental methodology for studying remembering and to
emphasize characteristics of it in his theory that Ebbinghaus had ignored, such as meaning,
context and social relations.
In this chapter, I discuss the key features of Bartlett’s approach to remembering and in
so doing contextualize them within his time as well as some present-day theorizing. For
an extended explication of Bartlett’s theory, its development and appropriation by others
readers should consult The Constructive Mind: Frederic Bartlett’s Psychology in Reconstruction
(Wagoner 2017a). The present chapter begins by describing the context and rationale for his
seminal ‘experiments on remembering’ from which he later developed his celebrated theory.
It then proceeds to outline Bartlett’s reconstructive theory of remembering mainly through an
analysis of his well-known concept of schema and how the memory researchers that followed
him have transformed it from his time until the present day. Finally, the chapter highlights the
often-neglected social aspects of his theory in order to show that Bartlett aimed to develop an
integrated sociocultural and psychological approach to remembering.
1. From philosophy to experiments on remembering

Bartlett entered the University of Cambridge in 1912 to do an undergraduate degree in “Moral
Sciences.” Having already completed degrees in Logic, Sociology and Ethics, he recalled hav-
ing at this time “a very predominant philosophical bias in [his] outlook upon life” (Bartlett
1936: 39). This was however short-lived. He was definitively turned off this path after two
encounters: first, he was asked by Russell, Moore and MacTaggart to weigh in on the question
of “whether the rats said to be seen in an advanced state of delirium tremens were real or not”
537
Brady Wagoner
during a Cambridge intellectual sparring activity called “squash.” After trying to agree with all
opposing positions, he went away feeling depressed. Second, he presented a paper defending
Bergson’s Creative Evolution (see Perri, Chapter 42, this volume) at the moral science club, for
which he was severely criticized by an audience already very unsympathetic to Bergson.
Fortunately, Bartlett was also becoming better acquainted with psychology. He took a class by
analytic psychologist and philosopher James Ward as well as a laboratory course in experimental
psychology run by C.S. Myers and assistant Cyril Burt. The latter class required heavy metho-
dological training based on German-style experimentation. Of particular influence here was the
Würzburg School’s studies and their emphasis on the active character of mind. This idea was
based on philosopher Brentano’s notion of “intentionality” (1874/1973). Bartlett’s own concept
of “attitude” (a holistic orientation to the world) is a development of this tradition: he used it to
both describe task-directedness (e.g., “I must accurately recall all the details”) and general feelings his
participants had of the stimulus material (e.g., “reminds me of what I read as a child” or “it is not
English”). Moreover, the notion of “intentionality” also comes close to his characterization of all
mental processes as involving “an effort after meaning,” whereby the person connects up what is
given with something else, described as a “setting,” “scheme,” or “schema” (see below).
On the opening day of a new well-provisioned psychology laboratory in 1913, director and
financer C.S. Myers asked Bartlett if he would carry out some visual perception experiments
for visitors. The diversity of interpretations of the same figures so fascinated Bartlett that it
inspired him to develop his own program of experiments on “some problem of perceiving and
imagining” (Bartlett 1916a), focusing on what the subject contributed to the response, as the
Würzburg School had done. In these experiments, he demonstrated the role of attitudes, values,
interpretations, and feelings on perceiving and imagining. These studies directly led Bartlett to
his famous experiments on remembering. It should be noted here that Bartlett considered the
mind as a unity (rather than separate mental faculties). He thought mental processes differ in
degree rather than kind, such that distinctions remain to some extent arbitrary. As such, studies
on perceiving and imagining directly contribute to his theory and analysis of remembering (for
a summary of the contemporary debate on whether imagination and memory differ in degree
or kind, see Perrin and Michaelian, Chapter 18, this volume).
Bartlett’s celebrated experiments on remembering were actually first framed as “contribu-
tions to the experimental study of the process of conventionalization” (Bartlett 1916b). Building
on a tradition of diffusionist anthropology, his mentor W.H.R. Rivers (1912) coined the term
“conventionalization” to describe the process by which foreign cultural elements (such as deco-
rative designs or folk stories) transform in the direction of the receipt group’s conventions when
they enter into it. Bartlett had actually hoped to become an anthropologist but Rivers convinced
him that rigorous training in psychology would be the best preparation for it. The anthropo-
logical influence can also clearly be seen in Bartlett’s choice of material for these experiments,
which comes from his wide ethnographic readings. Most famously he used a Native American
folk story from Boas’s Kathlamet Texts (1901) called “War of the Ghosts,” which contains many
narrative disjunctions, seeming lack of logic, strange and vivid imagery, supernatural events,
alongside other unfamiliar elements. French anthropologist Lévy-Bruhl (1926/1985) would
have considered it an excellent example of “primitive mentality”—an idea Bartlett (1923) later
refuted in his book Psychology and Primitive Culture.
In his experiments, Bartlett had people in Cambridge read the story twice at regular speed.
Then at an interval of typically 15 minutes they were asked to write down the story as best
they could remember it. With his “method of repeated reproduction,” he then had participants
write down the story several more times at increasing time intervals, such as one week, after
several months and even several years later. Thus, he could track how the story was transformed
538
Frederic Bartlett
through time. In another experimental variant called the “method of serial reproduction,” he
had the second participant read and then reproduce the first participant’s story. The second
participant’s reproduction was then given to a third and so on, like the party game “telephone”
or “Chinese whispers.” With both methods, Bartlett analyzed the series of reproductions for
what was omitted, added, and transformed from original to first reproduction, and from one
reproduction to the next. His analytic focus was on qualitative changes in single cases through a
series of reproductions. As with the anthropological process of conventionalization, the Native
American story “War of the Ghosts” was progressively transformed into a conventional English
story: many details dropped out, the disjunctive narrative structure was smoothed over, “hunt-
ing seals” became the familiar activity “fishing,” “war cries” changed to simply “noises,” the
unfamiliar proper names (e.g., “Egulac”) were forgotten, and the supernatural elements were
rationalized and then omitted. Changes were particularly “radical” using the method of serial
reproduction, which most closely paralleled the processes by which cultural designs, ideas, and
rumors are transmitted within and between social groups.
In addition to the methods of repeated and serial reproduction, Bartlett also developed the
“method of description” and the “method of picture writing.” With the former, he presented
participants with five faces of military men and after a delay of thirty minutes asked them a series
of questions regarding the faces’ details. In this experiment, he found widespread “transferences”
of details from one face to another and “importations” of features not present in the series. He also
noted how a subject’s “affective attitude” towards a face shaped the way it was later reproduced.
For example, a subject who had the impression of a face being “serious and determined” later
remembered it to have a firmer mouth, more prominent chin and square face when compared
with the original. Finally, when subjects were reminded of a conventional face of a common
British soldier, their reproductions changed towards that image. In the method of picture writ-
ing, subjects had to memorize image–word combinations and then reproduce the images when
they heard the words within a story dictated to them. Again, Bartlett found widespread omission,
blending, and confusion of details.
At the same time that Bartlett was conducting these experiments, he was also working with
traumatized soldiers at the Cambridge hospital. This experience taught him the role of personal
history in setting up a person’s internal conditions for behavior (e.g., attitudes, schema) and the
value of working with single cases, which was how he analyzed participant’s reproductions in
his experiments. Bartlett even compared an experimentalist to a clinician:
If the experimentalist in psychology once recognizes that he remains to a great extent

a clinician, he is forced to realize that the study of any well developed psychological
function is possible only in the light of consideration of its history.
(Bartlett 1932: 15)
His clinical experience also led him into the study of Freud’s work (see also Schwab, Chapter 41,
this volume). Although the majority of terms Bartlett uses to describe transformation in reproduc-
tions come from diffusionist anthropology (e.g., “simplification” and “elaboration”), a number
of terms are also employed from psychoanalysis, such as “condensations,” “transpositions,” and
“rationalizations” (see Wagoner 2017b).
The fate of Bartlett’s experiments on remembering can themselves be analyzed as a serial
reproduction experiment (see Ost and Costall 2002; Wagoner 2015). The first two decades of
replications and extensions of his studies focused on the role of group membership in shaping
what was remembered. For example, Nadel (1937) compared the memory of children in two
Nigerian tribes to illustrate how each group’s distinctive cultural patterns conditioned the result.
539
Brady Wagoner
Moreover, these earlier replications retained Bartlett’s analytic focus on qualitative changes in
single cases. In the late 1950s, both of these features disappeared from replications. These later
experiments turn away from a flexible experimental technique towards an approach that bet-
ter yields itself to inferential statistics. The varieties of qualitative changes in reproductions
become lumped under the term “distortion” or “error” (see, e.g., Gauld and Stephenson 1967).
Furthermore, “distortion” becomes a synonym for “constructive” or “reconstructive” remem-
bering, whereas for Bartlett (1932), it was a positive, future-oriented characteristic of memory
that could also lead to accuracy. How he theorized this and what factors lead to its misunder-
standing will be the focus of the next section.
2. Bartlett’s theory of reconstructive remembering

Bartlett consistently used the active verb “remembering” rather than “memory” to highlight
that it is a socially and materially situated activity that involves a myriad of different processes,
rather than a self-contained mental faculty sufficient unto itself. This perspective comes close to
Wittgensteinian (Hamilton, Chapter 46, this volume) accounts of remembering which underscore
that we are first and foremost speaking of a social practice; only after it is identified as such can we
even begin to search for neural correlates (Harré 2002). Thus, rather than thinking of memories as
stored as individuated traces in the organism (Robins, Chapter 6, this volume), Bartlett conceptu-
alized remembering as occurring at the dynamic intersection between a person and the world to
aid environmental adaptation. According to this perspective, remembering helps a person master
and enjoy a world filled with change. In contrast, recalling all the details of one’s life is highly dys-
functional, as bizarre clinical cases like the mnemonist Sherereshevsky reveal (Luria 1987). Bartlett
(1932: 16) thought our contemporary focus on literal accuracy was the product of an elaborately
guarded civilization, and in most situations was more of a hindrance than a help.
We should not, however, assume by this that our memories are necessarily inaccurate or
distorted. If nothing else, they are accurate enough to maximize our functioning in the world.
Actually, Bartlett (1932) discussed a number of specific cases of exceptional accuracy in remem-
bering (see also Ost and Costall 2002). In his experiments, he pointed out the “retention of
meaningless details,” as well as the persistence of an “attitude,” or accurately remembering the
“gist” of the material. Furthermore, from his field studies in Swaziland he noted that herdsmen
had a “prodigious memory” for characteristics of their cattle and exchange of them, because of
their central place in Swazi society. In short, accuracy is by no means a transparent notion: we
can speak of accuracy on different levels and aspects of our memories. Furthermore, the social
context (such as a court of law or discussion with friends) puts different constraints on how
accuracy is handled. Bartlett emphasized that remembering was socialized by the group and
modified according to the context (more on the social dimensions below).
To theorize this flexible and situation-dependent characteristic of remembering, he appropri-
ated the concept of schema from the neurologist Henry Head, who worked with brain-damaged
patients who had lost the ability to serially connect body movements. Have a patient close their
eyes and move his hand from one position to another and he will register it in the previous
location. Previous theories of body movement had argued that earlier movements were stored
as mental images in the brain but Head showed the image function remained intact in these
patients; thus, another explanation would have to be sought. Another set of patients had lost an
arm or a leg but continued to register it along with their changing body postures—the phantom
limp phenomenon. Adopting a term from Immanuel Kant’s Critique of Pure Reason, Head (1920)
argued that a postural schema was at work. In contrast to the idea that experiences were stored
in some location and later retrieved in roughly the same form as it was put in (i.e., the trace
540
Frederic Bartlett
theory of memory), schema implied an active and continuously revised form of memory that
operated at the transaction between person and world.
Bartlett generalized the concept of schema to argue that perceiving, imagining, and remem-
bering are organized into patterns for which the person is responsible (Wagoner 2013). People
do not passively receive impressions of the world on their sensory organs that are then stored
in memory—an idea that goes back to Plato’s metaphor of memory as a wax tablet in the mind
(Chappell, Chapter 30, this volume). Instead, both perception and memory imply a subject
engaged in the world according to a set of interests and carrying a past history. Schema is the
self-imposed context for action and experience, which guarantees both continuity and flexibility
through time. Memory is not a mental faculty but an activity guided by an adapted scheme or
plan—an idea that can be traced back, through Bartlett’s mentor James Ward, to the philosopher
Lotze’s notion of the “activity of memory” (Northway 1940). Through schema, the past is car-
ried forward en masse and adjusted to the present situation. With it we construct a recognizable
world filled with both permanence and change, where ambiguous objects are shaped into the
whole and those that do not in any way fit are left out. This pattern making or organized quality
of experience was at the same time being explored by the Gestalt psychologists (see, e.g., Koffka
1935). Bartlett’s approach differs in his explicit rejection of the trace theory of memory and his
attention to how social life shapes this schematic organization.
Thus, remembering is not about bringing up old images of past impressions; instead memory
images come into being through schematic organization. One’s memory for a visit to a restau-
rant will take on characteristics of other occasions, in which the most recent visits will exert the
strongest influence on the whole. Bartlett does give a particular function to images of “picking
out” aspects of schema, so that particular episodes from the past can be remembered and inserted
into the present context of action. However, as his experiments aptly showed, images are moving
and changing with our interests; thus, he is not here bringing in the trace theory of memory. In
addition to “schema” and “images,” Bartlett also placed importance on the concept of attitude—
or “summary feelings” of some material—which explicitly puts feeling and emotion at the center
of his theory. He says an act of remembering begins by setting up an attitude, then proceeds to
reconstruct the material so as to justify the attitude and finally ends with an attitude in which no
more questions are asked. Bartlett succinctly summarized his theory as follows:
Remembering is not the re-excitation of innumerable fixed, lifeless and fragmentary

traces. It is an imaginative reconstruction or construction, built out of the relation
of our attitude towards a whole active mass of organised past reactions or experience
[i.e., schema], and to a little outstanding detail which commonly appears in image
or in language form.
(Bartlett 1932: 213)
It is important to point out here that Bartlett did not think that people were simply slaves to
schema, but could actively manipulate them. In this process a person “turns around upon sche-
mata and constructs them afresh” (1932: 206). In other words, we are also capable of reflecting
on and controlling remembering—what has been more recently discussed as “meta-memory” or
more generally “meta-cognition” (Nelson 1996). This is where human agency enters into the
process of remembering. For Bartlett, “turning around upon schema” points to a more radical
form of reconstruction than the simple activity of schema, in that multiple schemata are brought
into play checking and supporting each other in the act of remembering. In characteristically
British fashion, Bartlett gives the example of a journalist’s account of a cricket match: “To describe
the batting of one man he finds it necessary to refer to a sonata of Beethoven; the bowling
541
Brady Wagoner
of another reminds him of a piece of beautifully wrought rhythmic prose written by Cardinal
Newman” (1935: 224). More recently, Wagoner and Gillespie (2014) have analyzed this process
in a conversational remembering experiment, extending Bartlett’s (1932) method of repeated
reproduction. In contrast to the practice of only looking at memory outcomes, this study shows
how memories emerge through a process of questioning, suggesting, and evaluating.
Schema theory changed when it became a central concept during the cognitive revolution in
the 1960s and 1970s. Moving away from Bartlett’s bio-functional approach, the primary meta-
phor of mind was a computer that processes information rather than an organism that makes
meaning to act in an environment. At this time, schema becomes a relatively static knowledge
structure with nodes that accept items of a certain kind and fill in default values when informa-
tion is missing. For example, the schema for a graduate student’s room includes desk, calendar,
pencils, books; we are more likely to remember a room that is consistent with this and may fill
in items that fit the schema even if they are not present (Brewer and Treyens 1981). Because
schema is assumed to be relatively static in this account, no one bothered to do repeated repro-
duction experiments. Moreover, Bartlett’s idea that we can “turn around upon” and reflect on
our schema is forgotten. Finally, the social dimensions of schema were not elaborated, and often
considered negligible. In more recent developments of schema theory, a social and cultural
focus has become central (e.g., Cole 1996; Wertsch 2002; McVee et al. 2005). The next section
will highlight how Bartlett theorized the social dimensions of remembering.
3. The social psychology of remembering

Bartlett’s (1932) famous book was explicitly “a study in experimental and social psychology,”
yet theorists have typically neglected the latter “social” contribution of the book. This was made
possible by the structure of Remembering itself: his experiments and theory of reconstructive
remembering were in Part I of the book, and his social theorizing was only explicitly addressed
in Part II, thus making it easier to ignore. Bartlett’s experiments clearly illustrated how internal-
ized group conventions shaped what was remembered (e.g., in the story “War of the Ghosts,”
“hunting seals” is not a familiar English activity and hence it tended to be changed to “fishing”),
but these experiments largely left aside the role of the concrete social process happening between
people, except in how the context of an experiment created an attitude more focused on accu-
racy. Social groups put definite constraints on how and what we remember “by providing that
setting of interest, excitement and emotion which follows the development of specific images
and socially by providing a persistent framework of restrictions and customs which acts as the
schematic basis for constructive memory” (1932: 255).
At the same time Bartlett was developing his theory, Halbwachs published his classic work
Les cadres sociaux de la mémoire (translated as On Collective Memory) arguing, “It is in society that
people normally acquire their memories. It is also in society that they recall, recognize, and
localize their memories” (1925/1992: 38). According to Halbwachs, people remember through
“social frameworks,” which are first of all the people one interacts with and second a symbolic
framework of condensed images and structure to give them order and meaning (see also Nikulin,
Chapter 44, this volume). Even in an isolated experiment, people use language to remember,
which places them in social frameworks. The idea of social frameworks comes very close to
Bartlett’s description of the role of conventions in remembering. However, Bartlett thought
Halbwachs went too far in claiming that the group itself had memory: there is “memory in the
group, and not memory of the group” (1932: 296; original emphasis). This is not to say that we
cannot speak of customs, traditions, and institutes as literally properties of groups. Bartlett is clear
“Social grouping produces new properties both of behavior and of experience” (1932: 298).
542
Frederic Bartlett
But remembering is a process irreducible to purely social factors; it is simultaneously personal

and collective. Whereas Halbwachs focused much more on how groups organize memory,
Bartlett’s interest is in how social forces condition a given individual’s act of remembering.
Bartlett’s social psychology of remembering is analytically divided into two areas, which he
called the matter and manner of recall. The former concerns how social factors condition what
we remember and the latter how we remember it (what might be called “cognitive style” today).
These are mainly illustrated with examples from his 1929 trip to Swaziland. In regards to the
matter of recall, Bartlett describes how a group of Swazis that had visited England remembered
police officers as being particularly friendly. This was because the gesture the officers used to
stop traffic was similar to the one they used to greet their fellows, demonstrating how a social
framework determines what is selectively attended to and remembered. In terms of the amount
of material recalled, Bartlett found no major difference between Swazis and English memory for
a short message. However, there was one topic in which the Swazis remembered with much
greater detail and accuracy: transactions concerning cattle, which was an integral part of their
social framework. Bartlett tested a Swazi herdsman’s recall for his employer’s record of cattle
transactions from the previous year. The Swazi easily did so, making only two “trifling errors.”
With regards to the manner of recall, Bartlett argued that certain forms of social organization
and customs, where no dominant interest lies, led natives into rote recall. In other words, they
recalled all information in a chronological sequence, rather than picking out and reconfiguring
the relevant pieces of information. It is a bit like having to recite the whole alphabet to remem-
ber the placement of a particular letter. Bartlett can easily be criticized here for slipping into a
colonial mindset in his characterization of Swazi memory as having few interests and tending
towards rote recall. Cole and Gray (1972) have shown in their cross-cultural study of Kpelle
rice farmers that people living in traditional societies do not tend to remember by rote, although
they are also less likely to group items into semantic categories in remembering them.
Bartlett (1932), however, also gives a more plausible explanation for the relationship between
social organization and style of recall. He notes how, when a story is remembered in the presence
of others, certain characteristics will spring forth, such as the comic, the pathetic, or the dramatic,
depending on the interests of the audience. This is why when a story that was orally transmitted
is written down, it often loses its spirit, even if the content remains effectively the same. Inspired
by cybernetics (Norbert Weiner was a friend of Bartlett at Cambridge), the narrator is seen to be
in a feedback loop with the audience such that the story is adapted to conform to the audience’s
framework of understanding. In his earlier book Psychology and Primitive Culture, Bartlett (1923)
had argued that folk stories change as they enter new social groups. Different characters, events,
and morals take prominence as the stories are told so as to connect up with the new social frame-
work. He also returns to the theory put forward in Psychology and Primitive Culture (Bartlett 1923)
of how different forms of social relationship (i.e., dominance, submissiveness, and friendship)
affect the transmission of culture within and between groups in Remembering:
Change the audience to an alien group, and the manner of recall again alters. Here
the most important things to consider are the social position of the narrator in his own
group, and his relation to the group from which his audience is drawn. If the latter
group are submissive, inferior, he is confident, and his exaggerations are markedly
along the lines of the preferred tendencies of his own group. If the alien audience is
superior, masterly, dominating, they may force the narrator into the irrelevant, reca-
pitulatory method until, or unless he, wittingly or unwittingly, appreciates their own
preferred bias.
(Bartlett 1923: 266)
543
Brady Wagoner
4. Conclusion: towards an integrative theory of

constructive remembering
Constructiveness in Bartlett’s theory is not simply the reassembly of pieces according to a pre-set
pattern. Instead, it involves an innovative adaptation to the novel circumstances of the present in
anticipation of the future (see also Perrin and Michaelian, Chapter 18, this volume). This living
forward requires a flexible use of the past rather than literal reproduction of it. Remembering
emerges within the stream of living (i.e., present schema) in order to bring about a novel rela-
tionship to the environment and increase one’s avenues for action within it. The person “turns
around upon” their attuned flow of activity with the environment to discover new possibilities
for action therein. Social factors are key to this process by providing both the condition and
means for remembering to unfold. Bartlett’s experiments elegantly showed the prominence
of group conventions in memory reconstruction. His theorizing of the role of others in recall
also helps us to identify the more direct social forces on this process. But Bartlett was clear that
remembering cannot be reduced to social factors; it is simultaneously personal. An integra-
tive constructivist theory of remembering will have to show how the social and personal are
actively coordinated with one another in any act of remembering (for contemporary examples
see Wagoner 2017c). Theorists will themselves have to be constructive to adequately develop
this integrative approach.
Further reading
Wagoner, B. (2017). The Constructive Mind: Bartlett’s Psychology in Reconstruction. Cambridge: Cambridge
University Press.
References
Bartlett, F.C. (1916a). “An experimental study of some problems of perceiving and imagining.” British
Journal of Psychology, 8: 222–66.
Bartlett, F.C. (1916b). Transformations arising from repeated representation: A contribution towards an
experimental study of the process of conventionalization. Fellowship Dissertation, St. John’s College,
Cambridge.
Bartlett, F.C. (1923). Psychology and Primitive Culture. Cambridge: Cambridge University Press.
Bartlett, F.C. (1932). Remembering: A Study in Experimental and Social Psychology. Cambridge: Cambridge
University Press.
Bartlett, F.C. (1935). “Remembering.” Scientia, 57: 221–6.
Bartlett, F.C. (1936). “Frederic Charles Bartlett [autobiography].” In C. Murchison (ed.), A History of
Psychology in Autobiography, Vol. III (pp. 39–52). Worcester, MA: Clark University Press.
Boas, F. (1901). Kathlamet Texts. Washington, DC: G.P.O.
Brentano, F. (1874/1973). Psychology as an Empirical Science. London: Routledge.
Brewer, W.F., and Treyens, J.C. (1981) “Role of schemata in memory for places.” Cognitive Psychology,
13: 207–30.
Cole, M. (1996). Cultural Psychology: A Once and Future Discipline. Cambridge, MA: Harvard University
Press.
Cole, M., and Gray, G. (1972). “Culture and Memory.” American Anthropologist, 74(5): 1066–84.
Ebbinghaus, H. (1885/1913). Memory: A Contribution to Experimental Psychology. New York: Teachers
College, Columbia University.
Gauld, A., and Stephenson, G. M. (1967). “Some experiments related to Bartlett’s theory of remembering.”
British Journal of Psychology, 58: 39–49.
Halbwachs, M. (1925/1992). On Collective Memory. Chicago, IL: University of Chicago Press.
Harré, R. (2002). Cognitive Science: A Philosophical Introduction. London: Sage.
Head, H. (1920). Studies in Neurology. London: Hodder & Stoughton
Koffka, K. (1935). Principles of Gestalt Psychology. London: Routledge.
544
Frederic Bartlett
Lévy-Bruhl, L. (1926/1985). How Natives Think. Princeton, NJ: Princeton University Press.
Luria, A.R. (1987). The Mind of a Mnemonist: A Little Book about a Vast Memory. Cambridge, MA: Harvard
University Press.
McVee, M.B., Dunsmore, K., and Gavelek, J.R. (2005). “Schema theory revised.” Review of Educational
Research, 75: 531–66.
Nadel, S.F. (1937). “Experiments on culture psychology.” Africa, 10: 421–35.
Nelson, T.O. (1996). “Consciousness and metacognition.” American Psychologist, 51: 102–16.
Northway, M.L. (1940). “The concept of ‘schema’: Part I.” British Journal of Psychology, 30: 316–25.
Ost, J., and Costall, A. (2002). “Misremembering Bartlett: A study in serial reproduction.” British Journal
of Psychology, 93: 243–55.
Rivers, W.H.R (1912). “Conventionalism in primitive art.” Reports of British Association for the Advancement
of Science (Sección H), 599.
Wagoner, B. (2013). “Bartlett’s concept of schema in reconstruction.” Theory & Psychology, 23(5): 553–75.
Wagoner, B. (2015). “Qualitative experiments in psychology: The case of Frederic Bartlett’s methodology
[82 paragraphs].” Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 16(3): Art. 23,
http://nbn-resolving.de/urn:nbn:de:0114-fqs1503230.
Wagoner, B. (2017a). The Constructive Mind: Frederic Bartlett’s Psychology in Reconstruction. Cambridge:
Wagoner, B. (2017b). “What makes memory construction? A study in the serial reproduction of Bartlett’s
experiments.” Culture & Psychology.
Wagoner, B. (2017c). Handbook of Culture and Memory. Oxford: Oxford University Press.
Wagoner, B., and Gillespie, A. (2014). “Sociocultural mediators of remembering: An extension of Bartlett’s
method of repeated reproduction.” British Journal of Social Psychology, 53: 622–39.
Wertsch, J.V. (2002). Voices of Collective Remembering. Cambridge: Cambridge University Press.
545
46
LUDWIG WITTGENSTEIN
Andy Hamilton
Wittgenstein did not work very extensively on memory, but his views on the topic are
reasonably defined and important. Moyal-Sharrock’s summary is accurate: ‘[he] discredited the
storage and imprint models . . . dissolved the conceptual link between memory and mental
images [and] made room for a family resemblance concept . . . where remembering can also
amount to doing or saying something’ (Moyal-Sharrock 2009: ‘Abstract‘). This quotation also
summarises the issues addressed in the present chapter. One can treat Wittgenstein’s work
historically as part of ‘Wittgenstein studies’; or ahistorically, relating it to the increasingly
technical sub-disciplines of epistemology and philosophy of mind. This chapter tries to bring
together historical and ahistorical approaches.
Wittgenstein’s discussions of memory belong to his middle and later periods, where philoso-
phy of mind and epistemology became a more direct concern. The later philosophy in particular
is therapeutic in conception, holding that philosophical problems should be dissolved rather than
solved. Therapy can be interpreted in a quietist or combative way, corresponding to the faultline
between Pyrrhonian readings, which see Wittgenstein as sceptical about philosophy itself, and non-
Pyrrhonian readings that regard him as criticising certain traditional approaches to philosophy (see
Fogelin 1976, and 1994: 205; Hamilton 2014: Ch. 12). This chapter adopts the latter approach.
Wittgensteinian therapy aims to show how philosophical disputes rest on what he termed
‘grammatical illusion’. His anti-metaphysical predecessors, Kant included, regarded their oppo-
nents’ theories as false or misguided. Wittgenstein was the first clearly to formulate the view that
they are nonsensical – or given that nonsense takes many forms, that metaphysical theories are
linguistic nonsense. He is the major philosopher who most keenly interrogates the meaningfulness
of philosophical questions and claims. His philosophical stance, with its enduring antipathy to
scientism, rests on a clear distinction between the logical or conceptual on the one hand, and
the empirical on the other.
Scientism regards philosophy as part of science, thus conflating the conceptual and the empiri-
cal. It holds that the natural sciences offer the only guide to what exists or is known; nature is
what these sciences say it is. As Bertrand Russell writes: ‘what science cannot discover, mankind
cannot know’ (Russell 1935: 243). Humanism, in contrast, holds that the humanities constitute
a relatively distinct group of disciplines concerned with the human rather than physical world,
whose forms of explanation are irreducible to those of the natural sciences. The status of the
humanities was not a particular concern of Wittgenstein’s, but he rejected the scientistic view
546
Ludwig Wittgenstein
that philosophy itself is not a distinct discipline or activity – for him, scientism had a baneful
effect in encouraging metaphysics.
Wittgenstein’s critique of a scientistic conception of psychology leaves open the possibility
of a humanistic psychology as advocated by William James – though the prospect is now, sadly,
more remote than ever. (A humanistic interpretation of James is defended by Reed 1997: xv;
see Hacking 1995.) Wittgenstein writes that
The confusion and barrenness of psychology is not to be explained by calling it a ‘young

science’; its state is not comparable with that of physics, for instance, in its beginnings.
For in psychology there are experimental methods and conceptual confusion.
(Wittgenstein PI II: 232)
In particular, Wittgenstein criticises how psychologists have investigated memory. It is essential

to stress that his stance is anti-scientistic, not anti-scientific – it is opposed to philosophical over-
valuations, and false metaphysical interpretations, of science. Its radical, unsettling critique of
scientific psychology is one that many will reject.
One final introductory comment: this chapter refers to arguments that Wittgenstein presents
against certain conceptions of memory – yet some writers deny that he offers ‘arguments’ in any
traditional sense. One could describe them as ‘considerations’ instead of ‘arguments’; but the
Wittgenstein literature has surprisingly little to say on what is meant by ‘argument’ and ‘refuta-
tion’. Read, for instance, contrasts therapy with ‘argument as traditionally conceived’, leaving
the tradition unspecified (Read, accessed 2014). Those who criticise a writer’s work for lack of
argumentation often have in mind explicit premises and conclusions, or a style from which these
can readily be deduced – just as some music is criticised as unrhythmic, because its rhythm is
subtler or less obvious than the crudest rock drumming. What is lacking in these cases is explicit
or obvious argument or rhythm, not argument or rhythm as such.
1. Wittgenstein’s rejection of an image theory

Plato introduced the metaphor of memory as an imprinting on a block of wax. Socrates remarks
that ‘whenever we wish to remember something we see or hear or conceive . . . we hold this
wax under the perceptions or ideas and imprint them on it as we might stamp the impression of
a seal ring’ (Theaetetus 191 c–d). From Aristotle onwards, the metaphor began to be taken more
literally (Hacker 1996). Plato’s metaphor is the basis both for what I term the image theory, and
the trace theory. The image theory – which we examine first – can be described as indirect realism
about memory, and Wittgenstein offers a sustained critique. It is most commonly an empiricist
view, advocated from Locke onwards, based on the assumption that all knowledge is derived
from (present) experience. The theory treats remembering as essentially a present experience
from which one infers about past events.
Its proponents often claim that putative memories are introspectively distinguishable. Thus
Pollock argues that ‘memory must provide us with beliefs about what we “seem to remember”
and then we infer the truth of what are ordinarily regarded as memory beliefs from these appar-
ent memories.’ He maintains that, by introspection, we can distinguish beliefs we have on the
basis of memory from those we have on the basis of perception, ‘or for no reason at all’. Or
rather, we can distinguish putative memory-beliefs from other kinds: ‘I do not mean that we
can tell introspectively whether we are correctly remembering what we take ourselves to be
remembering’ (Pollock 1986: 51–2).1 This qualification avoids the implausible suggestion that
memory yields infallible knowledge of the past. Image theorists should say rather that although
547
Andy Hamilton
I cannot be sure whether the event happened or not, I can at least be sure that if it happened,
I know that it did on the basis of memory, rather than because I received indications after the
event that it happened.
The image theory that Wittgenstein targets is espoused by empiricist thinkers William
James, Bertrand Russell – two members of his select philosophical canon – and Gestalt psy-
chologist Wolfgang Köhler. Wittgenstein insists that images may accompany remembering,
but cannot constitute it: ‘When we narrate a set of events from memory we do sometimes see
memory pictures in our mind; but commonly they are only scattered through the memory
like illustrations in a story book’ (PG: 181).2 Wittgenstein insists, against empiricists, that
‘When I say: “He was here half an hour ago” – that is, remembering it – this is not the
description of a present experience. Memory-experiences are accompaniments of remembering’
(PI Part II: 231). He denied that introspectible images, or memory-experience in general, pro-
vide an inferential basis for memory-claims:
My memory image is not evidence of that past situation; as a photograph would be,
which, having been taken then, now bears witness . . . that this is how it was then.
The memory image and the memory words are on the same level.
(RPP i. para. 1131)
What it means to say that image and words are ‘on the same level’ is something that will
become clear shortly. Note that his objection is not that the image theory fails to distinguish
memories from imagination and so cannot guarantee memory claims as certain – that it cannot
refute memory-scepticism, such as Russell’s suggestion that the world was created five minutes ago
including all our ‘memories’. Wittgenstein’s view is rather that the image theory offers no basis
at all for knowing when to infer and when not to infer from an image to a memory claim. A
more sustained argument is found in PI II:
Remembering has no experiential content. – Surely this can be seen by introspec-

tion? Doesn’t it shew precisely that there is nothing there, when I look about for a
content? – But it could only shew this in this case or that.
And even so it cannot shew me what the word ‘to remember’ means, and hence
where to look for a content!
. . . Would this situation be conceivable: someone remembers for the first time in his
life and says ‘Yes, now I know what “remembering” is, what it feels like to remember’. –
How does he know that this feeling is ‘remembering’?
Compare: ‘Yes, now I know what “tingling” is’. (He has perhaps had an electric
shock for the first time.) – Does he know that it is memory because it is caused by
something past? And how does he know what the past is? Man learns the concept of
the past by remembering.
(PI II – note that Part II of PI was not a draft prepared for
publication by Wittgenstein himself)
RPP 837 amplifies this remark:
Tingling teaches him about the external world. Does remembering teach us in the
same way that a certain event took place in the past? – Then we would have to connect
it up with past events . . . Whereas it is really the criterion for the past.
(RPP 837)
548
Ludwig Wittgenstein
These remarks require careful elucidation. Wittgenstein is raising two related issues:
(1) The experiential content of memory.

(2) The analysis of memory as ‘image or feeling caused by past event’.
He regards them as related because the concept of remembering cannot arise from experience of
the past, as empiricists argue – remembering is itself a criterion of the past. However, Meixner
is wrong to assume that Wittgenstein endorses the radical opening claim, ‘Remembering has no
experience-content.’ Meixner is concerned with Wittgenstein’s opposition to phenomenology,
commenting that he briefly entertains the idea that one can know by introspection that ‘remem-
bering has no experience-content’, and that he dismisses the idea ‘not because introspection in
fact shows (as every Phenomenologist knows) the exact opposite . . . but because he considers it
to be . . . purely a matter of the linguistic meaning of the word “to remember” that “remember-
ing has no experience-content”’. For Meixner, this claim violates the phenomenology both of
linguistic meaning, and of conscious experience (Meixner 2014; URL accessed 2016).
In response, Wittgenstein would say both the phenomenology of linguistic meaning, and that
of memory, that experiences may accompany, but cannot constitute it. It is true that he rejects
the phenomenological view – if that is what it is – that what ‘remember’ means is discovered by
introspection. He also rejects the empiricist view that memory-states can be distinguished from
imagination by an accompanying feeling – that one can know, as he puts it, that ‘This feeling is
“remembering”’. But he does not insist that ‘remembering has no experiential content’, since he
allows that memory is accompanied by experiences.
Even more elusive is issue (2), and the intriguing claim that ‘Man learns the concept of the
past by remembering.’ This claim recurs in remarks collected in Zettel:
Remembering: a seeing into the past. Dreaming might be called that, when it presents
the past to us. But not remembering; for, even if it shewed scenes with hallucinatory
clarity, still it takes remembering to tell us that this is past.
(Z 662) [also RPP ii. 592]
Here, Wittgenstein is considering and rejecting the idea that remembering is a seeing into the
past. Interestingly, he allows that dreaming, in contrast, sometimes ‘presents the past to us’ –
perhaps he is referring to Freud’s ‘day residues’, events of the previous day that are represented
in a dream. ‘Presenting’ implies a passive subject; the dreamer need make no judgement about
the events, and need not recognise them for what they are. Remembering, in contrast, involves
recognising the scenes represented – it involves judgement.
Wittgenstein continues:
Z 663. But if memory shews us the past, how does it shew us that it is the past?
It does not shew us the past. Any more than our senses shew us the present . . .
665. I give myself an exhibition of something only in the same way as I give one to
other people.
666. I can display my good memory to someone else and also to myself. I can subject
myself to an examination. (Vocabulary, dates).
667. But how do I give myself an exhibition of remembering? Well, I ask myself
‘How did I spend this morning?’ and give myself an answer. – But what have I really
exhibited to myself? Remembering? That is, what it’s like to remember something? –
Should I have exhibited remembering to someone else by doing that?
(Z 663, 665–7; original emphasis)
549
Andy Hamilton
The key argument here is ‘still it takes remembering to tell us that this is past’ (Z 662), which
reiterates PI quoted earlier: ‘Does he know that it is memory because it is caused by something
past? And how does he know what the past is? Man learns the concept of the past by remem-
bering’ (PI II: 231).
Wittgenstein’s claim, against both image and trace theories, is that no non-circular analysis
of memory is possible; one must have a prior or – more plausibly – simultaneously acquired
understanding of ‘the past’. He is proposing what I have termed a conceptual holism. This is not
holism in Quine’s sense; conceptual holism is distinct from the more widely accepted holism
of belief. Something like the latter is expressed in Wittgenstein’s pregnant thought, ‘What we
believe is not a single proposition, it is a whole system of propositions. (Light dawns gradually
over the whole.)’3 This insight applies equally to the acquisition of a conceptual system. For
the language-learner, light dawns gradually over the system of beliefs or concepts that they are
acquiring. But the web of belief or concepts is a holism of attribution, not one of meaning; like
Davidson’s holism of desire and belief, it concerns the attribution of mental states or capacities
to subjects.
My present concern is with the very different notion of conceptual interdependence – the
way that concepts such as belief and assertion, or self-consciousness and self-reference, or
art and the aesthetic, are inter-defined. To propose a conceptual holism is to see concepts as
part of a practice or language-game. In defending a holism of belief and desire, Davidson is
not saying – implausibly – that one cannot define the concept of desire independently from
that of belief and vice versa. However, there is this connection between conceptual holism
and the holism of belief: conceptual holisms show how concepts form a network, which the
subject acquires gradually, as light dawns over the whole.
A conceptual holism involves an equivalence or interdependence between the concepts
concerned, therefore; neither is more basic than the other. Thus, a definition or understand-
ing of concept X makes essential reference to that of concept Y, and vice versa. There is a
relation of mutual presupposition; one can neither acquire, nor manifest understanding of,
one concept without the other. In contrast are concepts where presupposition is one-way
only. A photograph is a kind of picture, and ‘photograph’ cannot be understood without an
understanding of ‘picture’; but ‘picture’ is more basic, and can be understood without under-
standing ‘photograph’.
My suggestion is that a conceptual holist picture underlies Wittgenstein’s remarks on mem-
ory and the past. These are developed in PR into an intriguing view of memory as both source
of the concept of the past, and ‘picture of a physical event’: ‘We can look at recognition, like
memory, in two different ways: as a source of the concepts of the past and of identity, or as a
way of checking what happened in the past, and on identity’ (PR 19). Here Wittgenstein seems
to anticipate On Certainty: ‘the same proposition may get treated at one time as something to
test by experience, at another as a rule of testing’ (OC 98): ‘“I have hands” might be treated as
the rule, and used to test whether I can see; or one could use “I can see” as the rule, in testing
whether I have hands)’ (OC 125). These remarks express Wittgenstein’s well-known distinction
between criteria and symptoms. Many empirical generalisations, both mathematical and scien-
tific, have turned into rules of testing; at one time, the boiling point of water was a discovery,
but now has a rule-like status.4
Wittgenstein’s critique of the image theory – or at least, a Wittgensteinian critique – may
imply that the theory neglects much of what is ordinarily regarded as ‘remembering’, viz. what
is often called habitual or factual memory of information, which need not involve recalling the cir-
cumstances in which one acquired the information (Hacker 1996: 163–4). Perhaps the image
theory does neglect this kind of memory, and maybe rightly so, for the standard philosophical
550
Ludwig Wittgenstein
distinction between personal or autobiographical memory – witnessing and recalling – and factual
memory should be respected. (The distinction is recognised, perhaps less clearly, in psychology.)
For instance, I can answer the question ‘What did you have for breakfast last Friday?’ by infer-
ring from factual memory that I always have toast and marmalade for breakfast, plus perhaps
my inability to recollect anything out of the ordinary about that Friday. This is at best partially
personal remembering.
2. Direct knowledge: a Wittgensteinian alternative

to the image theory
One can interpret Wittgenstein’s compressed comment that ‘The memory image and the
memory words are on the same level’ (RPP i. para 1131) by appeal to Mounce’s claim that ‘the
image is not the ground from which I derive my knowledge but the form, on this occasion,
which my knowledge takes’ (Mounce 1994). That is, I am able to picture the scene because I
remember it, not vice versa; a verbal description, or a drawing, could equally be the form that
my knowledge takes. The claim that the image correctly represents the event is itself a memory
claim. One could say: to picture an event or experience is to remember it. Mounce’s view has
affinities with the direct knowledge accounts of Reid and Stout, according to which memory gives
us immediate knowledge of the past, not inferred from some present experience or image of
remembering. On this view, human beings have the ability spontaneously to make reliable past-
tense reports; personal memory involves spontaneous judgements about the past. This view parallels
direct realism in perception, in eliminating intermediate representations or experiences.
On a direct knowledge account, as opposed to the indirect knowledge of the image
theory, memory yields unmediated knowledge of the past. On this view, the immediate
objects of memory are past things and not present ideas. These objects are intentional, that
is, objects of thought or judgement, not awareness. Just as fictional entities such as unicorns
can be objects of thought, so can past objects. Thus remembering makes essential reference
to judgement. To take Reid’s example, to say that the smell of the tuberose is the ‘immediate
object of my memory’ is to say that my memory-belief is not inferred, nor otherwise justified,
from a present mental image, representation, or introspectible state. Personal remembering
and the earlier perception share the same object – that is to say, memory has past things for
its objects. (There are parallels here with Marie McGinn’s recent discussion of non-inferential
knowledge – McGinn 2012.)
A direct knowledge account exploits analogies between memory and perception; it treats
personal memory-judgements as past-tense claims which are spontaneous and reliable, in the
same way that perceptual judgements are. If I am asked to describe the view from the window
of the room I am in, I will mostly do so spontaneously, though partly via inference. (As in ‘That
square over there must be Red Lion Square, because I remember passing a street-sign with that
name on my way here’, and so on.) In normal circumstances it would be absurd to question
whether this really is what I am seeing – my judgements in this respect are authoritative. The
reliability and spontaneity of perception is shared by memory – hence to amplify my earlier
claim, there is a human capacity spontaneously and reliably to report both present and past
events. One can trust such statements made by others provided one recognises them as sincere
and spontaneous products of their witnessing. A spontaneous past-tense claim is one not made
simply as a result of receiving indications after the event, whether statements or evidence, that
the thing happened, and not based on an inference. The connection with perception is not just
an analogy, however. Though memory allows only limited re-focusing after the event, personal
remembering shares the same object as the earlier perception. (It may be better to say that we
551
Andy Hamilton
trust memory, without implying that we believe that it is reliable. Note also that we remember
thoughts, so the parallel with perception is incomplete.)
Wittgenstein would not have been comfortable with terms like ‘direct knowledge’ – his
Ordinary Language successor J.L. Austin questioned appeals to directness and immediacy in his
book Sense and Sensibilia (1962). And in contrast with traditional direct knowledge accounts by
Reid (2002) and Stout (1930), which downgrade memory-images and memory-experience,
Wittgenstein and Mounce offer a third way, distinct from the image theory. Image theorists
say that the memory-judgement is justified by memory-images – I can remember the event
because I picture it. Traditional direct knowledge theorists say that the image is justified by
the memory-judgement – I can picture the scene because I remember it – and that the image
is a correct representation of the event is itself a memory claim. The third way says that the
memory-image is an expression of the subject’s knowledge of the events that they witnessed or
experienced. (This position has affinities with the no-priority account of the Euthyphro dilemma,
discussed in Hamilton 2013: Ch. 3.) Note that this is a possible interpretation or development
of Wittgenstein’s view; to reiterate, quietist interpreters argue that he offers phenomenologi-
cal considerations and grammatical remarks, but not a philosophical ‘account’.
3. Wittgenstein’s rejection of causal and trace theories

Its early proponents did not regard the image theory as a causal theory – there was then no
causal theory. But causal versions of it are possible. Martin and Deutscher (1966) seem to have
originated the causal theory of memory, arguing that an experience is a memory-experience when
it is caused appropriately (internally). That is, there is a causal chain internal to the subject, link-
ing witnessing to present recollection, and standardly involving a memory-trace. Both image and
trace are ‘representations’ of the remembered scene or experience, and thus share its structure.
The term ‘impression’ might be ambiguous, as when Locke refers to the ‘storehouse of our
ideas’ – he seems to think that impressions can also be traces. But ‘impression’ normally means
‘image’, something of which one is conscious, while a trace is something in the brain – not an
experience, but a neuro-physiological entity.
Wittgenstein famously rejects the idea of a purely inner process as redundant:
305. ‘But you surely cannot deny that . . . in remembering, an inner process takes
place’ . . . What we deny is that the picture of the inner process gives us the correct
idea of the use of the word ‘to remember’ . . . [this picture] stands in the way of our
seeing the use of the word as it is.
306. Why should I deny that there is a mental process? But ‘There has just taken
place in me the mental process of remembering . . .’ means nothing more than: ‘I have
just remembered’.
(PI 305–6)
That the alleged process is epistemically as well as ontologically inner, is shown by Wittgenstein’s
well-known riposte: ‘“Are you not really a behaviourist in disguise? Aren’t you at bottom really
saying that everything except human behaviour is a fiction?” – If I do speak of a fiction, then it
is of a grammatical fiction’ (PI 307). That is, Wittgenstein is not saying that mental states – pains,
feelings, emotions, beliefs and so on – are fictional. Rather, he holds, philosophers misconstrue
the grammar of first-person psychological statements in the way discussed in Section 1 of this
chapter, leading them to make fictional or nonsensical claims about private objects.
552
Ludwig Wittgenstein
Elsewhere, Wittgenstein targets traces as neural entities – physically rather than epistemically
inner. He rejects not their possibility, but their necessity, denying that these causal intermediaries
between past experience and present recollection are the ‘essence of memory’:
I saw this man years ago: now I have seen him again, I recognise him, I remember
his name. And why does there have to be a cause of this remembering in my nerv-
ous system . . . stored-up there in any form? Why must a trace have been left behind?
Why should there not be a psychological regularity to which no physiological regu-
larity corresponds? If this upsets our concepts of causality then it is high time they
were upset.
(RPP 905; original emphasis. See also Z 610, where it is followed by
criticism of the ‘prejudice in favour of psycho-physical parallelism’.)
These remarks follow on from
No supposition seems to me more natural than that there is no process in the brain
correlated with associating or thinking; so that it would be impossible to read off
thought-processes from brain-processes. I mean this: if I talk or write there is, I
assume, a system of impulses going out from my brain and correlated with my spoken
or written thoughts. But why should the system continue further in the direction of the
centre? Why should this order not proceed, so to speak, out of chaos?
(RPP 903; also in Z 608)
A contemporary version of this standpoint is found in Epstein (accessed 2016). Since this fun-
damental critique undermines the rationale of much lavishly funded neural research worldwide,
it is unlikely to be heeded.
Wittgenstein suggests that, fatally, the trace theory involves an infinite regress of explanation:
An event leaves a trace in the memory: one sometimes imagines this as if it consisted
in [an impression] in the nervous system. As if one could say: even the nerves have
a memory. But then when someone remembered an event, he would have to infer
it from this impression, this trace. Whatever the event does leave behind, it isn’t the
memory.
(RPP 220)
If my possession of certain information requires explanation in terms of a neural inscription,

then how the inscription carries information itself requires explanation, and so on.
A related argument arises from the fact that the concept of a memory-trace is meant
to capture the idea of memory as retained or stored information. But it may be that on a
Wittgensteinian account, ‘retains information’ and ‘remembers’ are interdefined – that is, they
form a conceptual holism as defined earlier – so that the former is not more basic than the
latter. Thus the notion of a trace is redundant. (Note, however, that remembering what one
said or thought is quite different from remembering a scene perceptually; one cannot assume
that experience is given to us already interpreted, as a Kantian account would suggest.) This is
what Schulte argues, when he considers someone making a mistake in recalling: ‘Where does
my mistake reside? Did I make the error in recording fact? Has the storing caused a certain
defect? Or did I make the mistake in encoding the record? We shall never find out’ (Schulte
1996: 116).
553
Andy Hamilton
Those who sympathise with Wittgenstein have therefore rejected a causal analysis. Thus
Squires argues that in describing something as having retained a quality, there is no forced
reference to causal connections:
even if we accepted the metaphor of the memory as a storehouse containing abilities,

it would not follow that in order to retain an ability there needs to be a successive set
of states each producing the next. A sack of grain in the granary is not operative in
producing the same bag of grain in the granary for the rest of the year.
(1969: 178, 196)
Anscombe (1974, 1981) argues that proponents of a causal theory fail to grasp that memory is
an ‘original phenomenon of causality’, involved in how we come to understand what a cause
is. Thus she proposes – implausibly, I would argue – what is effectively a conceptual holism
between memory and causality. Like Wittgenstein, Anscombe rejects appeal to a present experi-
ence as justification of memory-reports, and denies the necessity of memory-traces.
Malcolm, another Wittgenstein student, asks: ‘Why are we not willing to allow that this
ability [to remember] is a natural human power?’ (Malcolm 1977: 89). This is the humanistic
thought also underlying Anscombe’s suggestion that ‘The past tense is heard by the learner of
the language, and then used by him to make spontaneous reports’, where ‘spontaneous’ means
that the person does not make the report as a result of receiving indications after the event,
whether evidence or testimony, that it happened. The fundamental sense of ‘retained infor-
mation’ here is personal, not neural – retention by a person, not by a brain (Hamilton 2013:
Ch. 2.6). The scientific psychological model of encoding, storage and retrieval may be rendered
non-explanatory, if it fails to recognise that information-retention and remembering are inter-
defined concepts.
The idea that past events cause changes in the observer, which enable the person to remem-
ber, has been popular at least since Plato’s metaphor of imprinting. Perhaps the explanation
sought by neuro-physiologists, when they attempt to describe the ‘mechanism’ of memory,
need not be circular. We now consider more recent discussions of localisation and connectiv-
ity, which may allow that discovering explanatory mechanisms is not the same as capturing the
essential features of remembering.
4. Critique of Wittgenstein’s view

Wittgenstein’s rejection of the trace theory has been criticised for denying the possibility of
scientific study of learning and remembering. It does indeed undermine scientistic psychology, as
its tendency is humanistic – but Wittgenstein’s position is, I believe, consistent with scientific
psychology. There are many important questions that psychologists could continue to investi-
gate: optimum conditions of recall, and how subsequent suggestion causes inaccuracies; learning
tasks that can be performed by amnesiacs; localisation of brain function governing different
recognitional capacities, and so on.
The trace-concept is an illegitimate extension of the concept of localised brain-function,
almost an article of faith within the brain sciences. Genuine localisation concerns particular capac-
ities such as recognition or memory. Because these clearly are capacities, there is less likely to be
a misunderstanding about what imputing localised function means, viz. that damage to a certain
area of the brain means that function will be impaired. Neural processes are not the essence of
facial recognition, for instance, but its causal condition. In fact, localised brain-function is not as
554
Ludwig Wittgenstein
well-founded as many suppose – once plasticity is admitted, claims of localisation look empty –
but the idea of a localised trace is even more dubious.
Understanding of localisation has been qualified in more recent psychological research, how-
ever. Most psychologists now advocate distributed rather than localised traces, espousing the
holistic, connectionist position that information is stored everywhere in the brain. For Sutton
(1998: 301, n3), Wittgenstein’s followers neglect this development, attributing views to their
opponents which they no longer hold, if they ever did; distributed models are not high-fidelity
reproduction systems, and psychologists now deny any correspondence between brain traces
and what is remembered, no longer assuming the engram as a copy of a particular memory.
Stern goes further, arguing that Wittgenstein’s remarks on memory may be interpreted as ‘[fore-
shadowing] some of the most promising contemporary work’ in connectionist cognitive science
(1991: 203).
This response seems questionable. Connectionism confronts the dilemma of ‘false or empty’.
Without a structural analogue, the ‘distributed trace’ is not a trace at all, but amounts simply to
the truism that, as Reid put it, ‘memory is dependent on some proper state or temperament of
the brain’ (Reid 2002: III.vii, 282). If one regards information as essentially a personal, not a sub-
personal or impersonal concept, ‘total neural activity’ cannot constitute memory-information as
proponents of distributed traces envisage. On a Wittgensteinian view, connectionism suggests
ways of theorising about the mechanisms of memory, not of capturing its essential features.
Notes
1 Pollock may be claiming that introspection reliably rather than infallibly tells us when a belief is based on
personal memory.
2 For detail on Wittgenstein’s critique, see Hacker (1996).
3 OC 141, discussed in Hamilton (2013: Ch. 6.4).
4 Discussed in Hamilton (2013: Ch. 6.5).
Related topics
• Memory traces
• Memory images
• Memory and self-consciousness
• Bertrand Russell
Further reading
Hacker, P. (1996) Wittgenstein: Mind and Will, Part I: Essays, Oxford: Blackwell.
Moyal-Sharrock, D. (2009) ‘Wittgenstein and the Memory Debate’, New Ideas in Psychology 27(2):
213–227 – surveys impact on neuropsychology of Wittgenstein’s elucidations of memory.
Schulte, J. (1996) Experience and Expression: Wittgenstein’s Philosophy of Psychology, Oxford: Oxford
University Press – scholarly interpretation of Wittgenstein’s remarks on memory and other topics.
Stern, D. (1991) ‘Models of Memory: Wittgenstein and Cognitive Science’, Philosophical Psychology 4(2):
203–18 – critique of model of memory as a store, and defence of a non-representational connectionist
alternative.
555
Andy Hamilton
References
Wittgenstein references
PI = (1953) Philosophical Investigations, Oxford: Blackwell.
PG = (1980) Philosophical Grammar, Oxford: Blackwell.
PR = (1980) Philosophical Remarks, Chicago: University of Chicago Press.
OC = (1969) On Certainty, Oxford: Blackwell.
RPP = (1991) Remarks on the Philosophy of Psychology Vol. 1, Oxford: Wiley-Blackwell.
Z = (1981) Zettel, Oxford: Wiley-Blackwell.
Secondary sources
Anscombe, G.E.M. (1974) ‘Memory, “Experience” and Causation’, in Lewis, H.D. (1974), also in her
(1981).
—— (1981) Collected Philosophical Papers Vol. II: Metaphysics and the Philosophy of Mind, Oxford: Blackwell.
Austin, J.L. (1962) Sense and Sensibilia, Oxford: Oxford University Press.
Epstein, R. (accessed 2016) ‘The Empty Brain’, Aeon, https://aeon.co/essays/your-brain-does-not-
process-information-and-it-is-not-a-computer.
Fogelin, R. (1976) Wittgenstein (2nd edn 1987), London: Routledge & Kegan Paul.
—— (1994) Pyrrhonian Reflections on Knowledge and Justification, Oxford: Oxford University Press.
Hacker, P. (1996) Wittgenstein: Mind and Will, Part I: Essays, Oxford: Blackwell.
Hamilton, A. (2013) The Self in Question: Memory, the Body and Self-Consciousness, Basingstoke: Palgrave
Macmillan.
—— (2014) Routledge Philosophy Guidebook to Wittgenstein and ‘On Certainty’, London: Routledge.
Malcolm, N. (1977) Memory and Mind, Ithaca, NY: Cornell University Press.
Martin, C. and Deutscher, M. (1966) ‘Remembering’, Philosophical Review LXXV: 161–96.
McGinn, M. (2012) ‘Non-Inferential Knowledge’, Proceedings of the Aristotelian Society 112(1): 1–28.
Meixner, U. (2014) Defending Husserl: A Plea in the Case of Wittgenstein & Company versus Phenomenology,
Berlin: De Gruyter, https://books.google.co.uk/books?id=nvHmBQAAQBAJ&pg=PT270&dq=%22
Remembering+has+no+experience-content%22+%2B+Meixner.
Mounce, H. (1994) ‘The Philosophy of the Conditioned’, Philosophical Quarterly 44(175): 174–90.
Moyal-Sharrock, D. (2009) ‘Wittgenstein and the Memory Debate’, New Ideas in Psychology 27(2): 213–27.
Plato (1973) Theaetetus, trans. J. McDowell, Oxford: Clarendon.
Pollock, J. (1986) Contemporary Theories of Knowledge, Totowa, NJ: Rowman & Littlefield.
Read, R. (accessed 2014) ‘Therapy’, www.academia.edu/205510/Therapy.
Reed, E. (1997) From Soul to Mind: The Emergence of Psychology from Erasmus Darwin to William James, New
Haven, CT: Yale University Press.
Reid, T. (2002) Essays on the Intellectual Powers of Man, D. Brookes (ed.), Edinburgh: Edinburgh University
Press.
Russell, Bertrand (1935) Religion and Science, London: Thornton, Butterworth.
Schulte, J. (1996) Experience and Expression: Wittgenstein’s Philosophy of Psychology, Oxford: Oxford
University Press.
Squires, R. (1969) ‘Memory Unchained’, Philosophical Review 77: 178–97.
Stern, D. (1991) ‘Models of Memory: Wittgenstein and Cognitive Science’, Philosophical Psychology 4(2):
203–18.
Stout, G.F. (1930) ‘In What Way Is Memory-Knowledge Immediate’, Studies in Philosophy and Psychology,
London: Macmillan.
Sutton, J. (1998) Philosophy and Memory Traces, Cambridge: Cambridge University Press; Ch. 16 available
at www.johnsutton.net/Sutton_1998_PAMT_4_16_Attacks-on-traces.pdf.
556
47
MARTIN HEIDEGGER
Taylor Carman
1.
In his 1927 magnum opus, Being and Time, Heidegger advances a detailed account of the
unique temporal structure that, he believes, constitutes what it is for a human being (Dasein)
to be. According to Heidegger, being is always intelligible only in terms of time (hence the
title of the book), but different kinds of being are defined by different forms of temporality.
So, for example, the being of objects with properties consists in their occurring at a now,
that is, a single moment in a linear sequence of moments. Human beings, by contrast, exist in
a distinctive temporal way by standing in relation to their own existence, by their existence
being an “issue” for them.1
Our being is an issue for us in three basic ways, each bound up with a different dimension or
“ecstasis” of time. The first is by our constantly having to press or “project” into possibilities—that
is, affordances, opportunities, courses of action, goals, purposes, or broadly speaking ends—which
constitute our sense of having a future. The future that we thereby “have” (or that in despair we
might feel we lack) is not the series of chronologically later moments in time that are not yet now,
but rather the horizon of goals, purposes, and other ends that give point to what we are currently
doing, and more generally how we are living our lives.
Another way our being matters to us is by our constantly “falling” (Verfallen), or being pulled
into an absorption in the present situation confronting us. Unlike the present understood as a
very short or perhaps instantaneous moment in a linear chronology, the now we inhabit in
ordinary life and experience is not a quantity of duration, but a situation defined by concerns,
needs, demands, obstacles, and opportunities. The now of what Heidegger calls “world time”
is what we understand when we understand that (say) now it’s time to get up, time to get to
work, lunchtime, bedtime, and so on.
A third way in which our existence is an issue for us lies in our affective disposition or
“attunement” to the way things already have been (the German word, Befindlichkeit means some-
thing like how things are going or how you’re doing). We are, Heidegger says, always in one or
another mood (Stimmung), which reveals our “thrownness” (Geworfenheit) into the world, and
so constitutes the most basic way in which we can be said to have a past.
Each of these three temporal ecstases—future, present, past—in turn has an authentic and
an inauthentic mode. To exist authentically, for example, is to project into the future by
557
Taylor Carman
anticipating or “running forth” (Vorlaufen) into one’s possibilities wholeheartedly, while being
inauthentic means merely expecting or “awaiting” (Gewärtigen) something. Likewise, whereas
our inauthentic relation to the present is simply to let given circumstances present themselves
to us (Gegenwärtigen), the authentic now is what Heidegger calls the “moment” in which one
discerns the concrete situation and responds to it resolutely (the German word for moment,
Augenblick, literally means glance of the eye). Finally, the inauthentic mode of “having been”
(Gewesenheit) is “forgetting” (Vergessenheit), while the authentic way of having a past is what
he calls repetition or “retrieval,” or (translating more freely) recollection (the German word
Wiederholung literally means taking up again), by which Heidegger means a kind of recovery or
reappropriation of who one has already been, “a coming back to one’s ownmost self” (SZ 339).
This, then, is Heidegger’s existential concept of memory: the retrieval or recovery of
oneself from an always already forgotten past. That idea, that one’s past could somehow be
“forgotten” a priori, before ever having been retained in memory, challenges common sense:
surely we can forget something only by first having it in mind, then losing it. On the contrary,
Heidegger insists, “remembering is possible only on the basis of forgetting, not vice versa” (SZ 340;
original emphasis).
To ease the apparent paradox, it is important to recall that Heidegger’s notion of ecstatic
temporality dispenses with the mathematical concept of time, according to which each moment
is a dimensionless point in a linear series of nows. Neither, therefore, is the kind of forgetting
that has primacy in his account an instantaneous erasure of experience immediately following its
occurrence in a punctual present. Does Heidegger’s notion of forgetting therefore presuppose
a kind of memory, after all, namely the short-term or working memory that allows us to retain
information for a few seconds or minutes, long enough that we can think clearly without losing
track of what we’re doing?
Not necessarily. Of course, whether we apply the term “memory” to the lingering horizon
of the proximal past, which lends the present its aspect of temporal duration and openness, is
a matter of terminological convention. There is, however, good reason to resist the conflation
of that past horizon of the now with the more robust sort of past from which we can retrieve
and recover things only by remembering—again, literally re-collecting—them. The reason for
this, as Husserl had argued in The Phenomenology of Internal Time-Consciousness, is that since
memory presupposes experience, the experienced present must already have its own span or
extension, with past and future horizons built into it a priori, in order to be experienced at
all, and subsequently recalled.2 Husserl therefore distinguishes memory and expectation proper
from the “retentions” and “protentions” essential to a present awareness of anything as abiding
and enduring in an experienced now. Husserl’s insight was to see that any account of temporal
awareness that fails to acknowledge at the outset this a priori horizonal dimension of present
experience will be doomed to the hopeless task of trying to specify how a purely punctual or
momentary content of consciousness (an impression, an idea, a representation) manages to refer
back to a past (or, for that matter, forward to a future) moment. Does it bear some distinc-
tive mark indicating its pastness?3 For Husserl, then, as for Heidegger following him, precisely
because the temporal openness of the present is a necessary condition of experience as such, it
must a fortiori be a condition of the possibility of remembering things experienced in the past.
We are now in a better position to see why Heidegger insists that memory presupposes
forgetting, rather than vice versa. What is forgetting? Neither the immediate obliteration of an
instantaneous present experience nor the subsequent loss of an already formed memory, but the a
priori back horizon of obscurity that circumscribes the present in which we understand ourselves
now as always already having been. Forgetting in this sense is not merely a failure to remember, but
instead positively situates us temporally, just as the boundaries of the visual field orient and direct
558
Martin Heidegger
our gaze and are not, as Merleau-Ponty observes, mere “imperceptions.”4 Unlike the sharp,
determinate edges of a frame around a window or a picture, that is, the boundaries of the visual
field are horizons in the same sense in which the temporal boundaries of the present are horizons.
For just as the horizons of the visual field are where things trail off into an obscurity in which
they are both unseen and yet there to be seen, namely if we turn and look at them, so too the
past horizon of the present is where things fall away into a kind of forgottenness in which they
are no longer present, though they remain there in time to be recovered, collected, remembered.
Heidegger’s claim that forgetting is more fundamental than remembering is not a psycho-
logical hypothesis, then, but a phenomenological description of our primordial experience of
time. Or, to be more precise, it is not even a description of our “experience” of time, if by that
we mean our momentary subjective consciousness of time passing. Heidegger’s claim instead
concerns the temporal structure of human existence, that is, the basic framework of intelligibility
that supports our understanding of ourselves as persons—subjective or objective, conscious or
unconscious. The deeper philosophical point of insisting that memory presupposes forgetting
is to argue that the temporality of existence is a condition of the intelligibility of any conscious
attitude we might have toward things in the past. The horizon of the past is itself there, as it were,
a priori—that is, “prior” to our capacity to remember things having happened in it. Heidegger’s
concept of forgetting is a phenomenological reminder of our most fundamental understanding
of the past as an open temporal horizon from which things can be recollected.
2.
Near the end of Being and Time, Heidegger expands his account of authentic temporality from
the context of individual existence to the collective historical life of a community. There he tries
to say what it is for Dasein to have a shared cultural memory, or “heritage” (Erbe). His account
can best be understood as an attempt to steer between two competing alternatives, or tempta-
tions. The first is traditionalism, the idea that we live in an already fixed, determinate cultural and
historical world whose past has normative authority in determining our future. The second is
voluntarism, the idea that history and cultural tradition do not really constrain us at all, that we
are radically free to choose or even create the possibilities that shape our future.
The semblance of truth in each view is perhaps merely the mirror image of the implausibil-
ity of the other. The kernel of truth in traditionalism, for example, is that a cultural world is not
something we just imagine or invent for ourselves. Instead, just as we find ourselves given over to
our individual situations in life, so too we are thrown into a world of established tradition, a world
that we cannot but take for granted, at least initially, as normal, natural, and proper. A phenom-
enologically sensitive account of collective historical existence, that is, needs to acknowledge the
givenness or facticity of tradition, as opposed to our simply fabricating or inventing it.
Heidegger, however, is no traditionalist; indeed, when he uses the words Tradition and
traditionell, they almost always carry negative connotations. For him, history is never just unprob-
lematically given and normatively authoritative—or rather, the idea that it is is symptomatic of
an inauthentic relation to the past. When we remember or recollect ourselves authentically,
although our past is ours partly in virtue of our simply finding ourselves thrown into it, it is
nevertheless we who must actively “take over” our history as a heritage, as opposed to merely
accepting it passively as an established tradition (SZ 383).
It is harder to distinguish Heidegger’s concept of heritage from voluntarism. For the vol-
untarist, although history provides resources for the choices we make, in truth it imposes no
real constraints or obligations on us. Appealing to it as either authoritative or inexorable, as
traditionalists do, is at best a kind of bad faith designed to conceal the unsettling fact of our
559
Taylor Carman
radical freedom. What Heidegger says about heritage sometimes sounds like voluntarism in
this sense. For on his account, to inherit possibilities authentically is not to receive something
ready-made from a tradition, but to choose to grant oneself what one takes up, and moreover to
transform it in the process: “Resolutely coming back to its thrownness involves handing down to
itself (Sichüberliefern) possibilities that have come down, though not necessarily as having come
down” (SZ 383; original emphasis). In authentically taking over a heritage, “Dasein hands itself
down (sich . . . überliefert) . . . in a possibility it has inherited, yet nonetheless chosen” (SZ 384;
original emphasis).
What Heidegger’s conception of heritage shares with voluntarism is the thought that human
temporality is most fundamentally futural. Heidegger stresses the importance of this idea by writ-
ing the following sentence entirely in italics:
Only an entity that is essentially futural in its being, so that it is free for its death and can let
itself be thrown back upon its factical There [Da] by shattering itself against death—that is, only
an entity that, as futural, is equiprimordially having been—can, by handing down to itself the
possibility it has inherited, take over its own thrownness and be in the moment [augenblicklich]
for “its time.”
(SZ 385)
In carving out a path between traditionalism and voluntarism in this way, Heidegger acknowl-
edges our thrownness into history and our free projection into possibilities, both of which he insists
are equally fundamental aspects of historical existence. Still, once we have seen that Dasein is
somehow fundamentally futural, that it “can let itself be thrown back upon its factical There,” but
that it ultimately chooses its own possibilities for itself, “handing them down to itself” from
the tradition, it is hard to see how history can still be said to constrain that choice, not just by
offering limited resources—exemplary works, precedents, role models, the wisdom of experi-
ence, and so on—but by making some possibilities genuine live options while reducing others
to nonstarters and dead ends.
What is at any rate clear on Heidegger’s account is that in order to be historically resolute,
we must somehow identify with our own contemporary world; its concerns must be relevant
to the world we share with others: “Our fates have already been guided in advance in being-
with-one-another in the same world, and in resoluteness for definite possibilities. . . . Dasein’s
fateful destiny in and with its ‘generation’ constitutes the full, authentic happening (Geschehen)
of Dasein” (SZ 384–5). Moreover, the constraints of heritage must somehow manifest them-
selves in our retrieval or recollection of possibilities from the past:
The resoluteness that comes back to itself and hands down to itself then becomes the
retrieval of an existential possibility that has come down. Retrieval is explicitly hand-
ing down, that is, going back to possibilities of Dasein that has been (des dagewesenen
Daseins). The authentic retrieval of an existential possibility that has been—Dasein
choosing its hero for itself—is grounded existentially in forerunning resoluteness; for
therein the choice is first chosen that makes one free for the struggle of following and
being loyal to what is retrievable.
(SZ 385; original emphasis)
Arising as it does from a resolute futural projection, the retrieval of a heritage does not merely
let something previously occurrent recur. Rather, it responds (erwidert) to the possibility of an
existence that has been. A resolute response to a possibility in the moment (augenblicklich), however,
560
Martin Heidegger
is at the same time a “revocation” (Widerruf) of “the past,” understood as that which is merely
unfolding and working itself out in the present. Authentic retrieval of a heritage, that is, neither
simply surrenders itself to the past nor aspires to novelty; in the authentic moment, its commit-
ment is neither to rigid tradition nor to arbitrary innovation (SZ 386). Authentic historical
resoluteness is a kind of loyalty (Treue), but it is a loyalty to itself, or more precisely to possibili-
ties it has retrieved for itself from its heritage: “Resoluteness constitutes the loyalty of existence
to its own self. As resoluteness that is ready for anxiety, loyalty is at the same time the possible
veneration of . . . retrievable possibilities of existence” (SZ 391; original emphasis).
3.
By the middle of the 1930s, Heidegger also began to extend his concepts of memory and
forgetting from the account of individual and historical existence in Being and Time to an inter-
pretation of the history of philosophy. The metaphysical tradition from Plato to Nietzsche,
he now says, amounts to a “forgetting of being” (Seinsvergessenheit), that is, a forgetting of the
wonder or mystery that there is anything at all rather than nothing. In construing the tradition
as a forgetting of being, Heidegger implicitly defines his own later philosophical project as an
effort of memory: an attempt to recall, recover, or recollect the mystery inherent in our ability
to understand that and what anything is.
In his interpretation of metaphysics as a forgetting of being, as in his earlier account of the
temporality of memory in Being and Time, Heidegger does not conceive of forgetting as a mere
loss of something previously remembered. For although, on his account, the Presocratic thinkers
of pre-Classical Greece were naively, unreflectively immersed in the wonder of being, they did
not explicitly entertain the question Heidegger now wants us to recall, namely, What does it mean
to be? Rather, by the time that naïve sense of wonder had begun to fade away—canonically, he
suggests, in the cave allegory in Plato’s Republic—it was no longer even clear what had been lost.
The most profound forgetting, one might say, is the forgetting that one has forgotten. So, here
again, for Heidegger, memory is made possible by forgetting, not vice versa.
Moreover, echoing his earlier argument that the temporal present is defined by its back hori-
zon fading into an initially unremembered past (and, so too, by its forward horizon of projected
possibilities anticipating an always as yet unrealized future), Heidegger maintains in his later
work that the mystery of being consists precisely in its constant retreat or withdrawal, its essen-
tial resistance to clarification and explicitness—in short, its continually and necessarily lapsing
into forgottenness. The forgetting of being characteristic of metaphysical thinking is thus no
mere historical accident, but a defining aspect of the elusiveness of being as such. Again, just
as you cannot turn your gaze to look directly at the boundaries of your own visual field, so
too being as such is the horizon of the intelligibility (and unintelligibility) of entities, a horizon
that cannot itself be thematized and understood as an entity. Cultivating a memory of being is
thus no ordinary effort to remember something definite or particular, but an attempt to remain
attuned to the essential unfathomability and ungraspability of being.
Heidegger’s reflections on the task of remembering being cannot, therefore, be a theoreti-
cal effort in philosophy in any familiar sense. It is instead a task he thinks is shared by thinkers
and poets, though they pursue it differently. What great thinkers do, Heidegger suggests, is to
articulate the understanding of being that holds sway, dimly yet pervasively, in their own age.
Poets, by contrast, are attuned to the presence and absence of the sacred. The modern poet par
excellence, according to Heidegger, was Friedrich Hölderlin (1770–1843), precisely because his
poetry contains in itself a questioning concerning the essence of poetry and the place and task of
the poet. The first two words in Hölderlin’s question, “Wherefore poets in a time of dearth?”
561
Taylor Carman
(Wozu Dichter in dürftiger Zeit?) provides the title for an essay in which Heidegger compares
the metaphysical thinking still at work in the poems of Rainer Maria Rilke (1875–1926) to
Hölderlin’s more profound, nonmetaphysical understanding of what it means to poetize in
modernity. Both Rilke and Hölderlin felt intense anguish at something like what Nietzsche
called the death of God, that is, the dissipation and loss of a sense of the sacred in European
culture. In the poem “Bread and Wine” (Brod und Wein), Hölderlin writes of the wine god’s
wreath of ivy “remaining and bringing a trace of the fugitive gods down to those bereft of gods,
in darkness” (Weil er bleibet und selbst die Spur der entflohenen Götter / Götterlosen hinab unter das
Finstere bringt). Echoing these lines, Heidegger writes, “Those who,” like Rilke and Hölderlin,
“bring mortals the track of the fugitive gods in the darkness of the world’s night . . . are ‘poets
in a desolate time.’”5 Hölderlin is foremost among them, but what such poets have in common
is that their poetizing (Dichten) is at once a thinking (Denken): “The distinctive mark of these
poets consists in the fact that for them the essence of poetry has become worth questioning.”6
In an essay on Hölderlin’s poem “Remembrance” (Andenken), Heidegger again returns to
the idea that a heritage retrieved from the past must be founded on a prior taking up of pos-
sibilities in anticipation of a future. A phrase in the poem, “Noch denket das mir wohl . . .” means
(something like) I remember well, or I still think back on, but its syntax literally represents some-
thing (das) thinking “to me” (mir), something to which the poet is now responding poetically in
his poem. His response, however, is no mere nostalgic, backward-looking gaze into an irretriev-
able past. Rather, as the closing lines of the poem say,
Und die Lieb’ auch heftet fleißig die Augen,

Was bleibet aber, stiften die Dichter.
And love lifts the eyes and holds them fixed,

but what remains is the mark of the poets.
In these verses, as throughout the poem, the poet looks both backward and forward—or bet-
ter, he looks backward by looking forward, at once remembering and anticipating something
“lasting” (was bleibt), something that poets alone—those of the past and those yet to come—
establish or “found” (stiften). The poet, Heidegger says, “must think of (denken an) what has
been and of what is to come, or indeed of what has been as what is to come.”7
Notes
1 Heidegger, M. Sein und Zeit (Tübingen: Niemeyer, 1927; 15th edn. 1979), 42 et passim (hereafter SZ,
with page references to the German edition, translations mine). Being and Time, J. Macquarrie and
E. Robinson, trans. (New York: Harper & Row, 1962.)
2 Husserl, E. Vorlesungen zur Phänomenologie des inneren Zeitbewußtseins. M. Heidegger, ed. (Tübingen: Max
Niemeyer, 1928; 2nd edn. 1980). The Phenomenology of Internal Time-Consciousness. J. S. Churchill, trans.
(Bloomington: Indiana University Press, 1964).
3 Hume, for example, wrestles with this problem in A Treatise of Human Nature, I.i.3 and I.iii.5. See Daniel
Flage, Chapter 40 of the present volume.
4 Merleau-Ponty, M., Phenomenology of Perception, D. Landes, trans. (London: Routledge, 2013), 11 transla-
tion modified.
5 Heidegger, Holzwege, 6th edn. (Frankfurt am Main: Vittorio Klostermann, 1980), 315. Off the Beaten
Track, J.Young and K. Haynes, trans. and ed. (Cambridge: Cambridge University Press, 2002), p. 240.
6 Ibid.
7 Heidegger, Erläuterungen zu Hölderlins Dichtung, 5th edn. (Frankfurt am Main: Vittorio Klostermann,
1981), p. 99 (emphasis altered). Elucidations of Hölderlin’s Poetry, K. Hoeller, trans. (Amherst, NY: Humanity
Books, 2000), p. 123.
562
48
PAUL RICOEUR
1. Introduction
The issue of memory is not addressed in the work of Paul Ricoeur in a continuous way. In
the volumes of Time and Narrative, published in French between 1983 and 1985, the question
of the narration is mostly seen in relation to temporality and to the writing of history and
fiction. The issue of memory can be considered as an underlying theme in the Ricoeurian
reflection, which finally leads in 2000 to the publication (in French) of Memory, History
and Forgetting (from now on, MHF). In many ways, this publication works as a concluding
reflection on the interlaced relationships between narrative, history and imagination, while
introducing a new perspective, or at least a more explicit perspective about the relationship
between history and memory as well as the notion of forgetting, previously ‘neglected’ in the
work of Ricoeur. According to the words of Ricoeur in the introduction (Ricoeur 2004:
xv), MHF thus functions as a necessary supplement to the three volumes of Time and Narrative
(1983–85 in French) and Oneself as Another, edited in French in 1990.
In this chapter, I propose to return to the construction and the role of the memorial dis-
course in Ricoeur’s work and its foundations, looking particularly at the last volume of Time and
Narrative (1988), Oneself as Another (1992) and of course MHF (2004). I believe the development
of the memorial discourse in Ricoeur is forged through three main axes that will also structure
this chapter. First, the question of being and identity, as well as the question of time and tempo-
rality, are fundamental in the development of Ricoeur’s discourse on memory. Then we will see
that the relationship of the memory discourse to history – both essential poles in the Ricoeurian
reflection on the representation of the past – is a tangled one. Finally, we will see that the
Ricoeurian thought on the representation of the past necessarily leads to the consideration of
the mediation of the past, which raises the question of the imagination and of the necessary criti-
cal perspective; this third axis is what we will discuss in the conclusion to this chapter.
2. Time and being

In a letter to Georges Izambard, on 13 May 1871, Arthur Rimbaud wrote these words now
engraved in the collective memory: ‘It is wrong to say: I think. It should say: one thinks me.
I is another’ (Rimbaud 2007).1 This formulation highlights the complexity of the issue of
563
identity, the instability of the question of identity and, in particular, the complexity of the
relationships between the self and the other, or from the self to the other. This question has
stimulated the philosophical debate since the famous cogito ergo sum of Descartes and has also
been fundamental in Ricoeurian thought. Making a short backtrack on different positions
in this debate, and in particular Husserl, Sartre and Levinas, will allow us to highlight the
uniqueness of Ricoeur’s position. The Ricoeurian discourse on otherness goes beyond the
necessity of the ‘domestication of the other’, which we find in Husserl and Sartre, or the idea
of ‘infinity in us’, which is central in Levinas.
According to Husserl, the other is first perceived by the subject in its otherness, Das Fremde,
and is then identified by reintroducing the same, Das mir eigene, which makes the other some-
how familiar. The other becomes in this way another me, an alter ego (Husserl 1997: 130). This
transfer from otherness to sameness happens, according to Husserl, thanks to the intervention of
the imagination; the other and the elsewhere become the same and the here. This transfer to the
same and the here, which Husserl calls aktuellen (Husserl 1997: 92), implies a certain appropria-
tion or integration of the otherness.
In Sartre, and especially in L’être et le néant (1943),2 we recognise the Husserlian approach
to the issue of otherness where any altruistic ethic seems impossible, since ‘the for-others is a
necessary condition for the constitution of the for-oneself as such’ (Sartre 1976: 139).3 Again,
the question of otherness seems here to only be considered in relation with the development of
the ‘I/me’ and necessary in the ‘domestication’ of the other.
Emmanuel Levinas, somewhat in the tradition of Kantian morality, sketches a different
approach to the issue of otherness. For him, it is not ‘the ego’ that grants meaning to the other,
the other has a meaning independent of the ‘I/me’. The other is a constituent part of the cogito
but this one always ends up escaping from the ‘I/me’. In Levinas, there is an ethical and moral
approach to the issue of otherness: the other is actually related to the presence of ‘the Infinite
in us’ (Levinas 1990: 58). According to Levinas, the question of otherness has to be considered
in relation to the idea of Infinity, which by definition escapes us, even though it is a part of the
me. It appears then, in Levinas, a paradoxical coexistence of the self and the other.
Ricoeur considers in Oneself as Another that the reflection around the cogito ergo sum has
stalled. According to Ricoeur, this debate witnesses a clash of positions between those who
consider, like Sartre, that any altruistic morality is impossible, and others, like Levinas, who say
on the contrary, that the other has an existence independent of the me who somehow ends up
escaping the ‘I/me’. Although my summary is necessarily brief, this short interlude will allow us
to understand the Ricoeurian position better.
Ricoeur proposes an original approach to the question of identity by introducing a ‘herme-
neutics of the self’, including the recognition of both a plurality and an otherness within the being.
The other is not only considered as an external element but also as an element of the being and
of the self. Ricoeur’s starting point will be to consider not only the ‘me/I’ or what he calls the
sameness or identity-idem (Ricoeur 1992: 1); that is to say, the deep constituent and unchange-
able characteristics of the person, but also the ‘self’ of the being, or what he calls ipséité (selfhood)
or identity-ipse, which contains otherness. Being able to consider otherness as both an element
outside the being, and as a component of the self is a central element of Ricoeurian philosophy,
which will also be important to Ricoeur’s philosophy of memory.
This perspective finally allows us to consider the question of identity through a double-
reflexive prism, the ‘myself’ and the ‘self-other’. In this connection, Ricoeur also makes an
essential precision when he affirms that ‘as’ in the title of his work, Oneself as Another, indicates
not only the idea of a comparison but also the idea of an entanglement: oneself similar to the other
gives way to the oneself as another (Ricoeur 1992: 3). The relationships of the being to otherness
564
Paul Ricoeur
and to its own otherness is also fundamental in the Ricoeurian notion of ‘narrative identity’, in
which he introduces the complex issue of mediation, or narration, of the identity.
According to Ricoeur, the telling of the self allows the subject who speaks or writes, to situ-
ate him/herself and to recognise him/herself in the story he tells (Ricoeur 1988: 247). Therein,
there is a Freudian inspiration (Ricoeur 1988: 246), since the fact of telling has the ability to
collect scraps or disparate elements in the memory and reconstruct them into coherent and
acceptable individual or collective histories. The mediation of narrative identity finally allows us
to consolidate, to confront, and to question the sameness and the selfhood that exists, in tension,
within the same entity (Ricoeur 1988: 245).
The narrative identity in Ricoeur has also a fairly similar role to that of memory, especially
if we consider how Jan Assmann relates the question of memory to the question of identity.
According to Assmann, it is because the ‘I/me’ is an unstable entity that it needs memory
and narration to offer a synthesis of identity, which is of course always partial and tempo-
rary (Assmann 2008: 109). Processes of telling and remembering can be considered as central
instruments that allow the subject to confront the question of identity, both individually, and
collectively (Ricoeur 1988: 247). But the question of identity, like that of memory, whether
individual or collective, cannot be considered statically; according to Ricoeur, the fragility and
the instability of the memory liaises with the fragility and instability of identity (Ricoeur 2004:
80). Telling as remembering are above all dynamic processes of change and renewal that allow a
subject to question or challenge the representations and the significances of the past. They allow
in that negotiating new ways of being with oneself and with others.
But telling as remembering are also processes in tension, not only because of the relation of
being to otherness, but also because of the relation of being to the question of time:
As the primary cause of the fragility of identity we must cite its difficult relation to
time; this is a primary difficulty that, precisely, justified the recourse to memory as the
temporal component of identity, in conjunction with the evaluation of the present and
the projection of the future.
(Ricoeur 2004: 81)
The following example of memorial testimony illustrates fairly well this temporal tension that
exists in narrative and memorial processes such as Ricoeur has stressed it:
I was just a young child during the 1914–18 war. I don’t really remember it very
clearly. I had an uncle who went off to the war. In 1914, a piece of shrapnel fell right
in front of me. It was lunchtime. The table was laid, and was then destroyed. When
we returned the dog had eaten everything! I remember that we were bombed, but
from afar. At the time, it was just guns; there were not many aeroplanes yet. I could
have been killed that day.
(Dessingué and Ryckebusch 2011: 190)
This particular testimony is the story of an elderly woman in the northern region of France
who transcribes in a letter her memories from the First World War. In this narrative, it would
be tempting to identify a before and an after, a time of the event and a time of the narration.
Relying again on Ricoeur, temporal relationships in this story are actually more complicated
than they appear (Ricoeur 1988: 241). In other words, temporality is dependent on narration
and therefore it cannot be considered as an axiological and linear phenomenon that would run
from past to future through the present.
565
In reality, the time of testimony and of narration is always the present; narrative identity is
built through a discourse at a defined time, which by definition is the present of the narrator.
The same is true for memory: remembering always happens in the present of the one who
actually remembers. In this sense, the Augustinian conception (see Chapter 35, this volume)
of temporality in The Confessions illustrates perfectly these words:
But what now is manifest and clear is, that neither are there future nor past things.
Nor is it fitly said, ‘There are three times, past, present and future’; but perchance
it might be fitly said, ‘There are three times; a present of things past, a present of
things present, and a present of things future.’ For these three do somehow exist
in the soul, and otherwise I see them not: present of things past, memory; present
of things present, sight; present of things future, expectation. If of these things we
are permitted to speak, I see three times, and I grant there are three. It may also be
said, ‘There are three times, past, present and future’, as usage falsely has it. See, I
trouble not, nor gainsay, nor reprove; provided always that which is said may be
understood, that neither the future, nor that which is past, now is. For there are but
few things which we speak properly, many things improperly; but what we may
wish to say is understood.
(Augustine 2009: XI, xx, 26)
In other words, telling’s relation to memory is essential since it allows the reintegration of past
events in the present. The act of remembering is fundamentally an act of (re-)actualisation of
the past. Remembering is committing an act of temporal destabilisation, which constantly con-
fronts us and which is an integral part of our lives. We live in the present and the stories we
tell/remember occur in our present time. The narrative identity, in turn, is concretised in this
temporal tension that arises in the present. Although we would seek to regain the fundamental
characteristics of a group or individual through writings and through time, which would allow
us to study the construction of a narrative identity over time, we would necessarily be led to the
(re-)actualisation of all these discursive data.
In my opinion, this is exactly what the First World War testimony shows us. All the ele-
ments of this testimony are actually passed through the filter of contemporaneity of the narrator
and the filter of memory. In this way, the hyletic phenomenology of Husserl is once more
jeopardised. The revelation and even the knowledge of what the Greeks called ‘the hyle’ (the
first original thing, the lived thing, the thing in its primitive state) actually integrates a necessary
action (act of representation/dissemination), which can be discursive, narrative or otherwise. It
is at this stage – what the phenomenology called ‘noesis’ on one side which is the very act of
knowing, and ‘noema’ which is the act of representation – that the modification of the ‘hyle’
happens. It is also this kind of modification to which we are attending in processes of remem-
bering, from the original event to its retransmission or representation. The narrative identity
therefore is situated at the intersection of these two ‘realities’ described by the ‘noesis’ and the
‘noema’; it is a phenomenon in constant tension.
The instability of the narrative identity created in this testimony about the First World War is
here not only due to the distance between the time of testimony and the time of the event, but
also to the process of actualisation which arises in the present of the narrator. Thus, we attend to
an obvious and permanent alternation between different temporal sequences in the testimony. It
includes both the accounts of the main event in the past, ‘In 1914, a piece of shrapnel fell right
in front of me’, and ‘It was lunchtime.’ But we also attend to the telling of contextual accounts:
‘I was just a young child’, ‘I had an uncle’, and finally there is a number of sequences in which
566
Paul Ricoeur
the voice and the conscience of the narrator are clearly perceptible: ‘I don’t really remember’,
‘I remember that we were bombed’, ‘There were not many aeroplanes yet’, and ‘I could have
been killed that day.’
In Ricoeurian reflection, the act of telling is characterised by a temporal tangle where
memory plays a key role of actualisation, re-actualisation and sometimes even re-activation
of the past; for instance, if we think about traumatic experiences. In Ricoeur, remembering
is closely linked to telling and to the telling of the self. Memory, associated with the idea of
narrative identity, is thus considered through the act of mediation, as a discursive and narrative
construction, arising from individual or collective intention. But they can also, as I shall show
in the following, be considered as the result of a socio-cultural and historical context.
3. Epistemology of history and phenomenology of memory

As Ricoeur says otherwise, ‘narrative identity’ is both a construction, but it also carries with
it the risk of becoming an assignation (Ricoeur 1988: 246) and therefore a risk of becoming
a casual and more or less passive convocation of a ‘me’ or of an ‘us’. Through the narrative,
whether fictional or inspired from reality, the individual or the community may be assigned an
identity or a history, which can sometimes be difficult to question or to deconstruct.
To prevent excesses of assignation, Ricoeur suggests first to consider the idea of a fair politics
of memory, and states his intention clearly in the first pages of MHF:
I continue to be troubled by the unsettling spectacle offered by an excess of memory

here, and an excess of forgetting elsewhere, to say nothing of the influence of com-
memorations and abuses of memory – and forgetting. The idea of a policy of the just
allotment of memory is in this respect one of my confessed civic themes.
(Ricoeur 2004: xv)
Furthermore, the question of the abuses of memory is also linked to the question of the duty
of memory, which Ricoeur answers through a three-step argumentation (Ricoeur 2004: 89).
First, the duty of memory, according to Ricoeur quoting Aristotle (see Chapter 31, this volume),
is related to the duty of justice; second, it is essential to open the concept of duty of memory
to the broader concept of debt: ‘The idea of debt is inseparable from the notion of heritage.
We are indebted to those who have gone before us for part of what we are’ (Ricoeur 2004:
89). The duty of memory is therefore not associated with an obligation of ‘conservation’ of
the past, but a recognition of obligation to those who have made us and made what we are.
The third element Ricoeur associates with the duty of memory (see Chapter 27, this volume)
is that of the need to take into account the concept of victim. If there is a duty of memory, it
should not only be in a more general context of generational inheritance, it should be linked
to the recognition of an individual or group of individuals. Ricoeur argues then that the
duty of memory also has the potential of being associated with the idea of forgiveness, but
forgiveness requires direct and individual recognition of the victim and of the offender. Such
process is the opposite of, for example, the amnesty process which imposes (if it is possible
to impose?) forgetting and forgiveness but also de-personifies (and de-humanises) forgiveness
(Ricoeur 2004: 452).
It should also be emphasised that the publication of MHF intervenes in a particular political
context in France. Since the Gayssot Act of 1990, which condemned among other things crimes
against humanity, France has developed a genuine ‘arsenal’ of laws by which the legislator has
tended to intervene incessantly in the historical debate.4 One can imagine that the reference
567
Ricoeur makes about a ‘fair memory politics’ in MHF is not independent from the repeated
intervention of the political and the judiciary in the historical and memorial debate in France
in this specific context.
To avoid being faced with the risk of ‘assignation’ or what he calls otherwise the abuses of
the manipulated memory (Ricoeur 2004: 81, 448), Ricoeur also strives to consider the repre-
sentation of the past through a dual prism, which includes a balance between an epistemological
approach to history and a phenomenological approach to memory. As posed by Ricoeur, one
of the risks of the representation of the past, is to consider memory and/or history as put ‘in the
service of the quest, the appeal, the demand for identity’ (Ricoeur 2004: 81).
In emphasising this, Ricoeur makes a real warning linked to the ‘risk’ that is at the back-
plane of a systematic opposition or disconnection between history and memory, or between
epistemology and phenomenology. Studying the representations of the past, without having
knowledge (epistemologically) about the past object/event that one wishes to problematise/
contextualise, could easily lead to what Bloch called a ‘moralism sin or anachronism sin’ (Bloch
1993: 176). But in the same way, the attempt of knowing a past object/event without prob-
lematising its production and perception, i.e. its relationship to temporality, or to the contextual
or time frame from which it is studied (phenomenologically) could easily raise some serious
methodological issues.
In Ricoeur, looking at the representations of the past includes the necessary consideration
of memory and history, epistemology and phenomenology as the two sides of the same coin,
each of them designing their own realities but at the same time also inseparable. One can find
in these Ricoeurian considerations an underlying Benjaminian posture. Walter Benjamin, in his
essay on the concept of history and his critique of historical materialism from 1940 reaffirms the
importance of thinking history as a dynamic field:
To articulate what is past, does not mean to recognize ‘how it really was.’ . . . For
historical materialism it is a question of holding fast to a picture of the past, just as if it
had unexpectedly thrust itself, in a moment of danger, on the historical subject.
(Benjamin 1969: 255)
Benjamin also underlines this ideological position in details in his work entitled The Origin
of German Tragic Drama published for the first time in 1928 where he reaffirms the need to
consider the source of the facts studied in their relation to ‘their history and their subsequent
development’ (Benjamin 2009: 46). In this debate between epistemology and phenomenol-
ogy, Ricoeur therefore seems to seek a balanced position. This led him, among other things,
to refer to the work of Marc Bloch who, in The Historian’s Craft (1993), compared the work
of the historian to that of an examining magistrate (and not of a court magistrate) and defined
history as ‘a science of traces’ (Ricoeur 2004: 13). These nuances are central as they are at
the heart of the differentiations that Ricoeur introduced between epistemology of history
and phenomenology of memory. He starts by stating a methodological similarity between the
historian and the judge: both have a responsibility to discuss the degree of falsifiability of the
gathered proofs. Are the material elements/traces true or not? What is the degree of truth-
fulness of the trace? It is also at this stage that the historical truth may actually be discussed
and verified, and that historical revisionism should be discussed and challenged. But once the
work on the trace is over, the historian’s relationship to the ‘truth’ no longer holds. Historical
discourse becomes part of a discursive and cultural canon and participates in what Assmann
calls the cultural memory (Assmann 2008). In other words, the ‘historical truth’ should be
seen in relation with the truth of the trace, but not the truth of the telling/narrative; that is
568
Paul Ricoeur
to say, in its relationship to the episteme. There is therefore a need to differentiate between
the relationships history maintains with ‘traces’ which corresponds to the epistemological
approach to the representation of the past, and the relationships history maintains with telling
which necessarily integrates a phenomenological and memorial perspective opening onto the
world of possibilities and potentialities.
For Ricoeur, then, the problematisation of the representation of the past acquires a soci-
etal importance and can be considered as a main societal challenge. The establishment in a
society of a fair politics of memory, and of a balance between the epistemology of history and
the phenomenology of memory, are linked to the question of the constitution of collective
identities. These issues are also related to the question of assignation of identity and therefore
to the question of cultural heritage as a factor of exclusion or of implication of the otherness.
4. Conclusion: the importance of the imagination

and the world of possibilities
As we have seen above, in Ricoeur, the constitution of narrative identity, like that of cultural
memory and history (or historiography), is elaborated through narration, discourse, or more
generally mediation. In this way, these issues carry an intrinsic creative potential, one could even
speak of a certain creative power.
The representation of the past in Ricoeur is not only considered in its relationships to the
question of the fictionalisation of history or the historicising of fiction, nor to the question of the
veracity of the trace, even if these elements, as we have seen, are an integral part of Ricoeurian
thought. The discourse (or the philosophy) of memory in Ricoeur is not an isolated or single
issue, but a central perspective in a wider topic which is the study of the representation(s) of the
past. This statement leads us to consider the past and the memory also as essential resources in
the reflection about the invention of the present and the future, following in this sense the reading
of Ricoeur by Hayden White:
It is my conviction that, like Heidegger, Ricoeur genuinely wanted to think ‘new’,

but not in the modernist sense in which the ‘new’ is identified with ‘the modern’
itself. He was prevented from doing so because to embrace the ‘new’ in the sense of
‘the modern’ would be to grant a presupposition of contemporary modernism that
demands a complete break with ‘the past’, with tradition, heritage, ancestry, the ori-
gin, the archaic, and everything else that presumes the relevance of ‘the past’ to the
understanding and valorisation of ‘present life’. But it was not to save the past either
that Ricoeur spent fifty years thinking about history.
(White 2007: 234)
There is in Ricoeur an idea of the appropriation of the past which is accompanied by an ambition
of identifying positively the ‘what should be done’. In this sense, the Ricoeurian reflection on
the representation of the past opens a new way for the multitude of possible uses of these repre-
sentations. One of the alternatives suggested by Ricoeur is to consider the past and the memory
in relation to the notion of the living tradition in Gadamer (Ricoeur 1998). This idea is also
reflected in MHF when Ricoeur, referring to Bergson (see Chapter 42, this volume), reaffirms
the creative potential of memory: ‘To memory that repeats is opposed memory that imagines’
(Ricoeur 2004: 25).
This creative power or potentiality has also to be associated with the importance of the
place occupied by the issue of the aesthetics in Ricoeur, and the necessary transfer from the
569
telling to the reading/reception of the representation of the past (Ricoeur 1988: 167). The
place of the reader, already central in Time and Narrative, is highlighted in the second part of
MHF dedicated to the question of history and epistemology, and especially in the last chapter
related to the issue of the représentance, a French neologism translated into English by the term
‘standing-for’ (Ricoeur 2004: 274). The idea of the standing-for in Ricoeur emphasises the fact
that the historical narrative as other narratives put the reader in a position of expectation: ‘The
historian’s representation is indeed a present image of an absent thing’ (Ricoeur 2004: 280). It
is this paradoxical and complex relationship of the historical narrative between the representa-
tion of an absent thing on the one hand, and the re-actualisation of this thing in the intimate
sphere of the reader/receiver on the other hand, that led Ricoeur to consider the limits of the
purely epistemological approach of the past. This is also what led him to continue his reflection
on the representation of the past through the question of the historical condition of the being where
memory and history meet again.
The idea of the historical condition of the being is what finally concludes the dialectical
relationship between phenomenology and epistemology in Ricoeurian thought: ‘We make
history, and we make histories . . . because we are historical’ (Ricoeur 2004: 284), that is
to say, the being is both in history, participating in the history; but also an actor of history,
necessarily creating history. The tangle of memory and history, of the epistemological and
phenomenological approaches to the representation of the past in Ricoeur leads to a neces-
sary ontological hermeneutics, where the last word belongs to the reader/receiver or what he
called in Time and Narrative III, the ‘condition of being-affected-by-the-past’ (Ricoeur 1988:
227). However, it should not be forgotten that the Ricoeurian reflection on memory and his-
tory also leads to the consideration of a necessarily critical hermeneutics already present in the
conclusion of Time and Narrative III: ‘The mystery of time is not equivalent to a prohibition
directed against language. Rather it gives rise to the exigence to think more and to speak dif-
ferently’ (Ricoeur 1988: 274). This position is also reaffirmed in his later work, since the last
element of the Ricoeurian triptych in MHF focuses on the potential problem of forgetting. In
Ricoeur, forgetting is not only considered as a problematic issue linked to the notion of dele-
tion. In proposing the term ‘forgetting as a reserve’ (Ricoeur 2004: 506), he points out that
forgetting should also be considered ‘as a temporary stage, a suspension of activation of a word
or thought which has a latent existence’ (Dessingué and Winter 2016: 4). This consideration of
forgetting as ‘silence’ and not only as deletion strengthens the need of critical awareness which
is required both from the perspective of the one who represents the past but also for the one
who receives it.
Notes
1 My translation.
2 Published for the first time in English in 1956.
3 My translation.
4 The Taubira Act of 2001 recognises slavery as crime against humanity, but also encouraged address-
ing the issue in school curricula. The Gayssot Act of 1990 was an attempt to strengthen the judiciary
possibilities/solutions against racial hatred or discrimination issues but did not integrate the question
of the curriculum or education. Subsequently, however, a new advancement was made with the law
called ‘Mekachera’ from 2005 that encouraged, before being amended, to recognise the positive role
of the French colonial presence in overseas territories. This law has provoked a strong reaction among
many intellectuals, as it crossed a new level, this time by linking the law to a judgment of value on
the past recognising a positive aspect to a given historical and political period. Finally, the last debate
was related to the law which recognised the Armenian Genocide but was at the end censored by the
Constitutional Council in February 2012.
570
Paul Ricoeur
Related topics
• A duty to remember
• Aristotle
• Augustine
• Henri Bergson
References
Assmann, J. (2008) ‘Communicative and Cultural Memory’, in Erll, A. and Nünning, A. (eds.), Cultural
Memory Studies. Berlin: Walter de Gruyter.
Augustine. (2009) Confessions, trans. Chadwick, H. Oxford: Oxford University Press.
Benjamin, W. (1969) ‘Theses on the Philosophy of History’, in Illuminations: Essays and Reflections. New
York: Schocken Books, pp. 253–65.
Benjamin, W. (2009) The Origin of German Tragic Drama. London and New York: Verso.
Bloch, M. (1993) Apologie pour l’histoire ou métier d’historien. Paris: Armand Colin (1st edn, 1949).
Dessingué, A. and Ryckebusch, O. (2011) Dunkirk – City of Memories. Stavanger: Hertervig Akademisk.
Dessingué, A. and Winter, J. (2016) Beyond Memory: Silence and the Aesthetics of Remembrance. New York
and London: Routledge.
Husserl, E. (1997) The Psychological and Transcendental Phenomenology and the Confrontation with Heidegger
(1927–1931), trans. Sheehan, T. and Palmer, R.E. Dordrecht: Kluwer Academic Publishers.
Levinas, E. (1990) Totalité et infini: essai sur l’extériorité. Paris: Le livre de poche.
Ricoeur, P. (1988) Time and Narrative, vol. 3, trans. Blamey, K. and Pellauer, D. Chicago, IL: University
of Chicago Press.
Ricoeur, P. (1992) Oneself as Another, trans Blamey, K. Chicago, IL: University of Chicago Press.
Ricoeur, P. (1998) ‘La marque du passé’, Revue de métaphysique et de morale, vol. 1, 1–25.
Ricoeur, P. (2004) Memory, History and Forgetting, trans. Blamey, K. and Pellauer, D. Chicago, IL:
Rimbaud, A. (2007) ‘Les lettres dites du voyant’, in Poésies. Paris: Hatier.
Sartre, J. (1976) L’être et le néant. Paris: Gallimard.
White, H. (2007) ‘“Guilty of History?” The longue durée of Paul Ricoeur’, Journal of History and Theory,
46(2): 233–51.
571
INDEX
Note: The following abbreviations have been used – f = figure; n = note; t = table
a posteriori (empirical) knowledge 388 Analogy of Religion, The (Butler) 408

a priori (conceptual) knowledge 388, 389, 397, Analysis of Mind, The (Russell) 187–8, 519, 523,
405–6 524, 526n13–14
A-theory of time 219, 220, 221, 226 analytic approach: induction 58
Abel, T. and colleagues (1997) 45 anamnêsis 387, 396, 397, 399, 403, 533
Abhidharma tradition see Indian Buddhist Angere, S. 320–1
philosophy anima/animus 442, 445
Abhidharmakośa-bhāsya 419 animal cognition see non-human animals
ability hypothesis 216n11 animalism 172
absolute standards 391, 395n7 Annales d’histoire économique et sociale 529
Access Internalism 285–6 Annales group (Strasbourg) 529
accuracy 51, 52, 55, 56, 58–60, 85 Annas, J. 406n4
acquaintance 475–6, 519, 520–3, 526n4;9;12 anoetic consciousness 12f, 155, 308n3
acquired intellect (understanding) 465–6, 467 Anschauung (intuition) 488
activity of memory 541 Anscombe, G.E.M. 61, 80, 210, 554
Addis, D.R. see Schacter, D.L. and colleagues anterograde amnesia 37, 176, 422
(2007) anti-realism 211–12, 216n8, 221–2, 222–3
affirmation 159–60, 162 anxious-ambivalent attachment 274
agency 28, 31n31, 505; duty to remember 357–8, appearances 299
360, 361; forgetting 364, 367–9; intellectual apprehension 158, 423, 462, 475, 478
memory 462, 465, 466; memory modification appropriation of memories 195, 202n4, 569
technologies (MMTs) 377, 378; no self theory Arango-Muñoz, S. 250
418; unconscious memory 496 archival view 51, 54, 56, 60n1
Aizawa, K. 35 Aristotle 79, 85n7, 155, 397–9, 406n4, 511;
alaya consciousness see storehouse consciousness Averroes and 458, 459n32–3; Avicenna (Ibn
Alciphron: or, the Minute Philosopher (Berkeley) 477 Sina) 466–7; collective memory 531; duty to
allocentric spatial information 119, 121 remember 355–6; imagination 127, 128, 132,
Alzheimer’s disease 55, 174, 248, 373 133, 136, 457; internal senses 448; philosophy
amnesia 13, 37–8, 39, 82, 137; anterograde 37, of mind and perception 399–403; Plato and
176, 422; childhood 195, 196; consciousness 396–7; recollection 403–6, 407n8, 533;
103, 106–7, 110, 111, 181; personal identity Thomas Aquinas 461, 462, 464, 465, 466;
173, 174, 176, 177; retrograde 176, 421–2; see also Epicureans; Plato; Stoics
temporality 212, 214, 215 arrow of time 223–5
amytotrophic lateral sclerosis (ALS) 174 ‘as if’ language 23, 27
572
Index
Asa ga (Shiqin 世亲) 423–4, 429 305, 306; Hume 481, 525; intentionality 88–9,
Assmann, A. 258 90–1, 91–2, 93–4, 95; memory as storage unit
Assmann, J. 258, 565, 568 342; metaphysics of time 221, 222, 223; Russell
association of ideas 481, 484n2, 516 524; see also justification of memory belief
associative memory 36 Bell, G. 245
Atkinson, R.C. 10, 11 Bender, D.B. see Gross, C.G. and colleagues
atrocities 359, 360 (1972)
attachment theory 268, 274–7 Benjamin, W. 255, 257, 510, 568
attention 413–15; emotion 155–6; Hegel 488–9; benzodiazepines 374
imagery 150, 151; Locke 472, 475 Bergmann, M. 289, 340, 346n15
Atthasālinī (The Fount of Meaning) (Buddhaghosa) Bergson, Henri 529, 533, 538, 569; theory of
412–13 memory 510–14, 515f, 516, 517n2–4;15
attitude 22, 27–30 Bergsonism (Deleuze) 511
attributionalism 236 Berkeley, G. 477
attunement 557 Bermúdez, J.L. 183, 186
Aufmerksamkeit (attention) 488 Bernecker, S. 71, 82, 124n4, 190n, 226, 291;
Augustine 133, 136, 439–41, 450, 458n9, 467; perspective 116, 117–18
conceptualization of memory 441–5, 446n2; Bernet, R. 517n12
temporality 566; see also Thomas Aquinas Bernstein, D.M. 54, 61n4
Austin, J.L. 552 Berrios, G.E. 175
authenticity 52, 54, 56–8; perspective and Beyond Good and Evil (Nietzsche) 257, 265n2
115–19 Bickle, J. 34
autobiographical memory 121–2, 230; bio-functional approach 542
coherentism 318, 319; foundationalism 296, biological memory 243, 246, 248–9, 252, 253n2;
297–8, 307; Halbwachs 531, 533; narrativity complementarity 244, 245
187–90, 199, 200; self-consciousness 180, Blendy, J. see Bourtchuladze, R. and colleagues
181–9, 190n9; social identity 270, 271; social (1994)
interactionist theory of 193–6; Wittgenstein Bliss, T. 42
550–1 Bloch, M. 568
autonoetic consciousness 12f, 13, 16, 104, 155; Block, N. 141, 143, 144–5, 146
episodic memory 421, 422–3, 424, 425; mental Blue Book (Wittgenstein) 182–3, 186
time travel (MTT) 232–3, 235, 409, 410, 411, Boas, F. 538, 539
414–15; see also consciousness body (sôma) 261, 398, 540
autonomy 154, 361 Bonjour, L. 284, 314
Averroes 448, 454–8, 459n32, 468n11 Book of Letters (Averroes) 459n32
Aviary imagery (Plato) 393–4, 395n8, 397 borrowed memory 257
Avicenna (Ibn Sina) 448, 459n19–20; recollection Bossis, A.P. see Landa, A. et al. (2012)
451–4; retention 449, 450–1, 454, 458n7; boundary extension 57
Thomas Aquinas 463, 465, 466, 467n1, Bourtchouladze, R. and colleagues (1994) 44–5;
468n11 see also Abel, T. and colleagues (1997)
Awakening of Faith in Mahayana (大乘起信论) Bovens, L. 320
435, 438n9 Bowlby, J. 274–5
Azeri, S. 485n9 Boyd, R. 14
Boyer, P. 107
B-theory of time 220, 221, 223 Boylan, L.S. see Landa, A. et al. (2012)
Baier, A. 212 Brandt, R. 59
Barad, M. see Abel, T. and colleagues (1997) Brecht, B. 519
Barrès, M. 262 Buckner, R.L. see Schacter, D.L. and colleagues
Barsalou, L. 414 (2007)
Bartlett, Frederic 116, 156, 230, 537–44 Buddhaghosa 412–15
basic perceiving mind (prôton aisthêtikon) 400 Buddhism see Chinese Buddhist philosophy;
Bechtel, W. 34 Classic Indian philosophy; Indian Buddhist
Bedford, F.L. 14 Philosophy
Beike, D. 333n9 Burke, P. 258
Being and Time (Heidegger) 557, 559, 561 Burnyeat, M. 401
belief 22, 392, coherentism 310–11; false beliefs Bussey, T.J. see Clayton, N.S. and colleagues
392, 393, 394; foundationalism 298, 299, 304, Butler, Bishop J. 170, 176, 183, 184, 408, 477
573
Index
Butler, S. 256 Classical Indian philosophy: episodic memory as

Byrne, A. 24, 29 attention 411–15; memory and self 408–11;
see also Chinese Buddhist philosophy; Indian
Ca++ influx 43 Buddhist Philosophy
Cacioppo, J. 31n34 Clayton, N.S. 108; and colleagues 12 (2003)
calmodulin kinase II protein (CaMKII) 44, 45 cloud technology 243, 244, 251
cAMP see cyclic adenosine monophosphate ‘cogitative’ (‘thinking’) faculty (fikr) 449
molecules cognition 22, 37, 106, 449; content and 30n5;6;
Campbell, G. 477 episodic memory 421; extended memory 243;
Campbell, J. 216, 222, 226n6 Hegel 487; non-human animals 12, 13; Plato
Campbell, K. 31n27 388–9; qualities 335; Russell 521
Campbell, S. 200, 202n12, 212, 215 cognitive enhancement/integration 251–2
Capgras, J. 157, 158, 211–12 cognitive penetration 306–7
Cartesianism 72, 73, 74 Cognitive Self Esteem (CSE) 247
Cassel, D. see Cassel, J.C. and colleagues (2013); coherentism 310–13, 321n1;8, 324; impossibility
Manning, L. and colleagues (2013) theorems 320–1; independence and partial
Cassel, J.C.and colleagues (2013) 440; see also reliability 315–18; psychological realism
Manning, L. and colleagues (2013) 318–19; theories 189–90; truth conduciveness
Cassirer, E. 265n3 313–14
causal chains 67–9, 70, 81 coincidence detectors 43
causal condition see memory causation Cole, M. 543
causal groundedness 15 collaborative recall 246–7
causal theory of memory 65–7, 68, 117, 224–5; collective determinism 258
causal gaps 80–1; Hume 480–3 484n1–2; collective memory 77, 255–7, 265n1, 559; as a
mental time travel 233, 236; opposition to category of contemporary social analysis 259;
71–2, 73; Wittgenstein 552–4 Halbwachs 255, 257–9, 261, 262, 528–34,
ceteris paribus clause 317, 318, 320 535n3;5;8;12;14–16;18–19; historical past and
chains-of-stimuli-response 38 261–2; mass media 262–5; social cohesion and
Chalmers, D. 248, 250 257–9; symbolic embodiment of 259–61
Chan School of Buddhism (禅) 429, 434, 435, Collective Memory (Halbwachs) 528, 532,
436, 437 533, 542
change blindness see fragile visual short-term ‘Collective Memory of Musicians, The’ (‘La
memory (fragile VSTM) mémoire collective chez les musiciens’)
change detection paradigm 149 (Halbwachs) 528
character traits 173–4, 176–7, 270 collective recollection 534
Charmides (Plato) 387 Collingridge, G.L. and colleagues (1983) 43
Cheng, S. 97–8n4; and colleagues 16 Commentary on Aristotle’s De anima (Thomas
Child, W. 72 Aquinas) 461
Chinese Buddhist philosophy 428–9, 437n1–2; Commentary on Aristotle’s Metaphysics (Thomas
Buddha nature and memory self 434–5; human Aquinas) 461, 462, 467n4
self and logical status of memory 429–33; non- Commentary on Aristotle’s On Memory and
self/memory-self 433–4; transcending memory Recollection (Thomas Aquinas) 461, 462,
435–7; see also Classic Indian philosophy; Indian 467–8n7
Buddhist Philosophy Commentary on the Posterior Analytics (Thomas
Chisholm, R. 299 Aquinas) 464, 465
Choi, H. 145–6 ‘common sense’ 481, 512; Avicenna 449, 450,
Christianity see Augustine; Thomas Aquinas 451; Thomas Aquinas 462, 463
Chudnoff, E. 299, 300, 301, 302 communicative memory 258
Chun, M.M. 413 complementarity 244–6
Churchland, P. 34 compositionality 18n1
cinematic point of view 120 compositive imagination (al-mutakhayyilah) 449
Cioffi, D. see Bourtchuladze, R. and colleagues comprehended history 491–2
(1994) conceptualism: content 30n8; holism 550;
circularity: personal identity 170–1, 176; self- interdependence 550; self-representation 409
consciousness 183–4, 185; skepticism 335, 337, conditioning 35, 36
338, 339–40, 345n14 cone of memory (Bergson) 515f
Clark, A. 243, 248, 249, 250, 253n2 Conee, E. 286, 290, 304–5
574
Index
Confessions (Augustine) 439, 440, 441, 444, cyclic adenosine monophosphate molecules
445, 566 (cAMP) 41–2, 45
confident memory beliefs 60, 85 Cyclic AMP-responsive element-binding-protein
connectionism 555 (CREB) 44, 45
consciousness 12f, 13, 98n6, 111, 285; Bergson
512–13, 514, 516; Chinese Buddhist Dacheng Qixinlun (大乘起信论) 435, 436
philosophy 429, 430, 431–3, 434, 437; defining Dante 161
memory 103–4; episodic memory 104–8; Daoism 433
Freud 496, 497, 498, 500; Global Workspace Darden, L. 35, 36f, 37–8, 43
theories 141; Hegel 490–1, 492–3; iconic DARPA see Defense Advanced Research Projects
memory 141–2, 148–9; imagery 151, 152; Agency (United States)
Indian Buddhist philosophy 423–5, 426n8; Dasein 557, 559, 560
Locke 476; ‘movie-screen’ model 147; non- Data Protection Directive 95/46/EC (European
human animal memory 103, 108–10; see also Commission) (1995) 369–71
autonoetic consciousness daydreaming 198, 472, 473
consequentialism 358–60, 360–1 de Anima (Aristotle) 398, 457, 466
Constitutive Dependency Thesis 213 De Brigard, F. 51, 124n5, 138; narrativity 198,
Constructive Mind: Frederic Bartlett’s psychology in 199–200, 202n8
reconstruction, The (Wagoner) 537 De sensu et sensato (Aristotle) 461
constructive remembering 544 Dean, John 55
constructivism 116, 137, 273; preservation 326, Debus, D. 70, 75n9, 117, 160
327; truth 51, 54, 56, 57 decay 10, 134, 144, 147, 150, 483
contemplation 472, 473, 474, 513 declarative memories 37, 38, 77, 172, 181;
content 22–7, 30n2;5;6, 173–4, 224; emotion 158, 163; episodic memory 9, 10, 11,
autobiographical 195–6, 201; consciousness 13–14; narrativity 192, 193; semantic memory
105–6; imagination 127, 128, 132; truth 9, 11; skepticism 338, 339
54, 56–7 deep brain recordings 39
content-based taxonomy 12 deep brain stimulation 374
content identity 56, 61n8 Default Mode Network (DMN) 198
context-shock associative memory 36, 40 Defense Advanced Research Projects Agency
contextual conditioning 35, 36, 40 (DARPA) (United States) 374
contextual information 109 definitio per genus proximum et differentia specifica 9
contextualism 345 Deisseroth, K. see Liu, X. and colleagues (2012)
contiguity 81 déjà vu 498, 501, 511
continuism 228–36, 237n1, 445, 517, 525 Deleuze, G. 511, 517n7;10
continuous memory trace 68 denoting 520
contraction memory 511, 512, 517n7 dentate gyrus 39, 40, 42
conventionalization 538 deontology 358, 360, 361
Conway, M.A. 318, 319 Dependency Thesis 207, 208–13, 215, 216n2;5;8
Conway, M.C. 61n4 dependency view 25, 26
cooperativity 42 depersonalization disorder 175
copy theory of ideas 60n1, 481, 483, 485n4–6, depression 200, 230
524 depth perception 302
Corballis, M.C. 19 Descartes 133, 136, 284–5, 389, 473
Craver, C.F. 212–13; and colleagues (2014) 107, description method 539
214, 215; science 34, 35, 36f, 37–8, 43 description-independent representation 177
Creative Evolution (Bergson) 511, 538 descriptionalism 57
creative visualizing 400 Dessingué, A. 565
creativity 197, 202n7, 456–7 determinable/determinate properties 25
CREB see Cyclic AMP-responsive determinism 222, 223, 258
element-binding-protein Deuel, T.A. see Abel, T. and colleagues (1997)
credibility 310–11, 315 Deutscher, M. 31n31; memory causation 65–7, 68,
Critique of Pure Reason (Kant) 540 69, 70, 74, 75n10; memory traces 78, 82, 83
Croce, B. 256 Developmental Dependency Thesis 213, 216
Crombag, H.F.M. and colleagues (1996) 55 deviant causal chains 67–9, 70, 81
cultural memory 258, 270–1 Dhammasaṅgaṇi 411
cybernetics 543 diachronic approaches 323–5
575
Index
dialectical memory 493 empiricist theory of memory 208, 209, 216n4;5

Dickinson, A. 108 emptiness (sunyata/kong 空) 428–9, 430, 434, 435
Dickinson, A. see Clayton, N.S. and colleagues encoding 15, 51, 83
(2003) Encoding-Storage-Retrieval (E-S-R model) 83,
‘Dictionary of Misunderstood Words’ 84, 85n16
(Kundera) 156 Encyclopedia of the Philosophical Sciences (Hegel)
diffusionist anthropology 538, 539 487, 492
Digha Nikaya 412 endorsement 248, 250
digital memory 245, 262–5 engrams 77, 84, 85n5
Dilthey, W. 256 enhanced memory see memory modification
direct knowledge 551–2 technologies (MMTs)
direct realism 234, 470, 519, 524, 551 enlightenment (wu 悟) 428, 430, 432, 433–4, 436
discontinuism 228–36, 237n1; memory trace 68 Enneads (Plotinus) 441
disjunctivism 234, 287nt Enquiry concerning Human Understanding (Hume)
Diski, J. 113, 114, 115, 117 484n2
dispositional preservationism 305 enrichment 23–4
Disputed Questions on the Soul entorhinal cortex 39
(Thomas Aquinas) 461 entropy 224
distorted memories 118, 124n5, 137, 198, 202n8, Epicureanism 356–8; see also Aristotle; Plato; Stoics
442–3, 540 epilepsy 40
Distributed Cognition 243 episodic memory 21, 77, 155, 228, 473; Classical
divine illumination 443–4 Indian philosophy 410, 411–15; consciousness
Doctrinal Treatises (Augustine) 445 104–8, 110; emotion 155, 156, 158, 159,
‘Doctrine of One Mind with Two Functions’ 435, 161; foundationalism 296; Freud 498, 501; as
438n9 a generative capacity 208–10; grasp of time
Dokic, J. 156, 414 as a domain 210–15; imagination 137; Indian
Donald, M. 243, 244, 249, 252 Buddhist philosophy 421–6; intentionality
Dostrovsky, J. 38, 39 88, 89–90, 91–6, 97n1, 97–8n4, 98n8;
dreams 472–3, 482, 498–9, 500t, 501, 513 memory modification technologies (MMT)
Dretske, F. 98n6, 141, 143 378–9; mental time travel 229, 230, 231, 232,
Duerlinger, J. 419, 420, 426n3 233–4, 235, 236; metaphysics of time 221, 224;
Dummett, M. 211, 216n8 narrativity 196–9, 201; personal identity 172,
Durkheim, E. 528, 529 173, 174, 175–6, 177; perspective 115, 119;
duty of memory 567 preservation 326; Reid 477–8; Russell 521;
duty to remember 351–4; classical antiquity self-consciousness 181, 186, 188–9, 190n9; self-
philosophy and 354–8; modern ethical theories implication 409; simulation theory of 197, 198,
358–61; see also remembering 199, 233, 234; taxonomy 9, 10, 11, 12f, 13–15,
dynamic memory traces 78, 85n6 16, 17; temporality 13, 15, 105, 106, 108, 207,
dynamic model of the soul 496 212, 215, 216n1;3;7;8;10; see also semantic
memory
E-memory see extended memory epistemic activity 364–6
easy identity 170, 171, 172 epistemic badness 305–7
Ebbinghaus, H. 537 ‘epistemic boost’ problem 291
ecstases: temporality 557–8 epistemic elitism 301, 302
ecstasy 472 epistemic feelings 156–7, 225–6, 249, 250, 251
Edgerank algorithm (Facebook) 251 epistemic justification 287, 288, 289–92, 293n1
effortless action 21 epistemic seemings 302
Ego 496 epistemology: episodic memory and 91–7; of
ego-consciousness (kliśṭa-manas) 423, 424–5, 426n9 history 567–9; internalism/externalism 60,
egocentric spatial information 119, 121, 122, 123 283–6, 287t, 288–9; justification of memory
eikon 132 belief 289–92
elaborative reminiscing 271, 272 Epitome of the Parva naturalia (Averroes) 454,
emotion 77, 107, 154, 163, 376; Augustine 444; 456, 457
forgetting 159–60; norms of regret 160–2; EPSPs see excitatory post-synaptic potentials
personal/subpersonal 155–9; perspective 118–19, Epstein, R. 553
120, 122, 124n10 Erinnerung (recollection) 487, 488, 493, 494n4
576
Index
errors 540; from omission/commission 84, 124n3; externalists 136

mental time travel 231, 234, 237n4; Russell eye consciousness (caksurvijñāna) 424
521–2, 524, 525; truth 55–6, 61n6 eyewitness testimony 190, 259
Essay Concerning Human Understanding, An (Locke)
134, 471 Facang (法藏) 435
Essays on the Intellectual Powers of Man (Reid) 408, face recognition 157
476, 477 factive verbs 52, 53, 129–30
estimative faculty (wahm) 449, 451, 452, 453, 456; factivity 8, 15, 52, 160; imagination and 128,
Thomas Aquinas 463, 464, 465, 467 129–30; mind-dependent/independent 220,
Eternal Return thought experiment 159 226n2; perspective and 114, 115–16, 118
Eternal Sunshine of the Spotless Mind, The (film) factual memories 193, 550; memory causation
159–60, 161 63–4, 65–6, 67, 70, 74n5
ethics 356–8, 363n2, 373 fair memory politics 568
European Commission (EC) 369–71 false memories 16, 40, 190, 457; Augustine 442–3;
evaluative properties 29 belief and 392, 393, 394; consciousness and
Evans, G. 182, 185, 189, 206, 209 104, 110; imagination and 127, 130; memory
event-memory 123n2 modification technologies 374–5, 376, 377–8,
events 15, 17, 27, 70; assessing accuracy 51, 379; memory traces 80, 84–5; preservation 326;
52, 55, 56, 58–60; highly distressing 276; personal identity 170, 175–6; perspective and
imagination 130; intentionality 90, 97–8n4; 113–14, 115, 118, 124n3;5; Russell 521–2,
mental time travel (MTT) 230, 232; negative 524, 525; truth and 52, 53, 54–5, 56, 59–60,
experiences 107; publicly significant 259; what- 61n4;6, 64
where-when criterion of events 12 familial order 506–7
evidence-insensitive seemings 302, 303–4 familiarity 28, 29–30, 31n34, 78, 208, 305; feelings
Evidentialism 290 156, 157; Locke 475
ex ante justification 331, 334n18 fear conditioning 36, 40
excitatory post-synaptic potentials (EPSPs) 41, 42 fear-related memories 374, 453
expectation/opinion (elpis/doxa) 399 feelings 28–30, 31n32;34, 157
experience 8, 63, 70, 89, 118–19, 172; emotional Feldman, R. 286, 290, 304–5
29; foundationalism 297; Hegel 490–1; Locke Fernández, J. 105, 124n5
473; non-human animals 453; pastness 23, field and observer perspectives 114–15, 124n10
24, 25, 103, 104–5, 110; perceptual/memory first-order desires 378
beliefs 343; postdictive 145–6; quasi-transparent first-person perspective 53, 77, 114–15, 155,
simulation 16; social identity 269–70; source 283, 284
monitoring errors 55; subjective experience 13, Fisher, R.P. see Vrij, A. and colleagues (2010)
14; temporality 220, 221; Thomas Aquinas 464; fission 184
unpleasant memories 159–60; Wittgenstein 548 Fitzgibbon, O. see Hayne, H. and colleagues
experiential basis 186 (2011)
experiential contents 22, 24–7, 29, 30n7;8;12, five senses see external senses
31n15;19;21 Fivush, R. 194, 196, 200, 201n4, 270, 276
experientially awareness 232 Flage, D. 485n6
explanatory irrelevance objection 80 ‘f lashback’ memories 157, 158
explicit memory 11, 37, 38, 158, 181 FMTT see future-oriented mental time travel
explicitly recollective memories 182, 186 folk psychology 252
extended memory 243–4; cognitive enhancement/ forgetting 82, 159–60, 358, 504, 570; agency
integration 251–2; complementarity and and sensitive information 367–9, 371n2–3;
practical incorporation 244–6; extended mind Augustine 444–5, 446; Bergson 511–13;
(EM) 247–51; forgetting sensitive information Halbwachs 532–4; Heidegger 558–9, 561;
367–9, 371n2–3; memory modification human rights and 369–71, 371n4–6; morally
technologies (MMT) 373–9; transactive obligated epistemic activities 364–9; realism
memory (TM) 246–7 and 366–7; see also memory modification
extended mind (EM) 244, 247–51 technologies (MMTs)
external memory 243, 244, 249 forgiveness 567
external senses 449, 450, 462–3, 465 forms theory: Aristotle 397–8, 399, 406;
externalism see internalism/externalism Plato 390, 391, 396n6; Thomas Aquinas
externalist reliabilism 339 465–6, 467
577
Index
foundationalism 296–7, 308n1; evidence resistance Greco-Roman philosophy see Aristotle;

of phenomenal seemings 301, 302f, 303f, Epicureans; Plato; Stoics
304; memory seemings as justifiers 299, 300f, Green, S. 44
301; metacognition and problem of forgetting Greenwald, A.G. 318, 321n9
304–5; phenomenal dogmatism 297–9; transfer Gregory, D. 116
of epistemic badness 305–7 Grice, H.P. 170
fragile visual short-term memory (fragile VSTM) grid cells 39
141, 142, 147, 148f, 149–51, 152 grief 356, 357
fragility of identity 565 Gross, C.G. and colleagues (1972) 40
François, A. 510 Gross, J. see Hayne, H. and colleagues (2011)
Frenguelli, B. see Bourtchuladze, R. and group mind 246
colleagues (1994) Guzman, D.A. 421–2
Freud, Sigmund 156, 158, 496–9, 500t, 501,
507n7–9;11; see also Lacan, Jacques habit memories 74n3, 533, 550; Bergson 511, 514,
Friedman, L. 485n6 515
friendship 356, 357, 362n6 Haeckel, E. 256
Frijda, N. 29–30, 154, 155 Haggard, M. 34
‘From the History of an Infantile Neurosis’ Halbwachs, M. 542, 543; collective memory 255,
(Freud) 499–500 257–9, 261, 262; theory of memory 528–34,
frontotemporal dementia 174 535n3;5;8;12;14–16;18–19
Fumerton, R. 288, 340 hallucinations 234, 498
functional neuroimaging 39 Hamilton, A. 54–5
functionality 180 happiness 356
Furlong, E.J. 136 Hard, B.M. 123
future thinking 137, 161 harm to self/others 375–7, 378, 379, 380n35
future-oriented mental time travel (FMTT) 228, Harrod, R. 337
229, 230, 231, 232, 233–4, 235, 237n1 Hartmann, S. 320
Hasse, D.N. 458n2
Galloway, B. 425 Hayne, H. and colleagues (2011) 174
Ganeri, J. 420 Hazlett, A. 52, 60n2, 129
Garcia-Marquez, T. 31n34 Head, H. 540–1
Gayssot Act (1990) 567–8, 570n4 head orientation 39
Geary, P.J. 535n14 head trauma 175, 214, 422–3, 540
Gedächtnis (memory) 487 Hebb, D. 42, 43
gender-typed identities 273 Hegel, G.W.F. 256, 257, 487–93, 494n2;4;6;8, 501
gene intervention techniques 45 Heidegger, Martin 501, 505, 507, 557–62
gene targeting 44 Henri, V. and C. 121–2, 124n9
generativism 298, 323; preservationism 326, 327, Hering, E. 256
328, 329–32, 333n6 heritage 559–61
genetic modification 374 hermeneutics of the self 564
Gerrans, P. 198, 202n11 hierarchical taxonomy 9, 11, 12f, 14
Gilby, T. 395n6 hieroglyphs 7
Gillespie, A. 542 Hilts, P. 176
gist processing 14, 15, 55, 145 hippocampus 13, 16, 17, 84, 106; memory
Glass, D.H. 313 research 37, 38–41, 44, 45
global amnesia 37 hippocampus generating spatial maps 38–41
Goldie, P. 116–17, 122 Historian’s Craft, The (Bloch) 568
Goldman, A. 288 historical memory 559–60, 568, 570; collective
Goodman, N. 265n3 memory 257, 258; Halbwachs 531, 535n12;14
Gounden, Y. see Nicolas, S. et al. (2013) history of philosophy
Graci, M.E. 276 Hobbes 134, 136, 137
Graf, P. 136 Hoerl, C. 27, 28, 193
grammatical illusion 546, 552 Hölderlin, F. 561–2
Granhag, P.A. see Vrij, A. and colleagues (2010) holism 550
Gray, G. 543 Holland, R.F. 24, 135
Greco, J. 289 Hollingworth, A. 144, 150
578
Index
homeostatic property cluster view 14 Hume 480–3; memory indicator 27, 28, 29,
hope 22 31n28;29;31; memory traces 78; mental time
Huayan School of Buddhism (华严) 435–6, 437 travel (MTT) 230–1, 233, 234; phantasia
Huemer, M. 292 399, 400, 401, 402, 448; phenomenology of
Huineng (Œd) 429, 430, 434, 435, 436, 437 memory 21, 24, 26, 31n20;24; science and
Human Knowledge: its Scope and Limits (Russell) 136–8; taxonomy 13, 16; Thomas Aquinas
526n15 463–4, 468n8
human rights 359, 360; forgetting 369–71 immediacy 79–80, 92, 93–4
human self 429–33 immortality of the soul 389, 390
humanism 546–7 immunity property 181, 182–7
Hume, David 24, 31n16;17, 160, 227n11, 418, impermanence 424, 428, 474
470; acquaintance 522; imagery 524, 525; implicit memory 11, 14, 38, 181
imagination 134–5, 136, 137; memory traces impossibility theorems 320–1
78, 79, 85n8; temporality 208; theory of Incoherence of the Incoherence (Averroes) 454
memory 480–3, 484n1–2, 485n3–9; truth 51, incorporability 244–6
58; see also Locke, John; Reid, Thomas independent memory 315–17, 319
Hurka, T. 162 Indian Buddhist philosophy 426n1, 428–9, 431,
Husserl, E. 558, 564, 566 episodic memories 421–6; no-self theory
Hutchins, E. 243 (anātmavāda) 416, 417–21; see also Chinese
Hutto, D.D. 269 Buddhist philosophy; Classical Indian
hylomorphism 397–8, 467, 468n11 philosophy
hypnotism 64, 66, 156 indirect memory 8
Hyppolite, J. 512 individual memory 530, 531–2
hysteria 505, 508n41 induction 58
inequality: memory modification technologies
I-consciousness (mana 我识) 431–2 (MMTs) 376, 379
I-thoughts 177 inferential memory 57, 118, 124n4, 157, 220, 522;
Ibn Sina see Avicenna preservation 324, 328–9, 333n3; skepticism
iconic memory 141, 142, 143f, 144, 145, 147, 337, 338
148–9, 152 information: mis-match in processing 142, 146–7;
Id 496, 499 preservation of 76; retention 65, 70–1, 77, 78,
ideal theory 477 82, 83; retrieval 15, 16, 51, 57–8, 83, 137; self-/
identification agnosia 157 other-generated 136; sensitivity and forgetting
identification judgments 185 364, 367–9, 371n2; temporality 209, 263, 264
‘identity theory of memory’ 56, 60n1, 61n8, 565 informational persistence 143, 144
imagery 141–2, 263, 463; Augustine 439, 441, inhibitory post-synaptic potentials (IPSPs) 41, 42
442–4, 446; Bartlett 541; episodic memory intellectual inquiry 386–9
106, 111, 156; fragile visual short-term memory intellectual memory (memory per accidens) 461,
(fragile VSTM) 147, 148f, 149–51, 152; 462, 463, 465–6, 467n3
Freud 498, 507n8; Hume 524, 525; iconic intellectual prudence 467n5
memory 141, 142, 143f, 144, 145, 147, 148–9; intelligence 488
imagination 132; intentionality 89, 96; Lacan intelligent soul (intelligent memory) 441–2,
502f; Locke and ‘painting’ metaphor 474–5; 453–4, 457–8
memory traces 77, 78, 79, 80, 85n8; mis-match intentionality 88–91, 538; Averroes 455, 456, 457;
problem 142, 146–7; perspective and 113–14, Avicenna 448, 449, 450, 451, 452–3, 458n2;10,
120, 121, 123; phenomenology of memory 23, 467; episodic memory and 91–6, 97n2, 97–8n4;
24–5, 30n9;10, 144; postdictive experiences Locke 472; Thomas Aquinas 463, 467–8n7
145–6; Russell 521, 524–5; self-consciousness internal senses system 448, 449–50, 454, 457,
187–8; temporality 213, 216n11; truth 57–8, 462–3, 463–5
69–70; Wax Tablet 392–3; Wittgenstein internal working model (IWM) 274–5, 276
547–52; working memory 141–2, 147–8, 149, internalism/externalism 283–6, 287t, 288–9;
150–1, 152 justification of memory belief 289–92, 293n4
Imaginary order 502f, 505 internalists 136
imagination: Averroes 456–7; Avicenna 450–1; Internet 243–4, 247, 251, 263, 264, 310; forgetting
collective memory 259; content 199, 202n11; sensitive information 367–9, 371n2–3;
distinction between memory and 127–36; responsibility to remember 377
579
Index
Interpretation of Dreams (Freud) 499–500 Larkin, P. 154

intimacy 78, 158 Lashley, K. 84
introspection 91, 92, 97, 328, 329; Augustine 439, Leal, S. see Vrij, A. and colleagues (2010)
440, 441, 444–5, 446; Halbwachs 529–30 learning 156, 473, 516; Averroes 454, 458;
intuition 328, 329, 488, 489–90 forgetting 364–6; memory modification
involuntary episodic memories 107 technologies (MMTs) 374, 377; Plato 385, 386,
Iowa Gambling Task 107 388, 390; Thomas Aquinas 463, 465, 466
IPSPs see inhibitory post-synaptic potentials Legendary Topography of the Gospels in the Holy
ironic gap 116, 117 Land, The (Halbwachs) 528
Isidore of Seville 440 Leibniz 134
Islamic philosophy see Averroes; Avicenna L’être et le néant (Sartre) 564
Levinas, E. 564
Jackson, F. 299 Lévy-Bruhl, L. 538
Jacoby, L. 136 Lewis, C.I. 310, 311–12, 314, 320, 321n2;7;
Jaini, P. 411–12 partial report paradigm 315, 317, 318
James, William 134, 136, 174, 510, 547, 548 Leyden, von W. 54
Johnson, M.K. 136–7, 413 Liefke, K. 9
Jollimore, T. 159–60 lifelong relationships 77
Judson, L. 525 Lin, P.-A. see Ramirez, S. and colleagues (2013)
justification of memory belief 283, 284, 285, 288, linguistics 128–9, 156, 386, 520; Lacan 502f, 503;
289–92; skepticism 337–8; preservationism 323, narrativity and 193–4, 199, 201n2, 202n10;
324, 325, 326, 327, 328–30, 333–4n3; Wittgenstein 546, 549
7–8;13;15–16;18; see also belief literacy 270
Liu, J. see Sparrow, B. and colleagues (2011)
Kandel, E.R. see Abel, T. and colleagues (1997) Liu, X. and colleagues (2012) 40, 380n29; see also
Kant 134, 256, 540 Ramirez, S. and colleagues (2013)
karma (action-deed) 428, 432, 436 Llansó, E. 370
Kathlamet Texts (Boas) 538, 539 localisation 554–5
Kehl, S.J. see Collingridge, G.L. and colleagues (1983) Locke, D. 59
kinase inhibitors 374 Locke, John 134, 169, 256, 403, 409, 470;
Klein, P. 311, 314 consciousness 103–4, 105; memory traces 78,
Klein, S.B. 17, 54, 421, 422, 426n5 79, 552; temporality 216n3; theory of memory
know-how 21, 181, 193 471–6, 479; see also Hume, David; Reid,
knowing 52, 181, 269, 394; emotion 157, 160; Thomas
phenomenology 28–9, 31n32 Loftus, E.F. 54, 61n4, 156
knowledge 210, 343–5, 385–6; a posteriori Loftus, G.R. 156
(empirical) knowledge 388; a priori (conceptual) logic 492–3, 530
knowledge 388, 389, 397, 405–6; Aristotle Lømo, T. 42
396, 405; Hegel 490–1; metaphysics of time Long Commentary on De anima (Averroes) 454, 456
223, 226 long-term memories 37, 45, 159, 374, 414, 432;
‘Knowledge by Acquaintance and Knowledge by preservation 326, 327; taxonomy 9, 10–11
Description’ (Russell) 520, 521, 526n10 long-term potentiation (LTP) 41–3, 44, 84,
Köhler, Wolfgang 548 85n17;18, 374
Korsakoff syndrome 55 loose consciousness (独散意识) 438n7
Kundera, M. 156 Lost-Sock Objection 386, 387, 388, 392
Kwan, D. see Craver, C.F. and colleagues (2014) Lotze, Hermann 541
Loveday, C. 61n4
Lacan, Jacques 501, 502f, 503–6, 507n1, LTP see long-term potentiation
508n30;41; see also Freud, Sigmund Luck, S.J. 144
Lackey, J. 327 Luria, A. 375
Lamme, V.A. see Landman, R. and colleagues
(2003); Sligte, I.G. and colleagues (2008; 2010) McAdams, D.P. 270
Lampinen, J.M. 333n9 McDermott-Sales, J. 200
Landa, A. et al. (2012) 158 McDowell, J. 211, 401
Landman, R. and colleagues (2003) 149, 160 Mackie, D. 31n34
landscape of consciousness 269 McLennan, H. see Collingridge, G.L. and
Langland-Hassan, P. 202n11 colleagues (1983)
580
Index
McNamee, S. see Hayne, H. and colleagues (2011) memory seemings 297, 298, 299, 300f, 305, 307;
Madhyamika 429, 431 phenomenal seemings 301, 303, 304
Makovski, T. 150 memory self 434–5
Malcolm, N. 59, 80, 554 memory skeptical possibility argument 344
Mann, S. see Vrij, A. and colleagues (2010) Memory Theory (MT) of personal identity 476, 477
Manning, L. and colleagues (2013) 446; see also memory traces 16, 17, 85n19, 156, 541, 569;
Cassel, J.C. and colleagues (2013) consciousness 105, 410, 411; digital 245,
Marr, D. 155 248, 251; Freud 497, 498, 507n8; memory
Martin, C.B. 31n31; memory causation 65–7, causation 68, 76–8; remembering and 78–85;
68, 69, 70, 74, 75n10; memory traces 78, Wittgenstein 552–5
82, 83 memory types see taxonomy
Martin, M.G.F. 25, 210–11 ‘Memory Unchained’ (Squires) 70–1
mass media 262–5, 310 memory of the will 511, 513
Material Intellect 458 Meno (Plato) 390, 392, 395n5, 443; recollection
maternal expositional style 195 386–9, 397, 399, 405, 406
maternal reminiscence 271, 272–3, 274, 277 mental consciousness 431
Matsukura, M. 150 mental hunting 404
Matter and Memory: An Essay on the Relation of mental states 170, 226; forgetting 369–71;
the Body to the Mind (Bergson) 510, 511, 512, foundationalism 296, 297, 302, 308n2;
513–14, 515f, 516, 533 imagination 127, 128, 131, 134, 136; Indian
Matthen, M. 105–6, 119 Buddhist philosophy 418, 420–1, 423–4;
maximality 16 intentionality 90, 91, 92–5, 97, 98n6;7;8;
medial temporal lobe amnesia 39 internalism/externalism 286, 289, 293n3
Medieval philosophical thought see Augustine; mental time travel (MTT) 13, 77, 137, 197–8,
Averroes; Avicenna; Thomas Aquinas 409; Augustine 440, 446; consciousness 109–10,
Meditations (Descartes) 133 111, 410; continuism/discontinuism 228–36;
Meixner, U. 549 phenomenology of memory 23, 24; Reid 478;
‘memorative’ faculty (dhikr/ḥif ) 449, 451, 455 see also metaphysics of time; temporality
memorial discourse 563–70 mentalism 286, 299
memorial justification 290–2 Merleau-Ponty, M. 559
Memory: A contribution to experimental psychology meta-cognition 541
(Ebbinghaus) 537 meta-cognitive feelings 156–7, 158, 160, 161, 249,
memory: defining 103–4, 374; linguistic 250; forgetting 304–5
considerations 7–10; philosophical theories 51, meta-representation 16
60n1; preserving role 23, 24; research 34–5; metabotropic (energy-requiring) receptors 41, 42
‘stomach of the mind’ 444 metaphysical approach 512, 517n9–10, 546
memory blockers 374, 376 Metaphysics (Aristotle) 464, 465
memory causation 63, 74n1;2; causal condition metaphysics of time 225–6; realism and
65–7; deviant causal chains 67–9, 70; necessity epistemology of memory 221–3; remembering
claim 64–5, 70–1; opposition to causal and causal theory of time order 223–5; role of
condition 71–4; sufficiency claim 69–70 memory in temporal awareness 219–21; see also
Memory Criterion 169–72, 176 mental time travel (MTT); temporality
memory for details 55 Michaelian, K. 14, 137, 202n11, 338–9; episodic
memory for gist 55 memory 421; extended memory 248, 249;
Memory, History and Forgetting (Ricoeur) 563, 567, perspective 117, 118; truth 51, 57
569, 570 Milner, B. 37, 161
memory indicator 27–30 Mīmā sā-sūtra-bhāṣya 410
memory judgments 180, 182–7 mind consciousness (yi-consciousness 意识) 431
memory markers 85n9 mind views 243
memory modification technologies (MMT) mind–body relationship 397–8
373–9; see also forgetting; remembering misattribution 16
memory par excellence 514 misidentification 181, 182–5, 186, 231, 234
memory per accidens (intellectual memory) 461, misinformation 16, 53
462, 463, 465–6, 467n3 mnêmê 386, 396, 397, 399, 400–1, 403, 533
memory per se (sense memory) 461–5, 466, 467 mnemic causation 81
memory of the present 511 mnemonic content/representation/simulation 15,
memory processing 15 16, 17, 104
581
Index
modes of presentation 30n2 Nicolas, S. et al. (2013) 121–2

Molaison, Henry (‘H.M.’) 37, 38 Nietzsche, F. 159, 257, 265n2, 358, 362n8
Monist, The (Russell) 522 Nigro, G. 114–15
Moore, G.E. 53, 129, 519, 520, 526n3 Nikaya 416, 426n1
moral character traits 173–4 NMDAR see N-methyl-D-aspartate receptor
morality: duty to remember 357, 358, 360, 361, ‘no-ownership’ see no-self theory
377; forgetting 364–9, 371n2, 376 no-self theory (anātmavāda) 416, 417–21, 426n1–2,
Morris, T. et al. (2005) 120–1 433–4, 437
Morris water maze 35, 36f, 37–8, 43, 44, 45 ‘nocturnal pit’ 489–90
Moser, E.I. 39 noetic consciousness 12f, 13, 155, 308n3, 410
Moser, M.B. 39 nominal content 22
motion 134; detection of 10; monitoring of 39 non-causal information retention 70, 71, 74
motor memory 11, 12f, 37, 77; Bergson 511, non-conceptual content 30n8
513–14 non-declarative memories 9, 11, 12f, 192, 228
Mounce, H. 551, 552 non-deviant causal chains 68, 69, 75n8
‘movie-screen’ model of consciousness 147 non-egocentric information 119, 124n6
Moyal-Sharrock, D. 546 non-experiential contents 22, 23–4, 25, 27, 29,
Müller, J. 467, 468n11 30n7;8
Müller–Lyer illusion 131, 301, 302f, 303f, 307 non-factive states 286, 293n3
multi-level memory research 34 non-factive verbs 52, 53, 60n2, 60–1n3
multiple forms of memory 37, 510–14, 515f, 516 non-human animals 12, 13, 77, 174; consciousness
Munsat, S. 128 103, 108–10, 111; memory and experience
Musset, A. de 161 453, 456, 459n19; perception 449, 451, 463,
Myers, C.S. 538 464, 467, 468n9
Mysticism and Logic (Russell) 526n10 non-inferential memory 56–7, 118, 124n4, 339
non-personal memory 406n4
N-methyl-D-aspartate receptor (NMDAR) 43–5 non-propositional memories 9, 30n8
Nadel, S.F. 539–40 non-recollective past-tense memory
Nagarjuna (Longshu 龙树) 428 judgments 187
Naiy yiakas 416, 418, 419, 420, 421, 426 non-spatial memory 36–7
Nanay, B. 31n20 normative defeater 327
narrativity 192–3, 432; attachment theory Nuzzo, A. 493
275–7; autobiographical memory 187–90; early Nyanaponika Thera 412
parent–child reminiscing 268, 270–4; episodic Ny ya-Vai e ikas 416, 417, 421, 419, 422,
memory 196–9; narrative identity 269–70, 426n2
376; Ricoeur 565, 566–7, 569; roles and Nyāyabhāṣya 411
functions of narrative remembering 199–201;
social interactionist theory of autobiographical Oberauer, K. see Rerko, L. and colleagues (2014)
memory 193–6, 201n3, 202n5;6; socially object recognition 36–7
constructed life themes 277 obsessional neuroses 505–7
natural-kind based taxonomy 9, 14–17, 18, Odell, S.J. 56
155, 157 ‘Of Personal Identity’ (Butler) 477
Neisser, U. 55, 85n8, 114–15 O’Keefe, J. 38, 39
Nelson, K. 270; narrativity 194, 196, 197, 198–9, O’Keefe, T. 356
201n2;3, 202n4;9 Olick, J. 258
Neo-Platonism 439, 440, 441, 446, 448 Olsson, E.J. 313, 315, 317, 318, 320, 321n1;7
neural system 13–14, 15, 17, 84, 544–5; scientific On Certainty (Wittgenstein) 550
explanation 39–40, 41, 43–4 ‘On the Experience of Time’ (Russell) 522
neurodegeneration 174 On Memory and Recollection (de memoria et
neurology 497–8 reminiscentia) (Aristotle) 127, 132, 396, 399–405,
Neurophilosophy (Churchland) 34 406n2, 533; Averroes and 458, 459n32–3;
neuroses 499, 501, 505, 508n11 Thomas Aquinas 461
‘Neurotic’s Individual Myth, The’ (Lacan) 506–7 ‘On the Nature of Judgment’ (Moore) 519
neurotransmitters 41 ‘On Propositions: What They Are and How They
‘New Evil Demon’ problem 284–5, 286, 288, Mean’ (Russell) 523, 524
292, 303, 304 On The Soul (Parva naturalia) (Aristotle) 448,
Nguyen, P.V. see Abel, T. and colleagues (1997) 449, 454
582
Index
On the Trinity (De Trinity) (Augustine) 440, 442, perceptual skeptical possibility argument 343–4,
443, 445, 467 344–5
Oneself as Another (Ricoeur) 563, 564–5 ‘perfect’ animals 462, 463, 464–5, 467n4
ontological groundedness 15, 17 perfuming (vasara) 429
opsin gene 40 peripheral vision 25, 31n20
optogenetics 39–40, 45, 375 personal identity 105; Classical Indian philosophy
organic memory 256, 257, 265n2 408–9, 410; episodic memory 174–6;
organismic systems 155, 163n1 judgments of psychological continuity 172–4;
Origin of German Tragic Drama, The (Benjamin) 568 Locke 476, 477; Memory Criterion 169–72,
original justification 328 176; memory as evidence of 172; self-
ostensible memory 58–60 representations 176–8
otherness 564–5 personal memory 21, 406n4, 422, 529, 539,
Our Knowledge of the External World (Russell) 523 550–1
Outline of Philosophy, An (Russell) 526n15 ‘personal’ states 73, 74, 77–8, 79, 155–9
personality 270
Pacherie, E. 157 perspective 77, 113–14; emotion 118–19, 120,
Pang, P.T. see Liu, X. and colleagues (2012) 122, 124n10; field and observer perspectives
Pankejeff, Sergei (‘The Wolf Man’) 499, 114–15; plurality of 119–23; truth and
500t, 501 authenticity 115–19
Parfit, D. 170–1, 185 Phaedo (Plato) 389–92, 406
partial reliability 315–18 Phaedrus (Plato) 385, 391
‘partial report’ paradigm (Sperling) 145, 146, 147, phantasia 399, 400, 403–4, 406n5, 448
150, 151, 152; iconic memory 142, 143f, 144 phantasma 132, 133, 402, 403, 462
partially correct memory 55, 61n5 phantasy (phantasia) 463, 467–8n7
particularity 26, 27, 31n23;24 pharmaceuticals see memory modification
Passions (Descartes) 133 technologies (MMTs)
past-life beliefs 175–6 phase precession 17
past-tense beliefs 221, 222, 223 Phenomenal Conservatism 290, 291, 324
past-tense judgments 186, 187 phenomenal dogmatism 297–9, 303, 304,
pastness 28, 29–30, 134, 455; emotion 155, 160, 305–6, 307
162; experiences 23, 24, 25; ‘feeling of ’ 103, phenomenal seemings 301, 302f, 303f, 304
104–5, 106, 110; historical memory 257, 258, phenomenology 21, 30n1, 31n14, 549, 567–9;
531–2; memory traces 78; perception 53, 91, attitude 22, 27–30; content 22–7, 30n2;5;6;
92, 93–5, 96, 97; reliving/re-experiencing 16; episodic memory and 91–7; experiential
remembering past pleasures 357; representation contents 22, 24–7, 30n7;8;12, 31n15;
54, 78, 79–80, 106; self-consciousness 180, 188; phenomenological taxonomy 12f, 13
temporality 207, 208, 209 Phenomenology of Internal Time-Consciousness, The
Peacocke, C. 25, 209, 216n9 (Husserl) 558
Pears, D. 209 Phenomenology of the Spirit (Hegel) 256, 487,
perception 15, 37, 65, 147, 396; Aristotle 399, 490–2, 494n2;8
400, 401–2, 403, 406n4;7; attention 413–15; Philebus (Plato) 160–1, 394n2, 397, 399
Augustine 442–4; authenticity 56, 60; Averroes Phillips, I.B. 145, 146, 149
456; Avicenna 449, 452, 453–4, 458n6; belief Philosophical Investigations (Wittgenstein) 525, 548,
290; Chinese Buddhist philosophy 430–1, 436; 550, 552
definition of memory and 103–4; emotion Philosophical Remarks (Wittgenstein) 550
158, 159; experiential content 24, 25, 26, 27, Philosophy of Logical Atomism, The (Russell) 523
30n11;13, 31n18;19;22;27; Freud 497–8; Philosophy of Mind (Hegel) 488, 489
intentionality 89, 90, 91, 92–5, 97, 98n5, Philosophy of Symbolic Forms (Cassirer) 265n3
448–9; knowledge 391; Locke 471, 473, 474; phobias 499–500
memory traces 78, 79; no-self theory 417, 418, physical memories 83, 85n14
420; pastness 53, 105; preservation 326, 328, pictorialism 57
329, 330, 331, 332, 334n18; phenomenology picture writing method 539
23, 29; Plato 396, 397, 398; Reid 470, 477, Pignatelli, N. see Ramirez, S. and colleagues
478; reliability 338; temporality 211, 220, 221; (2013)
Wittgenstein 551–2; working memory 450 Piolino, P. see Nicolas, S. et al. (2013)
perception (aisthêsis) 399 ‘place’ fields/neurons 38–9
perception memory 511 plastic synapses 34, 42
583
Index
Platform Sutra of the Sixth Patriarch (Huineng 慧能) procedural memory 21, 38, 77, 181, 192, 296;
429, 430 emotion 155, 156; self-implication 409
Plato 131–2, 160, 394n1–3, 395n5–8, 443, 561; progressive brain disorders 55, 174, 248, 373
Aristotle and 396–7; memory as knowledge- ‘Project for a Scientific Psychology’ (Freud) 497–8
over-time 385–6; Meno and recollection 397, propositional content 22, 30n3–4;8
399, 405, 406, 386–9; Phaedo and recollection propositional memories 9, 56, 57, 61n9, 290;
386–93, 541; Wittgenstein 547, 554; see also coherentism 311–14; intentionality 89, 90,
Aristotle; Epicureans; Stoics 97n2; justification of memory beliefs 331,
pleasure principle 496 334n18; temporality 208
Plotinus 441 propranolol 374, 375
plurality: perspective and 119–23 prosopagnosia 157
poetry 561–2 prospective memory 244
point of view (POV) shots 120 protein kinase M-zeta (PKMzeta) 374
‘Points about Denoting’ Russell 526n4 Proust, J. 157, 161
Pollock, J. 547, 555n1 Proust, M. 155
possible intellect 462, 465–6 psychoanalysis 496, 498, 501
post-synaptic glutamate receptors 42 psychological approach 511–12, 517n9
post-synaptic receptors 41, 42 Psychological Continuity Criterion 171, 172–4,
post-traumatic amnesia 82 175–6, 183
post-traumatic stress disorder (PTSD) 158, 373, psychology: natural kinds 14
374, 376, 378, 379 Psychology and Primitive Culture (Bartlett) 538, 543
postdictive experiences 145–6 psychoses 508n11
posterior joint probability 311–12, 313–17, 318, PTSD see post-traumatic stress disorder
320, 321n5 pure memory 511, 513, 517n12
Poston, T. 338 Puryear, C.B. see Liu, X. and colleagues (2012)
postulational approach: induction 58 putative memories 547–8
postural schema 540–2 Pyrrhonian readings 546
pragmatic approach: induction 58
pre-synaptic receptors 42 quasi-memory 232, 408; personal identity 170,
Predictive Coding/Processing paradigm 198 172, 176, 184, 185
present-tense analog 186, 187, 221 quasi-transparent simulation 16
‘Presentation of the Suite’ (Lacan) 502, 508n30 Quine, W.V.O. 525, 550
presentational phenomenology 299–301, 304
presentism 221, 222–3, 226n5 radical skepticism 289
presentness 219–20 Ramirez, S. 40, 84, 380n30; see also Liu, X. and
preservationism 117, 124n5, 208, 235, 292, 305; colleagues (2012)
attractiveness of 332–3; characterizing 325–7; Ramirez, S. and colleagues (2013) 40
generativism and 323, 326, 327, 328, 329–32, Ramon y Cajal, Santiago 34
333n6; justification of memory belief 323, 324, RAP see Relationship Anecdotes Paradigm
325, 326, 327, 328–30, 333–4n3;7–8;13; ‘Rat Man, The’ (Freud) 506–7
15–16;18; remembering 371n1; synchronic/ Rational-Emotive Behaviour Therapy (REBT) 160
diachronic approaches 323–5 rationality 160–2, 328–9, 335; defensibility 287–8
Previous Awareness Condition (PAC) 473, 477 re-evaluation 160
Price, H.H. 58 re-experiencing 16, 228, 409; consciousness 104,
prima facie justification 328, 329–30, 331, 332, 333, 105, 106–7, 108, 111; narrativity and 198, 199
334n16 Read, R. 547
‘Primal Scene’ (Freud) 499, 500t, 501 realism 79–80, 234, 429–30, 433, 547;
‘primitive mentality’ 538 coherentism 318–19, 321n10; epistemology of
Principal of Ecological Assembly (PEA) 250–1 memory 221–3; forgetting 366–7; Hume 484;
Principle of Indifference 321n7 Lacan 503–4; reality principle 496; Reid 477–9;
Principles of Mathematics, The (Russell) 519, 520, 526n3 Russell 519; temporality 211–12, 215, 216n9,
Prior, A.N. 186 219–20, 226n1
Pritchard, D. 249–50 ‘Reappearance Hypothesis’ 85n8
‘probe’ trials 35 reasoning 27, 31n25
Problem of Forgotten Evidence 290–1 REBT see Rational-Emotive Behaviour Therapy
Problems of Philosophy, The (Russell) 207, 210, 212, recall 7, 375, 521, 543, 544
213, 321n9, 521 recognition 410, 411, 417–18
584
Index
recollection 7, 202n12, 248, 310–11, 316; ‘Remembering’ (Martin and Deutscher) 31n31;
Aristotle 399, 400, 402–3, 533; Augustine memory causation 68, 69, 70, 74, 75n10;
442–4, 446; Averroes 456–7; Avicenna memory traces 78, 82, 83
451–4; Bergson 510, 511, 514, 515; Freud remembrance see duty to remember
496, 501; Halbwachs 529–30, 532–4; Hegel ‘Remembrance’ (Hölderlin) 562
487, 488–93; Locke 472; memory causation reminiscence 268–9, 270–4, 275–7
63–4, 65–6, 67, 70, 74n4;5; Plato 386–92, remorse 161
394, 397; preservation 326; Thomas Renan, E. 256
Aquinas 464 repeated reproduction method 538–9, 542
reconstructive model 105, 116, 124n5, 305; representationalism 54, 328; consciousness 104,
emotion 157, 160; extended memory 248, 251; 107–8, 110–11; Hegel 487–8, 489, 490; memory
preservation 326, 327 traces 78, 79–80; narrativity 193; perspective
reconstructive remembering 540–2 115, 117, 123n1; Ricoeur 570; Russell 519
Redondo, R.L. see Ramirez, S. and colleagues repressed memories 16
(2013) Republic (Plato) 385, 388, 391, 397, 561
reference 402–3, 405 Rerko, L. and colleagues (2014) 150
‘Reference Back’ (poem) (Larkin) 154 Restoring Active Memory program (RAM)
reflection 471, 472, 473 (Defense Advanced Research Projects
reflexivity 95–7, 98n9–12, 200, 498, 501 Agency) 374
‘Refutation of a Theory of the Self’ (Vasubandhu) retained acquaintance 521
416–21, 426n2–3 retention 456, 465, 521, 540; Avicenna (Ibn Sina)
regret 160–2, 213–15 449, 450–1, 454, 458n7; Locke 474
Reid, Thomas 408, 411, 470, 551, 552; memory retrieval 83, 369, 392, 422, 433, 452; taxonomy
traces 79–80, 85n10; personal identity 172, 15, 16; truth 51, 57–8
174; theory of memory 476–9; see also Hume, retrograde amnesia 176, 421–2
David; Locke, John reverie 472
reiterative temporal rhythms 260–1 Ricoeur, P. 258, 563–70
‘Relation of Sense-Data to Physics, The’ (Russell) rightness 156, 157, 160
522–3 Rilke, Rainer Maria 562
Relationship Anecdotes Paradigm (RAP) 158 Rimbaud, A. 563
relearning 81–3, 85n13 Ritalin (stimulant methylphenidate) 374
reliability 297, 303, 521–2; coherentism 315–18, Rivers, W.H.R. 538
320; internalism/externalism 284–5, 287t, 292; Roberts, W. 108–9
skepticism and 336, 337–8, 339, 340–1 Robinson, J.A. 114
reliving the past 16 Rocha-Miranda, C.E. see Gross, C.G. and
Remarks on the Philosophy of Psychology colleagues (1972)
(Wittgenstein) 548–9, 553 rodent spatial memory 35, 36f, 37–8, 39
Remembering: A study in experimental and social Rodin, D. 370
psychology (Bartlett) 537, 542–3 Rooney, Wayne 121
remembering 7–9, 17, 21, 52, 64, 72; Aristotle Rosen, S. 370
459n33; attitude 22, 27–30; Augustine 441–2, Rosenbaum, R.S. see Craver, C.F. and colleagues
445–6; autobiographical 197, 198; Averroes (2014)
456; Avicenna 451–3; Bartlett 537–44; Roux, W. 256
constructivism 51, 54, 56, 57, 137; content Rowlands, M. 119
22–7; epistemic relevance 70; experiential rumination 473
contents 24–7; Freud 497–9, 500t, 501, 503, Russell, Bertrand 29, 31n33, 156, 292, 321n9,
507n7; imagination and 128–9, 130, 131, 132, 414; consciousness 104, 105; imagination 134,
135–6, 138; Lacan 503–5; Locke 470, 471, 136, 548; memory traces 79, 81; scientism
473–4, 475–6, 479n2; memory traces and 546–7; self-consciousness 181, 187–8, 189,
78–85; mental time travel 232, 233; narrativity 190n6; skepticism 344–5, 548; temporality 207,
201; preservationism 371n1; Reid 470, 478, 208, 210, 212, 213, 216n5; theory of memory
479; Ricoeur 566–7; roles and functions 519–25, 526n3–4;9–10;12–15; truth 58, 59
of narrativity 199–201; truth and 115, 116; Ryckebusch, O. 565
unconscious memory 496; Wittgenstein Ryle, G. 135
549–50; see also duty to remember; memory
modification technologies (MMTs) abara 410, 411
remembering intention 244 Sakyamuni Gautama 428, 429
585
Index
samsara (lunhui 轮回) 432 mental time travel (MTT) 228; narrativity
Sarathchandra, E.R. 412 193; personal identity 172, 173, 174, 176;
Sartre, Jean-Paul 501, 564 preservation 326–7; self-implication 409;
Saunders, J.T. 58 taxonomy 9, 11, 12f, 13, 14, 21; temporality
scaffolding 270, 271, 272, 273–4, 276 208, 211, 212, 214, 216n8; see also episodic
scalar taxonomy 9, 10–11, 14 memory
Schacter, D.L. 38; and colleagues (2007) 202n7 semantically remembered states of affairs 89, 97n3
Schechtman, M. 55 ‘Seminar on the Purloined Letter’ (Lacan) 502–3
Schema L (Lacan) 502f, 505 Seneca 356, 357, 359, 362n5
Scherer, K. 155 ‘Sensation and Imagination’ (Russell) 522
Scholl, B.J. 145–6 sense memory (memory per se) 461–5, 466, 467
Scholte, H.S. see Sligte, I.G. and colleagues SenseCam (digital camera) 245
(2008, 2010) sensitive soul (sensitive memory) 442
Schubert, S. 320 sensory memory (ultra-short-term memory) 10, 11
Schulte, J. 553 sensory states 106, 449; Hegel 488–9; Hume 482,
Schütz, G. 44; see also Bourtchuladze, R. and 484; Locke 471–2, 473
colleagues (1994) Sentencia Libri De Sensu et Sensato (Thomas
science 17, 34–5, 83–5 Aquinas) 461
Science of Logic (Hegel) 487, 492–3, 494n2 Sequence Analysis 14–15, 16, 17
scientism 546–7 sequentiality 15
scopolamine 374 serial reproduction method 539–40
Second Law of Thermodynamics 224 Shaw, W. 358
second-order desires 378 Shenxiu (神秀) 436
secure attachment 274–5, 276–7 Shiffrin, R.M. 10, 11
‘seduction scene’ (Freud) 499 Shimojo, S. 146
seemings 297, 298, 299, 300f, 302, 305; Shoemaker, S.: experience 473; self-consciousness
phenomenal 301, 303, 304 182, 183–4, 185; truth 59, 60, 61n10
Sein und Zeit (Heidegger) 558, 560, 561 Shogenji, T. 313, 315, 317, 318, 321n7
selectivity 42 Shore, B. 123
self-ascribed memory 27–8, 29, 31n26;27;30, 53 short-term memory 10, 11, 37, 374, 380n8, 432;
self-awareness 114 fragile visual short-term memory (fragile VSTM)
self-consciousness 180–90 141, 142, 147, 148f, 149–51, 152
self-deception 366–7 Shu Chen 433
self-examination see introspection Siegel, S. 305–7
self-historicization 502 Sierra, M. 175
self-identity 376, 432; Augustine 444, 445, Silence of the Lambs, The (film) 120
446n2; social identity 268, 269, 270–1, 272–3, Silva, A.J. 44; see also Bourtchuladze, R. and
276, 277 colleagues (1994)
self-image 319 Simple Internalism 285, 286
self-implication 408, 410, 411, 412 simple/similar content view 26
self-knowledge 31n28, 377, 378, 440, 443–4 simulation theory of episodic memory 414;
self-nature (zixing自性) 431, 434, 436 mental time travel 233, 234, 235; narrativity
self-realization 496, 502, 505 197, 198, 199
self-recognition 410 ‘Sing a Family Song’ (Haggard) 34
self-reference 95–7, 98n9–12, 110, 421–2, 455 Singer, R.A. see Zentall, T.R. and colleagues
self-representation 409 (2008)
self-specifying memories 181–2, 186, 188 single-cell electrophysiology 40–1
selfhood 361; Augustine 445–6; Chinese Buddhist size constancy 302
philosophy 428, 429, 430–3, 437; Classical skepticism 290, 323, 335, 379, 389; global
Indian philosophy 408–11 skeptical memory argument 336, 341; local
Sellen, A.J. 244 skeptical memory arguments 336, 341–2;
semantic memory 421, 473, 478; Classical Indian memory as storage unit 342; memory traces
philosophy 410, 411; consciousness 107, 79, 84; perceptual beliefs and memory beliefs
108, 109, 181; emotion 155, 156, 158–9; 343–4, 346n26–8; premise 337–41, 346n20;
foundationalism 296; intentionality 88–9, Russellian skepticism 344–5, 548
90–1, 97n1; memory modification technologies Slamecka, N.J. 136
(MMT) 378–9; memory traces 77, 78; sleep specific brain rhythms 374
586
Index
Sligte, I.G. and colleagues (2008, 2010) 149, philosophy 429, 430, 431, 432, 434–5, 436;
150, 151 Indian Buddhist philosophy 423, 424, 425
slow wave oscillations 374 storehouse metaphor 326, 333n9, 474–5, 552
Smith, M. 120 Stories of Mr. Keuner (Brecht) 519
social cooperation 107 Stout, G.F. 551, 552
social frames (cadres sociaux) 530–1, 542 ‘straightforward meditation’ (yixingsanmei
Social Frameworks of Memory, The (Les cadres sociaux 一行三昧) 436
de la mémoire) (Halbwachs) 528, 529, 532, 534 Strawson, G. 485n6
social identity 268–9; attachment theory 268, strong access internalism 298, 304
274–7; early parent–child reminiscing and strong connectedness 171, 172, 175–6
narrative development 268–9, 270–4; structural memory 498
narratives and narrative identity 269–70; studying see learning
socially constructed life themes and narrative Stuss, D.T. 421–2
identity 277 Styron, W. 377–8
Social Interactionist Theory (SIT) 193, 199, ‘sub-personal’ states; emotion 155–9, 163;
201n1, 361 memory causation 73, 74; memory traces
social psychology of remembering 542–3 77–8, 79
social recognition 37 subjective experience 13, 14
socially distributed memory systems 245–6 Suddendorf, T. 13; see also Cheng, S. and
socially entanglement: digital 245 colleagues
Soliloquies (Augustine) 441 Suh, J. see Ramirez, S. and colleagues (2013)
somatoform (bodily symptoms) 158 Summa Contra Gentiles (Thomas Aquinas) 461
Sophie’s Choice (Styron) 377–8, 378–9 Summa Theologiae (Thomas Aquinas) 461, 462–3,
soul 448, 449, 454; psychê 398, 440, 445, 457, 467–8n7
465–6, 496 Superego 496
sound-induced visual bounce 145–6 supernormal stimulus 247
source monitoring errors 55 supposing 22
source monitoring framework (SMF) 136–7 Sutton, J. 51, 245
Souza, A.S. see Rerko, L. and colleagues (2014) Swanson, K.L. 114
Sparrow, B. and colleagues (2011) 247 Swaziland 543
spatial cognition 115, 117, 421; maps 38–41, 84, Symbolic order 502f, 504–5
119, 122–3 symbolism 259–61, 262, 263, 264, 265n3
Special Theory of Relativity 225, 226n2 synapses 41
specious present 337 synchronic approaches 323–5
Spekreijse, H. see Landman, R. and colleagues systemic remembering 497
(2003)
Sperling, G. 145, 146, 147, 150, 151, 152; iconic Taber, J.A. 410
memory 142, 143f, 144 task-directedness 538
Spinoza 134, 162 task-irrelevant information 109
spirit 256, 487, 490–2, 494n2;8 taxonomy 7–10, 76–7, 446; content-based
Spittle, M. see Morris, T. et al. (2005) taxonomy 12; episodic memory 10, 13–14;
sports psychology 120–1 hierarchical taxonomy 9, 11, 12f; Locke 472,
Squire, L.R. 11, 12f, 38 474; natural-kind based taxonomy 9, 14–17,
Squires, R. 70–1, 71–2, 81, 554 18; phenomenological taxonomy 12–13; scalar
r dhara 420 taxonomy 9, 10–11
Stagner, J.P. see Zentall, T.R. and colleagues technology see extended memory
(2008) teleofunctional representation 104, 106, 107–8
static memory traces 78, 85n6 temporal decentring 23
Steindam, C. see Craver, C.F. and colleagues temporality 207, 216, 492; A-theory of time 219,
(2014) 220, 221, 226; autobiographical memories
Stern, D. 555 189, 190n10; autonoetic consciousness 421,
Stoics 356–8; see also Aristotle; Epicureanism; 422; Bergson 512–13; collective memory
Plato; 259–61, 262; displaced memories 55; episodic
storage 15, 83 memory and consciousness 13, 15, 105, 106,
stored knowledge 344 108; episodic memory as a generative capacity
storehouse consciousness (ālaya-vijñāna/alaya 208–10, 216n3;7; episodic memory and
consciousness 藏识); Chinese Buddhist grasp of time as a domain 210–15; Heidegger
587
Index
557–62; memory as knowledge 385–6; memory Tustin, K. see Hayne, H. and colleagues (2011)
per accidens (intellectual memory) 461, 462, Tversky, B. 123
463, 465–6, 467n3; memory per se (sense
memory) 461–5, 466, 467; Ricoeur 563–7; Uddyotakara 417–18
scalar taxonomy 9, 10–11; see also mental time ultima facie justification 329–30
travel (MTT); metaphysics of time ultra-short-term memory (sensory memory)
tenses 186–7, 190n4, 221 10, 11
Theaetetus (Plato) 388, 389, 392–4, 397, 398, 547 unbearable memories 378, 379
theoretical spirit 487, 488 unconscious memory 104; Freud 496, 497, 498,
Theory of Knowledge (Russell) 522 499, 500–1, 507; Lacan 502, 503, 504, 505,
theta sequences 17 507, 508n41
thinking consciousness 431 understanding 471, 500; acquired intellect
third-person perspective 77, 114 465–6, 467
36 pairings (dui 对) 430 unicity of the intellect 458
Thomas Aquinas 132–3, 136; intellectual memory unjustified belief 331–2
461, 465–6, 467n4, 467–8n7; sense memory unpleasant memories 159–60, 453; memory
461–5; see also Augustine modification technologies (MMT) 373, 375,
Thomson, J.J. 379 377–8
thought 488 Urmson, J.O. 27, 28, 135–6
Tiantai School of Buddhism (天台) 435, 436, 437 Utpaladeva 411
time see mental time travel (MTT); metaphysics of
time; temporality value 160, 162
Time and Free Will (Bergson) 511, 512 Van Koppen, P.J. see Crombag, H.F.M. and
Time and Narrative (Ricoeur) 563, 570 colleagues (1996)
Titchener, E. 29 Vandenbroucke, A.R. see Sligte, I.G. and
token reflexivity 95, 96 colleagues (2010)
Tolman, E. 38 ‘Various Aspects of Memory in Bergson’
Tonegawa, S. 40, 44; see also Liu, X. and colleagues (Hyppolite) 512
(2012); Ramirez, S. and colleagues (2013) Vasubandhu (Tipo 提婆) 416–21, 423, 424, 425,
Toobin, J. 370 426n2–3, 429, 432
Tower of Hanoi (block-moving game) 37 V tsy yana 411
traditionalism 559–60 Vendler, Z. 116, 129–30
traits see character traits verbs 52, 53, 60n2, 60–1n3
transactive memory (TM) 244, 246–7 video communication 264
transcranial stimulation 374 viewpoint-dependence 16
transference 158 vijñāna 424
transgene approach 45 virtue epistemology 292
transitivity 169–70, 171, 172 virtue ethics 358
transmitted justification 328–9 virus vectors 39–40
trauma 158, 373, 374, 376, 378 visible persistence 143–4, 146
Treatise of Human Nature (Hume) 480–3, visual experiences 25, 57, 424; foundationalism
484n1–2 299, 300, 301, 307; fragile short-term memory
tree rings 7 (fragile VSTM) 141, 142, 147, 148f, 149–51,
tropisms 154 152; iconic memory 141, 142, 143f, 144,
true self 376, 377, 378, 379, 430, 435 145, 147, 148–9; memory causation 63, 64;
trust and glue 248, 249, 253n1 perspective 114, 115, 116, 117, 119–20, 121;
truth 51–4, 288; assessing accuracy 51, 52, 55, 56, preservationism 330, 331; senses 10, 40, 41
58–60; authenticity constraint 52, 54, 56–8; vivacity 134–5
coherentism 310, 311, 313–14, 315, 317, 320; voluntarism 559, 560
episodic memory 15; foundationalism 301, 304, Vrij, A. and colleagues (2010) 122, 124n11
307; metaphysics of time 223; perspective and Vygotsky, L.S. 270
115–19; standards 54–6
Tulving, E. 83, 97n1, 137, 216n1, 228; Wagenaar, W.A. see Crombag, H.F.M. and
consciousness 103, 104, 108; emotion 155, colleagues (1996)
161; mental time travel (MTT) 230; personal Wagoner, B. 537, 542
identity 173; self-implication 409, 411; wakefulness (jue 觉) 433, 435
taxonomy 11, 12f, 13, 16 Wallace, J. 162
588
Index
‘War of the Ghosts’ (Boas) 538, 539 Winner, D. 121

Ward, A.F. 247, 251 Wittgenstein, L. 80, 525, 540, 546–55; Aristotle
Ward, J. 541 401, 403; self-consciousness 182–3, 186
Warfield, T.A. 311, 314 ‘Wolf Man, The’ (case history) (Freud) 499, 500t,
Warmelink, L. see Vrij, A. and colleagues (2010) 501, 506–7
Watanabe, K. 146 Wollheim, R. 116, 123n2
Watergate scandal 55 Wong, P.S. see Landa, A. et al. (2012)
Watt, A.P. see Morris, T. et al. (2005) working memory 10, 109, 374, 413; imagery
Wax Tablet imagery: Plato 392–3, 394, 395n8; 141–2, 147–3, 149, 150–1; perception versus
Aristotle 397, 398, 400, 401; Bartlett 541; retention 450
Wittgenstein 547, 554 world time 557
Ways of World-Making (Goodman) 265n3 worldly states of affairs 90, 92–5
weak access internalism 298, 299 Worms, F. 517n15
Wearing, C. 161 Würzburg School 538
Wedgwood, R. 286
Wegner, D.M. 244, 246–7, 249; see also Sparrow, ‘xerox model of memory’ 60n1
B. and colleagues (2011)
well-being 200 Yampolsky, P. 430
Werning, M. 997–8n4; see also Cheng, S. and Yogacara 429, 431, 432–3, 435, 436
colleagues Yogācārabhūmi 423, 426n7
wh-clause constructions 30n3, 53, 130
what-where-when criterion of events (WWW) Zajonc, R. 31n34
12, 108, 174, 409 Zentall, T.R. and colleagues (2008) 109
White, H. 569 zeta inhibitory peptide (ZIP) 374, 380n25
Whittaker, S. 244 Zettel (Wittgenstein) 549, 550, 553
Williams, B. 177 Zhiyi (智顗) 435
Winkielman, P. 31n34 Zola-Morgan, S. 11, 12f
589

The Routledge Handbook of Philosophy of Memory-Routledge (2004) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Routledge Handbook of Philosophy of Memory-Routledge (2004) PDF

Uploaded by

Copyright:

Available Formats

THE ROUTLEDGE HANDBOOK OF

• The nature of memory

THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF EMPATHY

THE ROUTLEDGE HANDBOOK OF EPISTEMIC CONTEXTUALISM

THE ROUTLEDGE HANDBOOK OF EPISTEMIC INJUSTICE

THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF PAIN

THE ROUTLEDGE HANDBOOK OF BRENTANO AND

THE ROUTLEDGE HANDBOOK OF COLLECTIVE INTENTIONALITY

THE ROUTLEDGE HANDBOOK OF LIBERTARIANISM

THE ROUTLEDGE HANDBOOK OF METAETHICS

THE ROUTLEDGE HANDBOOK OF PHILOSOPHY OF MEMORY

ISBN: 978-1-138-90936-6 (hbk)

Typeset in Bembo Std

List of figures and tables xi

Editors’ introduction: the philosophy of memory today 1

1 Taxonomy and unity of memory 7

2 The phenomenology of memory 21

3 Memory and levels of scientific explanation 34

4 Memory and truth 51

7 The intentional objects of memory 88

8 Memory and consciousness 103

9 Memory and perspective 113

10 Memory and imagination 127

11 Memory images 141

12 Memory and emotion 154

13 Memory and personal identity 169

14 Memory and self-consciousness 180

15 Memory and narrativity 192

16 Memory and the concept of time 207

17 Memory and the metaphysics of time 219

18 Memory as mental time travel 228

19 Extended memory 243

20 Collective memory 255

21 Memory and social identity 268

22 Internalism and externalism 283

25 Preservation and generation 323

26 Skepticism and memory 335

27 A duty to remember 351

28 An obligation to forget 364

29 The ethics of memory modification 373

32 Classical Indian philosophy 408

33 Indian Buddhist philosophy 416

34 Chinese Buddhist philosophy 428

36 Avicenna and Averroes 448

37 Thomas Aquinas 461

38 John Locke and Thomas Reid 470

39 David Hume 480

41 Sigmund Freud and Jacques Lacan 496

42 Henri Bergson 510

43 Bertrand Russell 519

44 Maurice Halbwachs 528

45 Frederic Bartlett 537

46 Ludwig Wittgenstein 546

47 Martin Heidegger 557

48 Paul Ricoeur 563

John Bickle is Professor of Philosophy, Adjunct Professor of Psychology, Head of the

Deborah L. Black is Professor of Philosophy and Medieval Studies at the University of

Taylor Carman is Professor of Philosophy at Barnard College, Columbia University, USA.

Monima Chadha is Head of Philosophy of the Philosophy Program at Monash University,

Felipe De Brigard (PhD) is Assistant Professor in the departments of Philosophy, Psychology

Ronald de Sousa is Professor Emeritus of Philosophy at the University of Toronto, Canada.

Alexandre Dessingué is Professor of Literacy studies at the University of Stavanger in Norway.

Paula Droege is currently a Fellow at the Institute for Advanced Study/Wissenschaftskolleg