(Routledge Research in Phenomenology) Marek Pokropski - Mechanisms and Consciousness - Integrating Phenomenology With Cognitive Science-Routledge (2021)

Mechanisms and Consciousness
This book develops a new approach to naturalizing phenomenology. The

author proposes to integrate phenomenology with the mechanistic frame-
work that offers new methodological perspectives for studying complex
mental phenomena such as consciousness.
While mechanistic explanatory models are widely applied in cognitive
science, their approach to describing subjective phenomena is limited.
The author argues that phenomenology can fill this gap. He proposes two
novel ways of integrating phenomenology and mechanism. First, he pres-
ents a new reading of phenomenological analyses as functional analyses.
Such functional phenomenology delivers a functional sketch of a target
system and provides constraints on the space of possible mechanisms.
Second, he develops the neurophenomenological approach in the direc-
tion of dynamic modeling of experience. He shows that neurophenom-
enology can deliver dynamical constraints on mechanistic models and
thus inform the search for an underlying mechanism.
Mechanisms and Consciousness will be of interest to scholars and
advanced students working in phenomenology, philosophy of mind, and
the cognitive sciences.
Marek Pokropski is Assistant Professor at the Faculty of Philosophy,

University of Warsaw, Poland.
Routledge Research in Phenomenology
Edited by Søren Overgaard, University of Copenhagen, Denmark,
Komarine Romdenh-Romluc, University of Sheffield, UK, and David
Cerbone, West Virginia University, USA
Husserl’s Phenomenology of Intersubjectivity

Historical Interpretations and Contemporary Applications
Edited by Frode Kjosavik, Christian Beyer, and Christel Fricke
Phenomenology of the Broken Body

Edited by Espen Dahl, Cassandra Falke, and Thor Erik Eriksen
Normativity, Meaning, and the Promise of Phenomenology

Edited by Matthew Burch, Jack Marsh, and Irene McMullin
Political Phenomenology
Experience, Ontology, Episteme
Edited by Thomas Bedorf and Steffen Herrmann
Levinas and Analytic Philosophy

Second-Person Normativity and the Moral Life
Edited by Michael Fagenblat and Melis Erdur
Philosophy’s Nature
Husserl’s Phenomenology, Natural Science, and Metaphysics
Emiliano Trizio
The Bounds of Self

An Essay on Heidegger’s Being and Time
R. Matthew Shockey
Towards a Phenomenology of Values

Investigations of Worth
D.J. Hobbs
Mechanisms and Consciousness

Integrating Phenomenology with Cognitive Science
Marek Pokropski
For more information about this series, please visit: https://www.rout-

ledge.com/Routledge-Research-in-Phenomenology/book-series/RRP
Mechanisms and
Consciousness
Integrating Phenomenology with
Cognitive Science
Marek Pokropski
First published 2022
by Routledge
605 Third Avenue, New York, NY 10158
and by Routledge
2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
Routledge is an imprint of the Taylor & Francis Group, an
informa business
© 2022 Marek Pokropski
The right of Marek Pokropski to be identified as author[/s] of
this work has been asserted by him/her/them 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.
Publication of this book was supported by Grant No. 2017/27/B/
HS1/00735 financed by the National Science Centre, Poland.
Library of Congress Cataloging-in-Publication Data
Names: Pokropski, Marek, author.
Title: Mechanisms and consciousness : integrating phenomenology with
cognitive science / Marek Pokropski.
Description: New York, NY : Routledge, [2022] | Series: Routledge research
in phenomenology | Includes bibliographical references and index.
Subjects: LCSH: Phenomenology. | Cognitive science | Mechanism (Philosophy)
Classification: LCC B829.5 .P6155 2022 (print) | LCC B829.5 (ebook) | DDC
142/.7--dc23
LC record available at https://lccn.loc.gov/2021032724
LC ebook record available at https://lccn.loc.gov/2021032725
ISBN: 978-0-367-46525-4 (hbk)
ISBN: 978-1-032-13792-6 (pbk)
ISBN: 978-1-003-03536-7 (ebk)
DOI: 10.4324/9781003035367
Typeset in Sabon
by SPi Technologies India Pvt Ltd (Straive)
To Paula and Malina
Contents
List of Figures x
Acknowledgments xi
Epigraphs xii
Introduction 1
PART I
Integrating Phenomenology with Cognitive Science 11
1 The Concept of Phenomenology 13

1.1 Introduction 13
1.2 What Is Phenomenology 14
1.2.1 Reduction and Variation as Phenomenological
Methods 14
1.2.2 Intentionality 18
1.2.3 Static and Genetic Approaches 19
1.2.4 Phenomenological Psychology 20
1.3 What Phenomenology Is Not 23
1.3.1 Phenomenology Is Not Introspection 24
1.3.2 Phenomenology Is Not About Qualia 26
1.3.3 Phenomenology Is Not Anti-Naturalistic 28
1.4 Why We Need Phenomenology in Explaining
Consciousness 31
1.5 Conclusion 36
Notes 37
References 37
2 Naturalizing Phenomenology Reconsidered 42
2.1 Introduction 42
2.2 Three Views on the Naturalization of Phenomenology 44
viii Contents
2.3 Mathematization 49
2.4 A Question of Constraints 52
2.4.1 Conceptual Constraints 52
2.4.2 Isomorphism 57
2.4.3 Neurophenomenology and Generative Passages 60
2.4.4 Neurophenomenology and Homeomorphism 63
2.5 Conclusion 66
Notes 67
References 67
3 Models of Explanation in Cognitive Science 72

3.1 Introduction 72
3.2 Scientific Explanation—Background 73
3.3 Models of Explanation in Cognitive Science 75
3.3.1 Deductive-Nomological Model of Explanation 75
3.3.2 Personal Explanations 78
3.3.3 Functional Explanations 83
3.3.4 Dynamical Explanations 87
3.3.5 Mechanistic Explanations 94
3.4 Explanatory Integration of Cognitive Science 110
3.4.1 Theoretical Reduction, Unification, and Integration 111
3.4.2 Mechanistic Integration of Cognitive Science 114
3.4.3 Autonomy 116
3.5 Is Phenomenology Explanatory? 118
3.5.1 Types of Phenomenological Explanation 119
3.5.2 Phenomenological Understanding 124
3.6 Conclusion: Toward Integration of Phenomenology With Multilevel
Mechanistic Explanation 127
Notes 129
References 129
PART II
Phenomenology and Mechanism: In Search of Constraints 137
4 Phenomenology and Functionalism 139

4.1 Introduction 139
4.2 Husserlian Phenomenology and Computational Functionalism 140
4.3 The Notion of Function and the Idea of Functional Phenomenology
in Husserl 145
Contents ix
4.4 Functional Analysis and Phenomenological Decomposition 148
4.4.1 Phenomenological Decomposition 150
4.5 Toward Functional-Mechanistic Naturalization 153
4.5.1 Example 1: First-Person Perspective 156
4.5.2 Example 2: Vision Studies 158
4.6 Conclusion 162
Notes 163
References 163
5 Phenomenology and Dynamical Modeling 167

5.1 Introduction 167
5.2 Integrating the Dynamical and Mechanistic Frameworks 167
5.3 Neurophenomenology and Dynamical Systems Theory 171
5.4 Discovering the Dynamics of Experience—Epilepsy Study 174
5.5 Micro-Phenomenology—Toward Diachronic Models of
Experience 180
5.6 Conclusion: Toward Phenomenologically Informed Dynamical
Constraints 184
Notes 188
References 188
6 Conclusion: Toward Methodologically Guided

Mutual Constraints 193
6.1 Constraining Phenomenology 195
References 199
Index 201
Figures
I.1 The fable of the blind men and an elephant 2

3.1 Watt’s centrifugal governor 89
3.2 Model of the perceptual categorization agent 92
3.3 Mechanistic explanation 97
3.4 Levels of mechanisms 106
5.1 Phase space of the (A) nonepileptic and
(B) epileptic brain model 179
5.2 Spatiotemporal evolution of migraine visual aura 1 186
5.3 Spatiotemporal evolution of migraine visual aura 2 187
Acknowledgments
I wrote a significant part of this book in my hometown, Warsaw, during

the COVID-19 pandemic lockdown. The pandemic changed my sabbati-
cal plans, but because I have been working with phenomenology and cog-
nitive science for some years now, I managed to discuss some of the main
topics of the book with numerous people before the outbreak. I would
like to thank Sorin Bangu, Michael C. Baumgartner, Anthony Chemero,
Piotr Kozak, Katarzyna Kuś, Piotr Litwin, Veli-Pekka Parkkinen, Witold
Płotka, Nicolas De Warren, Konrad Werner, and Michał Wierzchoń
for inspiring discussions and valuable advice. Special thanks to Shaun
Gallagher and Dan Zahavi, who influenced my way of thinking about
phenomenology and cognitive science, although I suppose they will find
many points they disagree with in this book. I would also like to thank
my students at the University of Warsaw who attended my lectures on
the naturalization of phenomenology and my seminar on explanations
in cognitive science for their insightful comments on these topics. I am
very grateful for thoughtful feedback on the early versions of the man-
uscript from Marcin Miłkowski, Mateusz Hohol, Carlos Zednik, Piotr
Suffczyński, and Rasmus Rosenberg-Larsen. I am also grateful to Søren
Overgaard and Andrew Wackenmann from Routledge for supporting
this book project, and many thanks to Kamil Lemanek for improving the
book’s English. Finally, I would like to express my deepest gratitude and
love to Paula for her patience and continuous support.
Chapters 1, 2, and 3 include some material from my paper
“Phenomenology and Mechanisms of Consciousness: Considering
the Theoretical Integration of Phenomenology with the Mechanistic
Framework,” (2019), Theory & Psychology, 29(5), 601–619. doi.
org/10.1177/0959354319868769
An early version of Chapter 4 appeared as “Phenomenology
and Functional Analysis: A Functionalist Reading of Husserlian
Phenomenology,” (2020), Phenomenology and the Cognitive Sciences,
19, 869–889. doi.org/10.1007/s11097-020-09694-y
Work on this book was supported by Grant No. 2017/27/B/HS1/00735
financed by the National Science Centre, Poland.
Epigraphs
For children in philosophy, this may be the dark corner haunted by

the spectres of solipsism and, perhaps, of psychologism, of relativism.
The true philosopher, instead of running away, will prefer to fill the
dark corner with light.
—Husserl (1969, p. 237)
What is characteristic of philosophy is not a special subject-matter,

but the aim of knowing one’s way around with respect to the subject-
matters of all the special disciplines.
—Sellars (2007, p. 370)
Introduction
Despite remarkable recent progress in neuroscience, such as research

on functional and structural networks in the brain and the human con-
nectome project, explaining consciousness and first-person experience
remains a hard nut to crack. Neuroscientists measure the activity of the
working brain, connectivity, and the biochemistry of neural cells and
detect events and patterns of neural synchronization, yet they often lack
the resources to give them mental interpretations. Moreover, when con-
sciousness enters the stage, it is difficult to find consensus in such key
issues as its definition, suitable methods of investigation, and models of
explanation.
There is also a strong tendency in contemporary cognitive neurosci-
ence to avoid mentalistic terms and first-person methods of investiga-
tion, and thus to either ignore or reduce the first-person perspective to
observable intelligent behavior. On the one hand, it is an understand-
able tendency. After all, “hard” science focuses on what is observable and
measurable, or at least consequences that can be empirically tested. On
the other hand, it seems that by explaining consciousness purely from
a third-person perspective, we miss something important. A metaphor
for the current situation in the neuroscience of consciousness is the old
Indian fable about the blind men and the elephant (see Figure I.1). In the
fable, an elephant appears in a village. Six blind men hear of the occur-
rence and, having never encountered an elephant before, seek it out. As
they try to understand just what an elephant is, each of them touches a
different part of the animal, and each of them describes it completely dif-
ferently. All of them were right in some sense, and yet all of them were
fundamentally wrong; their individual descriptions fail to capture what
the elephant is as a whole. The third-person methods of neuroscience are
similarly incapable of grasping the key experiential aspect of conscious-
ness and are limited to studying surface phenomena. The fable reminds
us of our partial and limited perspectives, and it also points to the impor-
tance of trying to integrate them.
DOI: 10.4324/9781003035367-1
2 Introduction
Figure I.1 T
he fable of the blind men and an elephant. From World Stories for
Children (p. 14), by S. Woods, 1916, Ainsworth & Co. In the public
domain.
Avoiding the difficulties inherent to the first-person perspective by ignor-

ing it will not help us in actually understanding first-person experience
or consciousness as such, any more than closing our eyes would help us
describe an elephant. Such an approach seems immature, and, as Edmund
Husserl (1969) puts it, “for children in philosophy, this may be the dark
corner haunted by the spectres of solipsism and, perhaps, of psychologism,
of relativism. The true philosopher, instead of running away, will prefer to fill
the dark corner with light” (p. 237). Phenomenology is that light, a coura-
geous endeavor to investigate consciousness, giving justice to its complexity
without taking the “easy” path of reduction. That being said, it seems clear
to me that phenomenology does not have the resources to explain conscious-
ness and other multifaceted mental phenomena on its own.
This book tries to break the deadlock in consciousness studies by
showing how we could possibly incorporate the first-person perspec-
tive into the explanatory framework of cognitive science. The proposed
solution is inspired by two apparently opposed approaches to the mind
and consciousness. The first one derives from the phenomenological tra-
dition initiated by Husserl in the beginning of the 20th century, deliv-
ering a methodological toolbox for studying consciousness from the
first-person perspective. The second one is the neomechanistic approach,
which considers the mind’s cognitive functions the results of working
Introduction 3
neural and biological mechanisms. Obviously, there is significant tension
between these approaches. On the one hand, phenomenology, especially
Husserlian phenomenology, is skeptical of reductive naturalistic explana-
tions of consciousness. On the other hand, the mechanistic approach con-
ceives of first-person experience as a sort of troublesome addition to the
observable behavior of mechanisms and thus tends to ignore or reduce it.
Despite that, however, both positions can be revised and brought closer.
Phenomenological skepticism appears outdated in the light of new, eco-
logical and embodied, trends in cognitive science, and the mechanistic
approach, when considered in the context of the theoretical integration
of cognitive science, is far from “ruthless” reductionism. In this book, I
consider theoretical integration of phenomenology with mechanism and
its limits and benefits for cognitive science.
The idea of integrating phenomenology with a mechanistic approach
to explanation in cognitive science motivates rethinking the naturaliza-
tion of phenomenology from the perspective of the philosophy of sci-
ence. The naturalization of phenomenology, understood as its integration
with the explanatory framework of cognitive science (Roy et al., 1999,
pp. 1–2), has been under discussion for over two decades now (for an
overview see Petitot et al., 1999), but it is still a controversial topic. Even
the very possibility of such a project is controversial. Some reject it as
being contrary to the transcendental character of phenomenology (e.g.,
Moran, 2013); for others, its possibility is conditional on an appropri-
ate reformulation of phenomenology (e.g., Reynolds, 2017) and finding
methodological bridges making phenomenology compatible with the
naturalistic framework of cognitive science (e.g., Gallagher & Brøsted
Sørensen, 2006; Roy et al., 1999; Varela, 1996). Exploring the possibility
of naturalizing phenomenology requires both a deep understanding of
the explanatory framework of cognitive science and recognition of the
theoretical possibilities of integration. This is a clear weakness of the nat-
uralizing phenomenology debate; it does not address in detail the various
models of explanation applied in cognitive science, the various available
ideas for theoretical unification and integration of research fields, or how
phenomenology can ultimately fit into this picture. This book fills this
gap. The idea is to reconsider the project of naturalizing phenomenol-
ogy in the context of the recent debate on explanation and integration in
cognitive science. Thus, my considerations mainly concern methodology,
explanatory models, and the philosophy of science rather than specific
accounts of mental and cognitive phenomena, although I refer to them as
examples of selected explanatory strategies. From among the variety of
explanatory models applied in cognitive science, I focus especially on the
mechanistic model of explanation, which offers an integrative approach
to cognitive science (e.g., Craver, 2007). The main claim of this book
is that phenomenology, understood as a theory of and research method
concerning acts and objects of consciousness, can be integrated with the
4 Introduction
mechanistic framework of cognitive science, as it can provide constraints
on mechanistic models.
I defend the mechanistic approach to the integration of cognitive
science (Craver, 2007) because it allows one to integrate various fields
of research into one multilevel explanation of the target phenomenon.
I consider mechanistic integration nonreductive, at least where reduc-
tion is understood in the strong eliminativist sense or in the theoreti-
cal reductionist sense endorsed by the deductive-nomological model of
explanation. The mechanistic approach acknowledges that mechanisms
have different levels of organization that are conceptualized and studied
in different research fields with varying methodologies. The mechanisms
studied in the new mechanistic philosophy (Glennan, 2017) are far from
simple, responsive, deterministic, and context-independent machines. On
the contrary, the mechanisms responsible for mental phenomena are typi-
cally considered to exhibit dynamic and active behavior; have a complex
organization, often including stochastic processes; and be situated in the
body–environment context (e.g., Bechtel, 2008). Taking into account
the character of the mechanistic model of explanation, I find it a reli-
able approach to complex mental phenomena and a plausible candidate
framework for the integration of cognitive science.
In the literature, the mechanistic framework is typically thought of as
incompatible with the phenomenological one. For instance, it is argued
(e.g., Thompson, 2007) that mechanism is committed to the decompos-
ability of cognitive systems, whereas phenomenology, as applied in the
tradition of embodied and enactive cognition, rejects the decomposability
of cognitive biological systems in favor of dynamism. As I will argue, that
is a false dichotomy, and the relation between explanatory frameworks,
such as mechanism and dynamism, is much more complex and interest-
ing. An exception in thinking about phenomenology and the mechanistic
framework is Joseph Neisser’s (2015) book The Science of Subjectivity.
Neisser proposes a phenomenologically valid naturalistic account of
subjectivity (understood as the first-person perspective), which is based,
on the one hand, on phenomenological analyses of time-consciousness
and, on the other, on the evolutionary developmental framework and
theory of homologous mechanisms. In this perspective, phenomenology
is not opposed to naturalistic explanations; it can be complementary if
phenomenological categories are operationalized and reformulated in
neuro-psychological terms. Accordingly, “ecologically considered, sub-
jectivity serves an orienting function for an animal in the environment of
its concern” (Neisser, 2015, p. 3). Such a functionalization of a phenom-
enological category opens the possibility of functional decomposition of
the target capacity and mapping the subfunctions onto neurobiological
mechanisms developed in the course of evolution.
The perspective that I adopt in this book is more general. I am not try-
ing to give a naturalistic account of phenomenological subjectivity; I am
Introduction 5
trying to say what the theoretical and methodological grounds are for
integrating phenomenological analyses of consciousness with the mecha-
nistic framework of cognitive science. The key to pursuing that integration
is the constraints that various research fields, often founded in differ-
ent explanatory strategies, impose on the space of possible mechanisms
responsible for a given phenomenon (Craver, 2007). Such constraints can
be of various sorts. For example, they may provide specific information
about a mechanism’s components, their size, and their organization; they
may also be more general and concern the way we think of what the
hypothetical mechanism does, what functions it realizes, or the temporal
character of its activities (e.g., sequential, cyclical, etc.). In other words,
various constraints shape the space of possible mechanisms. Therefore,
the question of the possibility of integrating phenomenology with cogni-
tive science takes the form of the question of whether phenomenology
can provide such constraints. My cautious answer to this question is affir-
mative: phenomenological analyses may, in principle, provide constraints
on multilevel mechanistic models. Furthermore, I think that these con-
straints may be stronger than conceptual ones derived from descriptions
of first-person experience. First, I argue (Chapter 4) that, from a certain
point of view, phenomenological analyses may be considered analogous
to functional analyses and thus provide sui generis functional constraints.
Second, I argue (Chapter 5) that phenomenological methodology may
contribute to building dynamic models of mental phenomena and there-
fore provide dynamic or temporal constraints on mechanistic models.
If that is the case, and if phenomenology may provide such constraints,
then my proposal may be read as a new way to approach the naturaliza-
tion of phenomenology.
Outline of This Book

This book is written for both philosophers and scientists working in
various fields of cognitive science. Finding a common language between
phenomenology and empirical science required compromises and modi-
fying some of the presented positions. The order of the chapters presents
successive parts of the argument. The book consists of two parts. In the
first part, I introduce the theoretical background, discuss the concept of
phenomenology and review two key debates, namely, the naturalizing
phenomenology debate and the debate surrounding explanatory models
in cognitive science and the prospects for integrating cognitive science,
in particular those of mechanistic integration. Readers who are famil-
iar with phenomenology and the project of naturalization are invited
to skim the first and second chapters. And readers acquainted with the
explanatory models presented in the third chapter are welcome to skip
that part. In the second part, I elaborate on two possible ways of integrat-
ing phenomenology with the mechanistic framework, namely, first, that
6 Introduction
phenomenology can be read as a sort of functionalism and thus provides
functional constraints and, second, that phenomenology can contrib-
ute to dynamical models of mental phenomena, that is, models which
describe the dynamics of the behavior of the target system, and thus pro-
vide dynamical constraints on mechanistic models.
In Chapter 1, I introduce a moderate concept of phenomenology that
relies on Husserl’s (1977) idea of phenomenological psychology. I also
introduce the basics of phenomenological method and the key phenom-
enological category of intentionality. I argue that this weak notion of
phenomenology is amenable to naturalization understood as integration
with cognitive science. I also debunk some common myths concerning
phenomenology. In particular, I argue against reading phenomenology
as a form of introspectionism and against identifying phenomenology
with the problem of explaining the qualitative character of experience.
I also discuss the alleged anti-naturalism of phenomenology and argue
that although phenomenology was critical of reductive naturalistic
explanations of consciousness, it was open to integration with empiri-
cal psychology. I end this chapter with a brief elaboration of why we
need phenomenology in cognitive science, in particular, why we need it
in the field of consciousness studies, and why it is important to consider
naturalizing phenomenology, that is, integrating it with the explanatory
frameworks in use in cognitive science.
In Chapter 2, I review the naturalizing phenomenology debate
(Gallagher & Schmicking, 2010; Petitot et al., 1999). First, I explain the
basic idea of naturalism and the naturalization of phenomenology, fol-
lowed by a discussion of three general positions in the debate. I reject the
two extremes—that the naturalization of phenomenology is impossible
or that it requires a redefinition of the concept of nature—and defend
the moderate position between the two, according to which naturaliza-
tion is not only possible but also stands to provide benefits to both phe-
nomenology and cognitive science. These benefits are not simply mutual
inspirations but take the stronger form of constraints. Recognition of
the constraints requires establishing methodological common ground. I
use the notion of constraints as the main category in introducing sev-
eral proposals for naturalizing phenomenology, including front-loaded
phenomenology, formalization of phenomenological description, applied
ontology as semantic bridging, and neuro- and micro-phenomenology.
In a separate section, I address the naturalization of phenomenology
through mathematization. I argue that the types of constraints currently
being proposed are either weak or implausible. I conclude that we should
rethink the notion of constraints, as well as the very concept of natural-
ization, in the context of mechanistic integration of cognitive science.
In Chapter 3, I discuss various models of explanation applied in
cognitive science as well as the project of mechanistic integration. I
focus on several key explanatory frameworks: deductive-nomological
Introduction 7
explanations, personal explanations, functional explanations, dynami-
cal explanations, and, last but not least, mechanistic explanations and
consider their limitations. I argue that none of them is able to provide a
complete explanation of multifaceted mental phenomena. I defend the
position of integrative pluralism, according to which a multiplicity of
approaches to mental phenomena is valuable for explaining them. The
question is how to integrate results from various research fields. I intro-
duce the idea of mechanistic integration, according to which various
research fields contribute to the explanation of a target phenomenon by
providing constraints of different sorts on possible mechanistic models
(Craver, 2007). In the last part of this chapter, I consider whether phe-
nomenology presents its own model of explanation. I argue that phenom-
enology offers a constitutive understanding of mental phenomena rather
than providing a complete explanation. Finally, I consider how phenom-
enological analyses could contribute to multilevel mechanistic explana-
tions. I propose two hypotheses. Both argue that phenomenology can, in
principle, provide constraints on mechanistic explanatory models. The
first hypothesis states that the character of these constraints is functional;
the second hypothesis conceives such constraints as dynamical.
In the next two chapters, I elaborate on the two hypotheses with
respect to possible ways of integrating phenomenology with mechanism.
In Chapter 4, I consider the functional hypothesis and put forward a
new functional interpretation of phenomenology. First, I discuss the func-
tional-computational interpretation of Husserlian phenomenology pro-
posed by Hubert Dreyfus (Dreyfus & Hall, 1982) and Ronald McIntyre
(1986). I outline the deficiencies of Dreyfus’s and McIntyre’s positions
and propose to shift to a methodological perspective on the relation
between phenomenology and functionalism. I defend the claim that in
Husserl’s conception of “functional phenomenology” we find an original
notion of intentional function and that phenomenology offers a method
of decomposition, which can be conceived as analogous to the explana-
tory strategy of functional analysis (Cummins, 1975). I argue that the
proposed functionalist reading of phenomenology opens a new approach
to the integration of phenomenology with the mechanistic framework
of cognitive science. To illustrate how phenomenological analyses can
contribute to explanatory models, I discuss two examples: the phenom-
enology of vision (Madary, 2017) and the evolutionary-developmental
account of the first-person perspective (Neisser, 2015).
In Chapter 5, I discuss the second, dynamical hypothesis. I begin by
assessing the relation between dynamical and mechanistic approaches to
cognition. I defend the claim that, despite different assumptions, these
frameworks can be seen as complementary rather than exclusive and
argue that the dynamic mechanistic model of explanation is the new
emerging standard of explanation in cognitive science. Next, I evaluate
applications of the dynamical systems theory in neurophenomenology
8 Introduction
(e.g., Lutz, 2002; Varela, 1996) and micro-phenomenology (Petitmengin
et al., 2019) and argue that they are limited. In particular, I discuss neu-
rophenomenological studies of epilepsy (Petitmengin et al., 2007) and
confront their results with dynamical models of epileptic seizures in neu-
roscience. Finally, I consider possible revision of neurophenomenology
in order to provide stronger dynamical constraints and contribute to a
dynamical-mechanistic model.
In the sixth and last chapter, I summarize the results of chapters one
through five and consider the consequences of the proposed integration
of phenomenology with mechanism. In particular, I discuss the opposing
direction of the constraining relation, namely, how dynamic mechanistic
models constrain phenomenology. I argue that we can use these mod-
els to test phenomenological theories. Moreover, dynamical-mechanistic
models and simulations can reveal features of the target phenomenon
inaccessible to apprehension via phenomenological method, and thus,
they can be seen as extensions of the phenomenological toolkit.
References
Bechtel, W. (2008). Mental mechanisms: Philosophical perspectives on cognitive
neuroscience. Routledge.
Craver, C. F. (2007). Explaining the brain: Mechanisms and the mosaic unity of
neuroscience. Clarendon Press.
Cummins, R. (1975). Functional analysis. The Journal of Philosophy, 72(20),
741–765.
Dreyfus, H. L., & Hall, H. (1982). Introduction. In H. L. Dreyfus and H. Hall
(Eds.), Husserl, intentionality and cognitive science (pp. 1–27). MIT Press.
Gallagher, S., & Brøsted Sørensen, J. (2006). Experimenting with phenomenol-
ogy. Consciousness and Cognition, 15(1), 119–134.
Gallagher, S., & Schmicking, D. (Eds.). (2010). Handbook of phenomenology and
cognitive science. Springer.
Glennan, S. (2017). The new mechanical philosophy. Oxford University Press.
Husserl, E. (1969). Formal and transcendental logic (D. Cairn, Trans.). Springer.
Husserl, E. (1977). Phenomenological psychology (J. Scanlon, Trans.). Martinus
Nijhoff.
Lutz, A. (2002). Toward a neurophenomenology as an account of generative pas-
sages: A first empirical case study. Phenomenology and the Cognitive Sciences,
1(2), 133–167.
Madary, M. (2017). Visual phenomenology. MIT Press.
McIntyre, R. (1986). Husserl and the representational theory of mind. Topoi,
5(2), 101–113.
Moran, D. (2013). ‘Let’s look at it objectively’: Why phenomenology cannot be
naturalized. Royal Institute of Philosophy Supplements, 72, 89–115.
Neisser, J. (2015). The science of subjectivity. Palgrave Macmillan.
Petitmengin, C., Navarro, V., & le van Quyen, M. (2007). Anticipating seizure:
Pre-reflective experience at the center of neuro-phenomenology. Consciousness
and Cognition, 16(3), 746–764.
Introduction 9
Petitmengin, C., Remillieux, A., & Valenzuela-Moguillansky, C. (2019).
Discovering the structures of lived experience: Towards a micro-phenomeno-
logical analysis method. Phenomenology and the Cognitive Sciences, 18(4),
691–730.
Petitot, J., Varela, F. J., Pachoud, B. & Roy, J.-M. (Eds.). (1999). Naturalizing phe-
nomenology: Issues in contemporary phenomenology and cognitive science.
Stanford University Press.
Reynolds, J. (2017). Phenomenology, naturalism and science: A hybrid and hereti-
cal proposal. Routledge.
Roy, J.-M., Petitot, J., Pachoud, B., & Varela, F. J. (1999). Beyond the gap: An
introduction to naturalizing phenomenology. In J. Petitot, F. J. Varela, B.
Pachoud, & J.-M. Roy (Eds.), Naturalizing phenomenology: Issues in contem-
porary phenomenology and cognitive science (pp. 1–80). Stanford University
Press.
Sellars, W. (2007). Philosophy and the scientific image of man. In K. Scharp & R.
B. Brandom (Eds.), In the space of reasons: Selected essays of Wilfrid Sellars.
Harvard University Press.
Thompson, E. (2007). Mind in life: Biology, phenomenology, and the sciences of
mind. Harvard University Press.
Varela, F. J. (1996). Neurophenomenology: A methodological remedy for the
hard problem. Journal of Consciousness Studies, 3(4), 330–349.
Woods, S. (1916). World stories for children. Ainsworth & Co.
Part I
Integrating Phenomenology
with Cognitive Science
1 The Concept of Phenomenology
1.1 Introduction
In this chapter, I elaborate a moderate concept of phenomenology that
is amenable to integration with the explanatory framework of cognitive
science. I focus on Husserlian phenomenology, which in my view is the
most advanced and methodologically aware version of phenomenology.
In the broadest sense, phenomenology is the philosophical investigation
of first-person experience and subjectivity. Edmund Husserl is considered
to be the founder of the phenomenological movement, which includes
such important figures as Heidegger, Merleau-Ponty, and Sartre, and
developed an advanced philosophical approach driven by the reductive
method (phenomenological reduction). For Husserl, phenomenology was
a fundamental discipline that adopts the transcendental attitude; that is,
it investigates transcendental subjectivity and the conditions of possibil-
ity for experience as such.1 The concept of phenomenology I propose in
this chapter is weaker and adjusted to the aims of this book; it relies on
Husserl’s idea of phenomenological psychology and takes phenomenol-
ogy primarily as a theory of consciousness and cognitive acts, playing
down its transcendental aspect. Because our topic concerns the natural-
ization of phenomenology and its integration with cognitive science, I
mainly discuss methodological issues. As I argue, the proposed concept of
phenomenology is compatible with the naturalistic framework of cogni-
tive science.
The chapter is structured as follows: First, I introduce selected aspects
of Husserlian phenomenology, such as reduction and intentionality, and
elaborate the distinction between the static and the genetic approaches. I
also introduce the idea of phenomenological psychology, which occupies
the space between transcendental phenomenology and empirical psychol-
ogy and thus seems open to integration with naturalistic frameworks.
Second, I refute some common myths about phenomenology. I argue
against interpretations of phenomenology that see it as a form of intro-
spectionism that stress its alleged anti-naturalism and suggest that phe-
nomenological analysis concerns the qualitative character of experience.
DOI: 10.4324/9781003035367-3
14 Integrating Phenomenology
Finally, I explain why we need phenomenology in cognitive science, why
we need it in the field of consciousness studies in particular, and why it is
important to consider integration.
1.2 What Is Phenomenology
In this book, when discussing phenomenology, I mainly refer to the philo-
sophical tradition started by Edmund Husserl in the beginning of the
20th century. Of course, there were other important figures in the phe-
nomenological movement, but I find Husserlian phenomenology to be
the most advanced version of the position, the most driven by method-
ological considerations; hence, it fits well into the discussion surrounding
the methodological grounds for naturalization—granted, that is not to
say that it is the easiest version of phenomenology to naturalize. By natu-
ralization of phenomenology, I understand its methodological integration
with an explanatory framework for cognitive science (e.g., Roy et al.,
1999). I will return to the issue of naturalization in Chapter 2.
Phenomenology, especially Husserlian phenomenology, is a highly
advanced and sophisticated philosophy, and I cannot reasonably intro-
duce it here in its entire complexity, nor can I address all its important
contributions to various philosophical debates. That would certainly be
beyond scope of this book (for an overview of Husserlian phenomenol-
ogy, see, e.g., Landgrebe 1981, Moran, 2005; Zahavi, 2003). I focus on
methodological issues and those issues in phenomenology that can find
application in cognitive science. Thus, I hardly address the transcendental
level of phenomenological investigation. That level is also played down
in my considerations because, as I will argue, successful integration with
cognitive science requires it. That said, I do not deny that transcendental
phenomenology is an important philosophical endeavor, though separate
from the project of naturalization. I take phenomenology primarily as a
theory of consciousness and mental processes, as “the eidetic theory of
lived-processes” (Husserl, 1980, p. 38), which investigates the structure
of experience. This understanding of phenomenology is incorporated in
the idea of phenomenological psychology (Husserl, 1977), which I elab-
orate later. With respect to the phenomenological theory of conscious-
ness, I discuss mainly methodological topics relevant to naturalization
and integration with cognitive science, as well as selected issues such as
intentionality and consciousness of time, which are among what makes
phenomenology a research field of interest to cognitive science.2
1.2.1 Reduction and Variation as Phenomenological Methods

According to Husserl, creating a scientific approach to first-person
phenomena can be achieved only through the development of a rigor-
ous and intersubjective method of studying structures of experience.
The Concept of Phenomenology 15
Phenomenological method, in the broadest sense, relies on an iterative
process of reducing common beliefs and scientific knowledge about the
external world and the mind, describing the subjective experience of vari-
kinda seems
unqualified? ous objects and analyzing it in order to identify its invariant structures
and regularities. The first step in phenomenological thinking is typically
called the reduction, and the subsequent phenomenological descriptions
and analyses may include the method of imaginative variation. I intro-
duce them in turn.
1. REDUCTION Husserl actively developed the method of reduction for many years
and introduced several variants, such as reduction to the sphere of own-
ness, specially designed to address the problem of intersubjectivity (e.g.,
Husserl, 1960), or the eidetic reduction applied in phenomenological
psychology (Husserl, 1977). Furthermore, the theory of transcendental
reduction is involved in Husserl’s broader epistemological project, which
was thought to solve the problem of grounding the central belief of our
natural attitude, namely our belief in the world’s existence (e.g., Vincini,
2020). A difficulty in introducing different types of phenomenological
reduction is that Husserl often did not use clear-cut distinctions and that
their application overlaps. In the following, I focus only on the basics of
reduction, the understanding of which is necessary to apprehend the idea
of phenomenology amenable to naturalization.
Although phenomenological method is usually termed phenomeno-
logical reduction, it might be useful to distinguish two of its components:
epoché and the reduction (e.g., Gallagher & Zahavi, 2012; Zahavi,
2003). As Zahavi (2003) writes,
it is necessary to distinguish the two: The epoché is the term for our
abrupt suspension of a naïve metaphysical attitude, and it can con-
sequently be likened to a philosophical gate of entry (Hua 6/260). In
contrast, the reduction is the term for our thematization of the cor-
relation between subjectivity and world.
(p. 46)
Epoché can be seen as a necessary first step in initiating phenomenologi-

cal reflection. Its objective is to change our cognitive attitude from natu-
ral and naïve, in which the world is simply out there, filled with material
things, to the phenomenological attitude, in which objects of experience
are taken as phenomena appearing to consciousness (e.g., Husserl, 1982,
pp. 57–62). In order to do it, as Husserl puts it, “we put out of action
the general positing which belongs to the essence of the natural attitude”
(p. 61); namely, we suspend the general claim that the world exists. It
is important to note, however, that phenomenological exclusion of that
general claim is not an act of negating the world, nor does it remove the
world from the scope of phenomenological interests. On the contrary, the
world and how it appears to consciousness are the primary concerns of
phenomenology. In other words, the suspension of the natural attitude is
necessary to start analyzing how intentional functions of consciousness
produce the experience of the world.
Next to this “putting out of action” of that general claim, there is the
reduction or “suspension” of scientific and folk knowledge related to the
natural world and worldly objects. The point is not, however, to hold
that all our sciences are wrong. Our natural tendency is to see the world
through a scientific lens. Those judgments are bracketed in order to see
how objects appear without already being shaped by scientific and folk
concepts. The result of this reductive procedure is the phenomenological
domain of experience, where things are no longer objects out there in the
world but apprehended as objects for consciousness, that is, phenomena.
For example, a perceived apple tree is no longer considered a material
object, a plant of a certain species, growing in front of me in the garden.
Following reduction, the object is considered as perceived, as it perceptu-
ally appears to the subject, as experienced in specific modalities (vision,
smell), and, more important, how it is perceptually apprehended: through
successive adumbrations, from a certain point of view relative to the sub-
ject’s body and so on. Correlations between the structure of perceptual
consciousness and modes of appearing are what phenomenology wants
to investigate. In other words, reduction excludes from experience those
judgments that constitute the natural attitude, and by doing so, it enables
analysis of the very process of the constitution of experience.
An obvious objection is that the application of such a radical method
is impossible. We cannot simply set aside our knowledge and our deeply
rooted cognitive attitudes. This is certainly true. As Merleau-Ponty (2005)
puts it, “the most important lesson which the reduction teaches us is the
impossibility of a complete reduction” (p. XV). The reduction should be
understood as a normative and methodological rule—an ideal toward
which we aim and which tells us how phenomenological analysis should
proceed. But reduction is always incomplete, and the procedure needs to
be repeated every time one begins to analyze their experience.
As a result of applying phenomenological reduction, we obtain a field
of phenomena that can be analyzed for correlations and invariant struc-
tures. For the purpose of analysis, Husserl (1973) introduces another
method called imaginative variation. This procedure is crucial for so-
called essential seeing (para. 87), that is, disclosing the essence or, to put
it differently, the invariant structure of experience. According to Husserl
(1973), this operation
is based on the modification of an experienced or imagined objec-

tivity, turning it into an arbitrary example which, at the same time,
receives the character of a guiding ‘model’, a point of departure for
the production of an infinitely open multiplicity of variants. … We
produce free variants, each of which, just like the total process of
variation itself, occurs in the subjective mode of the ‘arbitrary.’ It
then becomes evident that a unity runs through this multiplicity.
(pp. 340–341)
In other words, the variation relies on imaginary modifications and

replacing parts of experienced objects; what remains unchanged through
these modifications can be treated as its invariant structure. The unity
that Husserl aims to apprehend thanks to this procedure is the “essence”
of the considered objectivity. It is important to notice that, accord-
ing to Husserl (1973), the meaning of essence should “be free from all
metaphysical interpretations” (p. 341). Imaginative modifications of an
experienced object inform us of how the object is constituted by various
activities of consciousness. In fact, this process of constituting experience
is the main explanatory aim of phenomenology. Interestingly, Husserl
(1980, pp. 44–45) sometimes considers imaginative variation a method
of phenomenological thought experiment on a par with the methods of
experimental psychology. In this sense, a phenomenologist experiments
with their experience in order to discover the generic structures of this
type of experience and the eidetic laws that govern it.
Although Zahavi (2003) argues that “both epoché and reduction can
consequently be seen as elements of a transcendental reflection” (p. 46), it
seems that the application of phenomenological method can be delimited
so that it does not have to lead to considering subjectivity transcendental.
First, it is worth noting that phenomenological method has several lev-
els of application as well as different degrees of completion (aiming for
the ideal of absolute reduction). The transcendental level is, so to speak,
the deepest and the most profound level, addressing fundamental philo-
sophical issues, such as investigating the subjective conditions of pos-
sibility for the world’s appearance. But we can apply phenomenological
method locally in order to describe and analyze experiences of a certain
kind, such as the visual experience. We can investigate how the experi-
ence is structured, for example, that it is perspectival, temporal, and so
on, without making any commitments regarding the transcendental sta-
tus of subjectivity. Second, separating the methodological steps of reduc-
tion from transcendentalism is important in the context of integrating
phenomenology with empirical research. This idea was already discussed
by Husserl (1977) in his project of phenomenological psychology, which
was to be a science located between transcendental phenomenology and
empirical psychology (Reynolds, 2017; Thinès, 1977). Phenomenological
psychology is not transcendental, yet it applies some aspects of phenom-
enological method and investigates the essence of mental states and acts
(Kockelmans, 1987; cf. Płotka, 2021). As I show in the following, the
task of phenomenological psychology is to perform “useful preliminary
work” (Husserl, 1980, p. 39) and deliver a theoretical framework (con-
cepts, typology of cognitive acts, etc.) to experimental psychology.
1.2.2 Intentionality
To see how phenomenological reduction works and what kind of analy-
sis it enables, we need to introduce one of the key structural features
of consciousness, namely, intentionality, that is, consciousness’s “about-
ness” or directedness toward objects. A major part of Husserl’s inves-
tigations concerned various forms of intentional apprehensions, for
example, perceptual, imaginative, or signitive. In the first book of Ideas,
after applying phenomenological reduction, Husserl analyzes the general
structure of intentionality and claims that intentional experience consists
of correlated noetic and noematic moments, which he calls, in short, a
noesis and noema (see Husserl, 1982, para. 87–127). Noetic moments
are “really inherent” components of mental processes, whereas noematic
moments are intentional correlates. To put it differently, noetic moments
are really inherent acts of consciousness (e.g., acts of perceiving, recall-
ing, imagining, etc.), whereas noematic moments are intentional contents
or meanings. Husserl emphasizes (1982, para. 98) that both, the noetic
and the noematic, are “non-self-sufficient”; they belong to each other in
the unity of lived experience. Multiple noematic moments amount to the
“full noema,” that is, different content, in which, according to Husserl,
we can distinguish a “noematic nucleus” or “noematic sense,” which is
the unit of meaning that remains the same across various mental acts
apprehending the same object. The rest of the noematic moments depend
on the mode of intentional apprehension of the object (e.g., perceptual,
linguistic, etc.). For example, I can apprehend the Palace of Culture and
Science perceptually, through imagination, or refer to it in a linguistic
expression when I say, “I was on the 40th floor of the Palace of Culture
and Science yesterday.” All these acts of apprehension, by means of a
fixed “noematic core,” refer to the same object, namely, to a historical
building in Warsaw, yet they all differ in some respect (due to the noe-
matic moments attached to the core and determined by specific modes
of apprehension).
This is how, in brief, Husserl analyzes the intentional structure of acts
of consciousness. It is important to notice that we are not aware of these
noetic and noematic moments in the natural attitude, which is the default
mode of our everyday activities. They are components of mental processes
in virtue of which consciousness refers to objects but can be decomposed
and analyzed by applying phenomenological reduction and attaining
the phenomenological attitude. In virtue of phenomenological reduction
and analysis, we can study types of noemata as well as different noetic
functions and ultimately create a general formal theory of noemata. The
theory would cover the typology of noetic functions and their noematic
correlates, as well as the eidetic laws governing these correlations, build-
ing a unity of mental processes (1982, para. 93). In Chapter 4, I return to
the issue of noema and discuss its functionalist-representational reading.
Three central ideas are given
particular attention: the
phenomenological reduction,
phenomena, and essence. -studying The Concept of Phenomenology 19
consciousness
1.2.3 Static and Genetic Approaches
Husserl offers two complementary approaches to studying consciousness in
which a phenomenologist can investigate structures of mental phenomena,
namely, the static and the genetic approaches. The static approach is related
to the active intentional constitution of objects of experience. By active level
of constitution, Husserl understands all processes that are related to cogni-
tive activities and are accessible to awareness, so, for example, all contents
that are attentively intended in perception and expressed in acts of judgment.
Analysis of intentional acts in terms of noetic and noematic correlations is
an example of the static approach. Another example of static analysis can
be found in Husserl’s (1997) lectures titled Thing and Space, where he ana-
lyzes how spatial orientation is correlated with bodily motility. According
to Husserl (1973), the static approach, which describes the constitution of
an intentional object as “the product of an objectivating operation of the
ego” (p. 72), should be supplemented by a genetic investigation explaining
the passive processes accompanying and influencing the acts of the ego.
The passive level of constitution concerns processes that are not explicit
components of cognitive acts but nevertheless determine the contents of
experience. For example, that includes the processes of different sorts of
associations, affectivity, kinesthesis, habitual determinations, and previous
experiences of the subject (see, e.g., Welton, 1982). So genetic phenomenol-
ogy enriches the static picture of hyletic content shaped by noetic activity. It
discloses that under noetic functions there are motivational and associative
functions, which also contribute to how and what we experience.
Importantly, Husserl considered the genetic approach explanatory, in
contrast to the static approach, which he considered descriptive. As he
writes in a short work titled Static and Genetic Phenomenological Method,
[i]n a certain way, we can therefore distinguish “explanatory” phe-

nomenology as a phenomenology of regulated genesis, and “descrip-
tive” phenomenology as a phenomenology of possible, essential
shapes (no matter how they have come to pass) in pure conscious-
ness and their teleological ordering in the realm of possible reason
under the headings, “object” and “sense.” In my lectures, I did not say
“descriptive,” but rather “static” phenomenology. The latter offers
an understanding of intentional accomplishment, especially of the
accomplishment of reason and its negata. It reveals to us the gradu-
ated levels of intentional objects that emerge in founded appercep-
tions of a higher level as objective senses and in functions of sense
giving, and it reveals to us how they function in them, etc.
(Husserl, 2001a, p. 629)
In other words, static phenomenology describes the contents of expe-

rience and its intentional formations through intentional functions,
whereas genetic phenomenology reveals the genesis of these functions
including their passive, pre-reflective, determinations.
The shift from static to genetic analysis in Husserl’s work can be seen
in the development of his model of time-consciousness (Husserl, 1991;
for an overview, see de Warren, 2009; Kortooms, 2002). A simple static
model of time-consciousness consists of intentional functions engaged in
the production of a unitary experience of a temporal object (e.g., a mel-
ody). Generally speaking, the experience of a temporal object is produced
by three functions of consciousness: (1) retention (retaining the parts of
the object which are no longer present), (2) protention (anticipating the
parts of the object which are about to become present), and (3) the “pri-
mal-” or “ur-impression” (sensory receptivity in the present moment).
According to this model, we do not experience the momentary “now” of a
temporal object as sensorially stimulating (the ur-impression is a product
of idealization); rather, the lived experience of the “now” is a product of
all these functions, which means that the sensory core of the “now” is
always accompanied by retentional-protentional formations. This static
model was later developed by Husserl and supplemented with the genetic
approach (Kortooms, 2002, pp. 175–284). For instance, he introduced
consideration of the issues of affectivity and embodiment to his concep-
tion of time-consciousness and showed their contribution to the passive
constitution of the experience of the present. Accordingly, bodily motil-
ity and affectivity involve internal dynamics that shape our intentional
experience in the present moment (Pokropski, 2015). I will return to this
issue in Chapter 4 and argue that the phenomenological model of time-
consciousness exhibits the method of phenomenological decomposition.
1.2.4 Phenomenological Psychology
Husserl often contrasted empirical psychology with transcendental
phenomenology and called for a revision of the former. As he writes in
“Philosophy as Rigorous Science,”
actually sufficient empirical science of the psychical in its relations to

nature can be realized only after psychology is built upon a systematic
phenomenology, thus if the essential formations of consciousness and
its immanent correlates, inquired into and fixed in systematic con-
nection by means of pure seeing, provide the norms for the scientific
sense and content of the concepts of all manner of phenomena, hence
of the concepts with which the empirical psychologist expresses the
psychical itself in his psychophysical judgments.
(Husserl, 2002, pp. 276–277)
In later works, Husserl came up with the idea of an eidetic and pure psy-
chology, also called phenomenological psychology, which was thought
to be the science mediating between transcendental phenomenology and
empirical psychology (Husserl, 1971, 1977; see also Reynolds, 2017;
Thinès, 1977). Accordingly, phenomenological psychology was thought to
‘‘supply the essential insights needed to give meaning and direction to the
research presented under the title ‘empirical psychology’’ (Kockelmans,
1987, p. 6). These insights concern the eidos of consciousness’ acts
expressed in the “scientific concepts of internality” (1977, p. 166), which
cannot be obtained in an inductive and naturalistic approach but are
apprehended in intuition.
Husserl (1977) characterizes phenomenological psychology as a priori,
eidetic, grounded in intuition or pure description, and focusing on inten-
tionality. Phenomenological psychology is an a priori science because it
is primarily interested in “all those essential universalities and necessities,
without which psychological being and living are simply inconceivable”
(Husserl, 1977, p. 33). To put it differently, phenomenological psychology
investigates generic structures of experience, such as the perspectival char-
acter of visual perception. It is eidetic because “only subsequently does it
proceed to the explanation of psychological facts, to theory, precisely their
eidetic explanation, which is naturally for us the first interest” (Husserl,
1977, p. 33) or, in other words, its objective is to explain psychological
facts not in terms of the particular mental states of an individual but in
terms of discovered generic structures of consciousness. Next, intuition
(eidetic seeing) designates the source of the a priori. Finally, intentionality
states that consciousness is always a consciousness of something. Thus,
investigations under the heading of phenomenological psychology con-
cern a twofold aspect: acts of consciousness and the related objects of
experience or, to put it in phenomenological terminology, the noetic and
the noematic. I will return to the issue of the explanatory force of phe-
nomenological “eidetic explanation” in Chapter 3. For now, it is impor-
tant to remember that phenomenological psychology is a nonempirical,
eidetic investigation of how the human mind functions and that it pro-
vides experimental psychology with key concepts, such as intentionality.
It is also important to remark that phenomenological psychology applies
a reduction that is different from the transcendental reduction of tran-
scendental phenomenology. Husserl calls it “eidetic reduction” (1971) or
“phenomenological reduction” (1977). Eidetic reduction relies on exclud-
ing (suspending) all reference to the physical basis of the mental and all
scientific prejudices concerning the physical and the mental. In this way,
the reduction opens the sphere of the purely psychological to intentional
analyses and imaginative variation. The results of these methods are then
described, conceptualized, and applied in empirical psychology. According
to Joseph Kockelmans (1987), phenomenological psychology, although it
employs the reduction, remains in the natural attitude (p. 20). He argues
that “the purpose of the phenomenological-psychological reduction is
not to bring the transcendental subjectivity to light. Phenomenological
psychology hopes to expose only the foundations of empirical psychology”
(Kockelmans, 1987, p. 21). Eidetic reduction makes it possible to see and
describe the structures of experience, such as noesis and noema, which in
the normal attitude are hidden. It can be applied to psychological research,
and it does not have to lead to transcendentalism.3 Phenomenological
analysis is therefore theoretical work preceding experimental research.
Husserl (1980, pp. 44–45) also suggests that psychological experiments
can be useful for phenomenology; thus, the relation can be considered
reciprocal. When we acknowledge this phenomenological middle ground,
phenomenology appears to be complementary to psychology rather than
opposed to it. As we will see, the difference between transcendental and
eidetic reduction is also of key importance for considering the integration
of phenomenology with contemporary cognitive science.
In the third book of Ideas, Husserl (1980) gives an example of how
empirical psychology and phenomenology are related to each other in
the study of perception:
The exploration of such psychic states, called perceptions, as states

of actual real individuals of the actual world, is a matter for psy-
chology, for inductive experiential science. … Perceptions, however,
can be explored not only as factually existing states in the nexus of
factual unities of consciousness, belonging to factual psycho-physical
individuals in the factual world, irrespective of whether in the singu-
lar individual case or in experiential-scientific universality; rather, we
can undertake an “eidetic reduction,” exclude all questions about real
factual existence, about the judgment-positing of the latter, and carry
through the attitude of purely eidetic investigation. We concern our-
selves then with the eidos, the essence “perception,” and with what
belongs to a “perception as such,” as it were to the sense, ever the
same, of possible perception in general. … We therefore differentiate
the “possible” perceptions in general according to basic types; for
each one we ask what belongs to it essentially and what it requires
according to its essence as necessarily belonging to it, what changes,
transformations, connections it makes possible purely through its
essence, whether with phenomena of the same sort or with those of
another sort, etc.
(p. 35)
Phenomenological psychology analyzes the eidos, the structure of percep-

tual acts, and in this way, it delivers concepts to experimental psychology.
These concepts include the general notion of perception, as well as typo-
logical concepts of different sorts of perceptual acts. Such analyses are the
theoretical work preceding experimental research.
For phenomenological psychology, the notion of motivation is of key
importance. Husserl introduces the notion of motivation to explicate
quasi-causal and lawful relations between mental states (e.g., Husserl,
1977, p. 108; 1989a, para. 56). An example of “motivation of reason” is
when a perceptual state motivates some belief, which, in turn, can moti-
vate a subject to be in a specific emotional state or to perform a particular
activity. The concept of motivation is applied to both active and passive
constitution. In the former, it concerns how different sorts of conscious
perception, beliefs, and attitudes lead to other beliefs or actions. The lat-
ter sheds light on how pre-egoic (unconscious or pre-reflective) states,
such as affections and associations, affect conscious apprehension of
objects and related expectations. Husserl also uses the notion of motiva-
tion to describe how external objects affect the subject. As he writes,
the subject comports itself toward the Object, and the Object stimu-
lates or motivates the subject. The subject is subject of an undergoing
or of a being-active, is passive or active in relation to the Objects
present to it noematically, and correlatively we have ‘effects’ on the
subject emanating from the Objects. The Object ‘intrudes on the sub-
ject’ and exercises stimulation on it (theoretical, aesthetic, practical
stimulation).
(Husserl, 1989a, p. 231, emphasis in the original)
Although Husserl claims that there is a “lawful regularity” in motivational

relations, it is weaker than the relation of physical causality. There is no
certainty that a mental state will always motivate the same set of other
mental states. Particular mental states are effects of a motivational nexus,
which consists of interrelated mental states with reference to personal his-
tory and habitual formations as well as the context of a particular situa-
tion. To put it simply, our mental life is too complex and context-dependent
to explicate it in strict causal or formal laws; nevertheless, Husserl formu-
lates ceteris paribus or eidetic laws governing subjective experience.
To sum up, phenomenological psychology occupies the space between
empirical psychology and transcendental phenomenology. For empirical
psychology it is foundational; it provides a theoretical framework and con-
cepts that do justice to the complex structure of consciousness. Such phe-
nomenological analyses of consciousness depend on a weaker version of
eidetic reduction and thus can be separated from the transcendental level
of consideration. But phenomenological psychology may be a prolegom-
enon, a preliminary exercise, which may evolve into a transcendental posi-
tion if it employs a more stringent reduction (Kockelmans, 1987, p. 22).
1.3 What Phenomenology Is Not

The concept of phenomenology requires clarification of a few misunder-
standings. First, phenomenology is unfortunately often confused with intro-
spection, which is a rather surprising misinterpretation given that Husserl
explicitly criticizes introspection as unreliable in his works. Second, phe-
nomenology is not concerned with the issue of the qualitative aspect of
consciousness, or, at least, it is not primarily so. Husserl was more interested
in generic structures of consciousness. The third misunderstanding directly
concerns the issue of naturalization and is related to Husserl’s critique of the
natural sciences. As I will argue, although Husserl was critical of naturalistic
explanations of consciousness, his position is not antinaturalistic.
1.3.1 Phenomenology Is Not Introspection

One of the biggest misunderstandings in the reception of phenomenol-
ogy in philosophy of mind is reading it as a form of introspectionism
(e.g., Dennett, 1991; Rupert, 2009). It is argued that if phenomenology
relies on introspection, then phenomenological insights are subjective
and unreliable, sharing the weaknesses of introspective method. The fact
of the matter is that although phenomenological method shares some
similarity with introspection, after all, both approaches are about expe-
riences and their objects, they should not be identified as one and the
same. A careful reader can even find passages in Husserl’s works where
he explicitly distinguishes phenomenology from introspective observa-
tion. For instance, in the third volume of his Ideas, Husserl (1980) writes:
Let us consider the state of affairs more closely. This is all the more
necessary since at the present time the naturalism predominating so
greatly among psychologists, as among all natural scientists, has as
its consequence an almost universal misunderstanding of the sense
of phenomenology and of its possible achievements for the psycho-
logical science of experience. With this is connected the basically per-
verted view that with phenomenology it is a matter of a restitution
of the method of inner observation or of direct inner experience in
general. Only in this way also are explained those superficial (indeed
not even superficial, because not understanding at all the sense of the
matters) literary rejections of the claim that phenomenology makes,
and must make through its own specific character, of paving the way
for a reform of psychology (as also, on the other hand, of philoso-
phy) that in the literal sense is fundamental and novel.
(p. 33)
Husserl not only opposed the introspectionist reading of phenomenol-

ogy but was, in general, also critical of introspection and methods of
self-observation, which, as he argues (1982, pp. 181–190), are unreliable
and focused only on individual experiences.4 On the contrary, phenom-
enology is primarily interested in generic structures or, to use Husserl’s
terminology, in the essences of various types of experience. For example,
when we phenomenologically investigate the essence of perception, we
differentiate the ‘possible’ perceptions in general according to basic
types; for each one we ask what belongs to it essentially and what
it requires according to its essence as necessarily belonging to it,
what changes, transformations, connections it makes possible purely
through its essence, whether with phenomena of the same sort or
with those of another sort, etc.
(Husserl, 1980, p. 35)
To use Husserl’s metaphor, a phenomenologist is like a geometer interested

in the general properties of figures and not in particular shapes drawn on
the board. In order to grasp these general structures, phenomenology
makes use of the method of imaginative variation, which excludes from
investigation particularities and enables “essential seeing.” Importantly,
the application of the imaginative method shows that in some sense the
phenomenological method is more similar to thought experiments than
to introspection.
One could argue that phenomenology was not a sort of introspection
in Husserl’s own opinion but that it may be understood as such from the
point of view of contemporary conceptions of introspectionism. I do not
think this is true either. Even when we take into account recent formu-
lations of introspection, phenomenology does not meet its criteria. For
example, Eric Schwitzgebel (2019) identifies three minimal conditions
for a process to be introspective. First, the mentality condition states
that introspection generates knowledge of mental states and processes.
Second, the first-person condition specifies that this knowledge is only
of one’s own mind. Third, the temporal proximity condition, introduced
due to research on memory fallibility, narrows the scope of introspec-
tive judgment to currently ongoing or immediately past mental processes.
Phenomenology only meets the first condition, as it provides knowledge
of our mental lives. Contrary to the second condition, phenomenologi-
cal knowledge does not just concern one’s own mind and particular
mental states. Through the application of phenomenological reductive
and imaginative methods, we are able to consider generic structures of
experience, which are common to other experiencing subjects as well.
Moreover, phenomenology does not fall under what Gualtiero Piccinini
(2009) calls privatism, that is, a conception according to which first-per-
son or introspective data are private and unverifiable. Phenomenologists
acknowledge that our mental states are private only in the sense that it
is the subject who lives through their experiences from the first-person
perspective. No one else can experience those particular states from that
perspective. But those experiences are not private in the sense that they
are unobservable to other subjects. On the contrary, our mental lives
are observable to others from the third-person perspective, as they are
expressed in our bodily behaviors (Krueger & Overgaard, 2012; Smith,
2010). In fact, Husserl (1960) argued that we are able to know other’s
mental states through special cognitive acts of empathy (Einfühlung).
Finally, phenomenological knowledge about our mental lives is not lim-
ited per the third condition, that is, to the narrow temporal window of
the present. It is true that phenomenologists often describe their experi-
ences lived in the current moment, but allow me to stress again that the
aim of doing so is to analyze these experiences and capture their generic
structures, including their temporal genesis. Furthermore, we can apply
phenomenological method to our past experiences as well, and, in this
way, study the intentional structure of acts of memory.
1.3.2 Phenomenology Is Not About Qualia

Phenomenology in contemporary analytical philosophy of mind and cog-
nitive science has come to be identified with describing the qualitative
character of consciousness, which is captured by the notion of qualia—
the famous “what it’s like” of experience (e.g., Chalmers, 1996; Nagel,
1974). Accordingly,
we can say that a mental state is conscious if it has a qualitative

feel—an associated quality of experience. These qualitative feels are
also known as phenomenal qualities, or qualia for short. The prob-
lem of explaining these phenomenal qualities is just the problem of
explaining consciousness.
(Chalmers, 1996, p. 10)
To put it differently, consciousness is a sort of inner awareness of qualia,

which is paradoxical in its nature and thus problematic or even impos-
sible to naturalistic explanation (Levine, 2001). That said, there have
been attempts at providing a scientific explanation of such phenomenal
properties. For example, an influential approach to consciousness called
integrated information theory (IIT; e.g., Oizumi et al., 2014; Tononi,
2004) aims to be not only a theory of consciousness, including its phe-
nomenal character, but also a method of measuring consciousness and a
research program for searching for the mechanisms responsible for the
phenomenon of consciousness. The notion of phenomenology used in IIT
is very broad, as synonymous with qualitative experience or conscious-
ness in general. Importantly, however, IIT attempts to give a quantita-
tive informational account of qualitative states. The main claim of IIT is
that “consciousness has to do with the capacity to integrate information”
(Tononi, 2004, p. 2). The idea behind it is that consciousness is basically
the ability to integrate and generate information, that is, to differentiate
various perceptual states and to integrate them in one coherent percept.
According to Tononi (2004), generating information is a common feature
of many living and artificial systems. Integrating information is, however,
not so common and is the key aspect of consciousness. The greater the
amount of integrated information, the higher the degree of conscious-
ness. The notion of a quale is understood by IIT’s proponents “as a shape
that embodies the entire set of informational relationships generated by
interactions in the system” (Balduzzi & Tononi, 2009, p. 2).5
For Chalmers, the phenomenal character of consciousness is what
makes consciousness hard to explain for cognitive science. Others, how-
ever, criticize the notion of qualia and reject its importance in explana-
tory pursuits (e.g., Dennett, 1991). For example, Neisser argues that the
problem of qualia should be separated from the issue of subjectivity. As
he writes,
we will not have a complete scientific explanation of consciousness as

long as the problem of qualia persists. But I argue that inner aware-
ness is not the essence of subjective thought and that the first-person
perspective can be treated independently of the problem of qualia.
(Neisser, 2015, p. 6)
It seems to me that this sort of separation is also acknowledged in clas-

sical phenomenology, although for different reasons. Husserl clearly
separates the qualitative dimension of experience from the intentional.
Although the former is an indispensable aspect of our mental lives, only
the latter is key to explaining consciousness. In Ideas I, Husserl describes
the hyletic (sense-data) layer of experience as “senseless stuff,” which is
shaped by intentional (noetic) functions giving meaning to the object of
experience. He distinguishes hyletic-phenomenological considerations
from noetic-phenomenological ones, which are “incomparably more
important and richer” (Husserl, 1982, p. 207). For Husserl, analysis and
categorization of the noetic functions is necessary for explaining the key
feature of consciousness, namely how consciousness can intentionally
apprehend objects. In his later works (e.g., in Ideas II), Husserl adopted
the genetic approach and modified his position, emphasizing the impor-
tance of the hyletic foundation of consciousness. He was especially inter-
ested in affective and bodily, especially kinesthetic, sensations. But even
then, he was interested in the general constitutive character of affectivity
and embodiment, not the quality of particular sensations. Contemporary
positions that rely on the phenomenological tradition, not neglecting that
our experiences have a qualitative dimension, separate phenomenality,
understood as qualia, from phenomenality, understood as the first-person
subjective perspective. For example, Zahavi and Kriegel (2015) argue that
“experiential for-me-ness is not a quality or datum of experience on a par
with, say, the taste of lemon or the smell of crushed mint leaves. In fact, it
is not supposed to be any specific qualitative content at all,” and further-
more, “one does not grasp for-me-ness by introspecting a self-standing
quale”; rather, “one grasps such experiential elements as lemon-qualia
and mint-qualia by appreciating what varies across such phenomenal
characters, … grasps what for-me-ness is by appreciating what remains
constant across them” (p. 38). I think this quotation perfectly illustrates
the idea behind phenomenological studies and their aims, which is to
capture an invariant structure of experience of a certain kind, not its
particular qualitative character.
1.3.3 Phenomenology Is Not Anti-Naturalistic

One of the key controversies surrounding the naturalization of phenom-
enology concerns the allegedly antinaturalistic character of Husserlian
phenomenology. Contemporary phenomenologists like to emphasize
Husserl’s critique of naturalism and his contrasting phenomenology with
natural science (e.g., Moran, 2013; Zahavi, 2013). Accordingly, phe-
nomenology is in critical opposition to the natural sciences, which are
committed to the naturalistic attitude. Phenomenology, on the contrary,
endorses the phenomenological attitude and is committed to transcen-
dentalism, which investigates the very possibility of knowledge, including
scientific knowledge. If that is the case, then naturalizing phenomenology,
that is, integrating it with natural science, is impossible. I find this opin-
ion exaggerated. In my view, Husserl’s approach to naturalism and the
natural sciences was not so straightforwardly critical. First, notice that
Husserl distinguishes naturalism, as a philosophical position, from the
natural sciences. As he writes,
when it is actually natural science that speaks, we listen gladly and

as disciples. But it is not always natural science that speaks when
natural scientists are speaking; and it assuredly is not when they are
talking about ‘philosophy of Nature’ and ‘epistemology as a natural
science.’
(Husserl, 1982, p. 39)
Husserl clearly acknowledges the importance of the natural sciences

and is impressed by their progress, especially in physics. As he argues
in Philosophy as Rigorous Science, “in the whole of modern life there is
perhaps no idea that is more powerful, whose advance is less resistible
than that of science” (Husserl, 2002, p. 256). But the idea of modern sci-
ence is problematic too. It involves a conception of infinite progress, that
is, a never-ending process of accumulating knowledge, spanning genera-
tions (e.g., Husserl, 1989b). Thus, both formal and natural sciences are
assumed to be flawed and imperfect because they are just a stage in the
infinite progress of scientific reason.6 The development of natural science
was possible, according to Husserl, thanks to the application of a mathe-
matical method. But, although modern mathematization provided a vital
impulse for the development of natural science, it also contributed to the
crisis of human rationality (Husserl, 1989b). This fundamental problem
was compounded by the fact that, the incredible success of the natural
sciences led to the popularization of naturalism, a philosophical position
as common in Husserl’s day as it is today.
A naturalist conceives every object of study as a part of nature, gov-
erned by strict laws and describable in the formal language of mathemat-
ics. For Husserl (2002), it is a naïve position because it assumes that
“the nature into which it wants to inquire is simply there for it” (p. 257).
Moreover, it creates the illusion of one universal scientific method that
may be applied to any domain. Husserl is especially interested in critiqu-
ing naturalism in psychology. Roughly speaking, the naturalization of
consciousness is problematic for two reasons.
The first refers to the naturalistic assumption, which Husserl rejects,
that we can describe our mental lives with a finite list of formal or natural
laws in an analogous manner to mathematics or physics. It is true that
Husserl (1982, pp. 14–16) does also use nomological categories when
speaking about phenomenology, but the laws that phenomenology dis-
covers are of a different nature than the laws of the natural sciences—
they are eidetic laws. Furthermore, although Husserl uses a mathematical
analogy in thinking about phenomenology and its relation to the natural
sciences, he sees its limitations. Phenomenology should not be consid-
ered a mathematics of consciousness, because it is a different type of
exact science (e.g., Husserl, 1980, p. 38). Phenomenology is, similarly
to mathematics, an eidetic science, but contrary to mathematics, which
investigates “formal essences,” it investigates “material essences,” that
is, essences apprehended in intuition, which in Logical Investigations,
Husserl (2001b) calls “inexact essences” (Investigation 3, para. 9). These
“inexact essences” require a different methodology and conceptual appa-
ratus. In order to explicate them, phenomenology uses descriptive notions
that differ from the idealized notions of mathematics. Mathematics is
a deductive science, whereas phenomenological insights are grounded
in intuition, that is, in the essential seeing of structures of experience.
Husserl (1982, para. 72–73) also doubts that we can recognize a closed
set of phenomenological axioms, analogous to the axioms of geometry,
because the variety of the forms of mental phenomena is potentially infi-
nite. That said, although Husserl indicates essential differences between
mathematics and phenomenology, he is not opposed to applying some
mathematical terms to phenomenological investigations. As a trained
mathematician, he frequently used quasi-mathematical formal notations
and graphs to visualize structures of experience, for example, temporal
structure (e.g., Yoshimi, 2007). As I will argue, the mathematization of
phenomenological descriptions in terms of dynamical systems theory is a
potential way toward naturalization.
The second aspect of naturalism in psychology, also relevant today,
which Husserl resisted is the tendency in experimental psychology to
explain mental phenomena in purely physical or biological terms. A
naturalist conceives of mental states as biological facts that can be stud-
ied experimentally and explained in causal terms. As Husserl (1999)
writes in The Idea of Phenomenology, “here I remind the reader of the
favorite ploy of grounding epistemology on a psychology of knowl-
edge or on biology. In our day, reactions against these fatal prejudices
are frequent. And prejudices they are indeed” (p. 20). In short, accord-
ing to Husserl, consciousness is not merely another object in the world,
thus reductive naturalistic approaches to it are flawed. It is important
to understand, however, that Husserl does not neglect that the mind is
founded in the physical world. As he writes, “the existence of mental
realities, of a real mental world, is bound to the existence of a nature in
the first sense, namely that of material nature, and this not for accidental
but for fundamental reasons” (Husserl, 1980, p. 104). And in the second
book of Ideas (Husserl, 1989a), he defends the key role of embodiment
in the constitution of perceptual experience. Moreover, he discusses the
ingestion of santonin, which results in an altered perception of colors
(Husserl, 1989a, pp. 67–68), and speculates about the physical processes
in the brain correspondent with these mental processes (pp. 172–173).
It seems, therefore, that Husserl was not opposed to empirical studies in
psychology and finding psychophysical dependencies, but only opposed
to the reductive approach that tries to explain mental phenomena purely
in physiological terms.7
Although critical of reductive naturalistic explanations of conscious-
ness, Husserl acknowledged the importance of experimental psychology.
As he writes in the introduction to the first book of Ideas,
my criticism of psychological method did not at all deny the value of

modern psychology, did not at all disparage the experimental work
done by eminent men. Rather it laid bare certain, in the literal sense,
radical defects of method upon the removal of which, in my opinion,
must depend an elevation of psychology to a higher scientific level.
(Husserl, 1982, p. XVIII)
And in the third book of Ideas, Husserl (1980) argues that “experimen-
tal psychology should not be abandoned, but rather made incompara-
bly more fruitful through the phenomenological founding” (p. 42). This
higher scientific level of psychology and the fruitfulness of its research
can be achieved by integrating experimental psychology with what
Husserl (1971, 1977) calls phenomenological or eidetic psychology. As
shown earlier, phenomenological psychology would be an a priori science
that provides experimental psychology a systematic theoretical frame-
work, including a typology of intentional processes and their composi-
tion (Husserl, 1971, p. 79).
To sum up, the Husserlian view of naturalism is twofold. On the one
hand, it acknowledges the importance and achievements of the natural
sciences, including empirical psychology. On the other hand, it recog-
nizes the naturalistic attitude as naïve, because it assumes the existence
of nature as simply “ready-made” and encourages the application of its
method in studying every phenomenon, including consciousness. These
naturalistic fallacies result in attempts at reductive explanations of men-
tal phenomena, which Husserl is opposed to. Thus, he calls for rethinking
the naturalistic attitude and revising empirical psychology in the light of
phenomenology. In particular, he develops the project of phenomenologi-
cal psychology, which he considers a means to integrating phenomenol-
ogy with empirical psychology.
1.4 Why We Need Phenomenology in Explaining Consciousness

In this section, I consider phenomenology in the context of consciousness
studies and argue that we need phenomenology for several reasons. In
thinking about contemporary attempts to explain consciousness, it is use-
ful to refer to the distinction proposed by Chalmers (1996, 2010) between
the hard and easy problems of consciousness. The easy problems concern
cognitive capacities, such as perception, categorization, or reportability of
internal states. They are still difficult to explain, but at least researchers
have an idea of the methods of investigation and models of explanation
that can be applied to these phenomena. One possible model of explana-
tion relies on describing the mechanisms, for example, neural mechanisms,
responsible for the target cognitive function. The hard problem is related
to the question, “Why is the performance of these functions accompanied
by experience?" (Chalmers, 2010, p. 8) or, to put it differently, why the
performance of these functions is related to some qualitative feel or, for
short, qualia. It is an enigmatic issue because we have neither a good sci-
entific theory of phenomenal consciousness nor a consensus concerning
which model of explanation to apply. Thus, the problem of experience is
often referred to as the explanatory gap (Levine, 1983).
According to Chalmers (2010), there are several strategies we can
adopt to cope with the hard problem, including two opposites: (1) simply
denying that there is a problem, that is, we can deny that first-person
experience is something that science should explain (e.g., Dennett, 1991),
or (2) claiming that we can explain the phenomenal character of con-
sciousness in a scientific way (e.g., Balduzzi & Tononi, 2009). But there
is another position, one that I defend in this book. It acknowledges that
the hard problem is too hard for contemporary science and thus that we
should focus on easy problems, the explanations of which are within our
reach, and hope that they will shed new light on approaching the hard
problem.
The revival of interest in Husserlian phenomenology in the 1990s was
closely related to interest in consciousness, which reemerged in science and
philosophy after years of neglect. But can phenomenology help in finding
solutions to the hard and easy problems of consciousness? Which strategy
toward the hard problem is adopted in the naturalization of phenomenol-
ogy debate? The editors of Naturalizing Phenomenology argue that filling
the “explanatory gap” is the main motive behind naturalization. According
to their argument, naturalistic cognitive science is incapable of explaining
the phenomenal, qualitative character of consciousness, and “the attempt
to naturalize Husserlian phenomenology might usefully be seen as an
attempt to close this explanatory gap” (Roy et al., 1999, p. 3). Varela (1996)
explains the objective of his version of naturalized phenomenology in a
similar manner, a position called neurophenomenology, which is thought
to be a “remedy for the hard problem.” I find these positions misleading for
that reason: They suggest that phenomenology holds the key to solving the
hard problem of consciousness. I do not believe that phenomenology on its
own has the resources to approach the hard problem of consciousness. It
only has access to one side of consciousness, that is, to the experiential men-
tal side without a clear connection to the other physical side. Furthermore,
as I argued earlier, phenomenology, especially Husserlian phenomenology,
is not about the problem of qualia but primarily about the structure of
consciousness. In other words, phenomenology is primarily interested in
the generic structures of experience and not in the particular qualities that
fill these structures. Moreover, phenomenological reduction suspends claims
about the natural causes of consciousness. Thus, phenomenology cannot
help answer the question of why and how the qualia of experience emerge
from the brain. In my view, phenomenology can, however, contribute to
solving the easy problems, which, after all, are not that easy. That is not to
say that phenomenology completely abandons the problem of experience.
Instead of trying to answer why there is some qualitative feel accompanying
our experiences, it delivers a nontrivial concept of consciousness and meth-
ods for describing and analyzing the structure of first-person experience.
Describing and defining an explanandum, that is, the phenomenon we
want to explain, especially when it concerns such enigmatic phenom-
ena as consciousness and first-person experience, is not an easy task.
Three common mistakes in attempting to do so are misidentification,
underspecification, and taxonomic errors (Craver, 2007; Irvine, 2013).
Examples of taxonomic errors include incorrect categorizations, for
example, when a phenomenon is thought to be singular and homoge-
neous when, in fact, it is a collection of several distinct phenomena or,
vice versa, when several phenomena are thought to be under investiga-
tion while actually being one multifaceted phenomenon. An example of
the first case is research on memory, which was initially thought to be a
single phenomenon and is now considered a collection of various phe-
nomena, for example, the ability to remember bodily skills or episodic
recollections of childhood, which have different underlying mechanisms
(Schacter, 1996). Underspecification is when a multifaceted phenomenon
is characterized superficially, with the omission of some of its key aspects.
It is often found in bottom-up theories built on empirical research, in
which the concept of consciousness is narrowed down to a single capac-
ity such as attention or perceptual awareness. Another example of
underspecification is when we define the target phenomenon in terms
of dispositions without identifying its conditions of occurrence, modula-
tion, and inhibition. Finally, the error of misidentification concerns situa-
tions where scientists try to explain a phenomenon that, in fact, does not
exist. Historically, these errors usually resulted in the elimination of the
misidentified phenomenon from science, as was the case with phlogiston.
According to Irvine (2013), all three types of errors are being made in
consciousness studies, including the most serious one, namely, misiden-
tification. Her solution is to eliminate the concept of consciousness from
scientific research and replace it with a collection of precisely defined
and operationalized cognitive functions. Although it is possible that there
is no single, homogeneous phenomenon behind consciousness and that
there are many phenomena that amount to being a conscious-minded
being, I believe that elimination is too quick a move. Instead of eliminat-
ing the concept of consciousness from cognitive science, we should try to
make it more precise and recognize its various forms. Phenomenology,
with its first-person methodology, can help with that and provide us with
a better understanding of what we actually want to explain.
Recently, there has also been growing interest in second-person meth-
ods of phenomenological interview, which may also deliver fine-grained
descriptions of experience. In particular, second-person phenomenologi-
cal interviews proved their value in studies of complex and multifac-
eted phenomena to which researchers may not have first-person access,
such as epilepsy (Petitmengin et al., 2007), schizophrenia (Parnas et al.,
2005), depression (Ratcliffe, 2014), psychotic states (e.g., Sass, 2014),
drug addiction (Moskalewicz, 2016), sensory substitution (Kałwak et al.,
2018), and music absorption and cerebral palsy (Høffding & Martiny,
2016). An advantage of phenomenological interview is the incorporation
of the reductive method taken from Husserlian tradition, that is, bracket-
ing or suspending, as much as is possible, both scientific and folk knowl-
edge about the mind and focusing on lived experience itself. Descriptive
categories are usually elaborated in a second-person fashion that helps
express various aspects of experience that are otherwise difficult to expli-
cate in scientific, for example, medical, terminology. And using key phe-
nomenological categories, such as sense of bodily agency and ownership,
affectivity, and intentionality, helps express and understand, for both the
subject and the researcher, the features of the studied experience. The
objective of the phenomenological interview method does not have to
be limited to describing first-person experience for scientific research. It
could also be directed toward improved self-cognition, which often has a
therapeutic effect, which is itself an important part of phenomenological
practice (e.g., Høffding & Martiny, 2016; Petitmengin et al., 2007).
Phenomenological methods can provide not only first- and second-
person descriptions of experience but, more important, a method of
analysis that also allows for capturing its structural properties. The phe-
nomenological literature, including the classic works of Husserl (e.g.,
1982, 1989a, 1991), is rich in analyses revealing the conditions of the
constitution of various types of experience and their invariant structures.
As I argue in Chapter 4, phenomenological analyses of mental phenom-
ena share similarities with functional explanatory strategy (Cummins,
1975), and thus, they may provide characterizations of explananda that
are useful for formulating hypotheses and guiding further empirical
research.
Contemporary second-person phenomenological methods also offer
advanced methods for analyzing described experiences. For example,
Petitmengin et al. (2019) propose methods of synchronic and diachronic
analysis. The former’s objective is to describe the structure of experien-
tial content at a given moment in time; the latter concerns the evolution
of experience in time, for example, how the intensity or modality of a
particular experiential state changes over time. In Chapter 5, I argue that
diachronic analysis may provide an insightful characterization of experi-
ence dynamics and contribute to building dynamical models.
As an example of the potential contribution of phenomenology to con-
sciousness studies, we can focus on the importance of phenomenology in
explaining abnormal experiences related to mental maladies, although
the point extends to other multifaceted mental phenomena. Without
first- or second-person descriptions of a patient’s experience, we can-
not capture the key symptoms of a given mental malady, and without
phenomenological analysis, we cannot capture its structural properties,
on the basis of which we can formulate hypotheses and heuristics to
guide treatment and empirical research (e.g., Moskalewicz et al., 2018;
Ratcliffe, 2014; Sass, 2014). It is also argued that a complete explanatory
model of a mental malady, such as schizophrenia, should include multi-
ple aspects, including psychological, phenomenological, neurobiological,
social, and anthropological factors (e.g., Engel, 1977). Can phenomeno-
logical descriptions and analyses contribute to such an explanation? On
the one hand, Sass (2014) argues that the phenomenological model of
schizophrenia is explanatory. According to that model, schizophrenia is
a disturbance of the “minimal self” that involves malfunctions such as
hyper-reflexivity (exaggerated self-consciousness resulting in the objecti-
fication of experiences that are lived through implicitly in normal condi-
tions) and diminished self-affection (a decline in one’s sense of existing
as a unified subject; Sass, 2014, p. 368). Sass uses a phenomenological
conceptual apparatus, mainly taken from Husserlian and Heideggerian
tradition, and argues for the autonomy of such phenomenological expla-
nations. On the other hand, according to Matteo Colombo and Andreas
Heinz (2019),
phenomenological descriptions, analyses, or interpretations may not
constitute explanations. After all, phenomenology is often character-
ized as a purely descriptive enterprise distinct from explanation. But unspecific
they can still constrain and inform causal, computational, or other
types of explanations of mental maladies. In fact, phenomenologi-
cal results are often used to clarify the structure of the experiences
involved in mental maladies, to interpret experimental results, and to
inspire hypotheses for further research.
(p. 700)
I agree with the latter view. Phenomenological models of a mental mal-

ady may provide a structural characterization of the studied malady; they
may also reveal aspects that a standard third-person scientific approach
may ignore or leave unnoticed. That being said, phenomenological mod-
els lack explanatory power because in their pure form, they have nothing
to say about the causal mechanisms responsible for the relevant distur-
bances of the self and experience. Similarly, psychological explanations
of schizophrenia in terms of beliefs (e.g., Campbell, 2002) are incomplete
because they capture only one aspect of the malady. It seems, therefore,
that the best strategy for building a complete explanation of a mental
malady is integrating results from various fields of research. so true
Only an inte-
grative approach can do justice to the true complexity of these kinds of
maladies, because it promises to account for various aspects, including
not only the causal-biological but also the psychological and phenom-
enological. An example of an integrative framework for mental mala-
dies is the mechanistic property cluster proposed by Kenneth Kendler,
Peter Zachar, and Carl F. Craver (2011). Accordingly, a mental malady
is a cluster of interrelated causes, properties, and symptoms produced
by underlying causal mechanisms. A complete explanation integrates
multiple aspects of the malady in question and describes the relations
between various causal factors and produced mental states, which, in
turn, produce symptoms.
I believe that the discussion surrounding explanations of mental maladies
tells us something important about the desired model of explanation for
consciousness in general. We see that multifaceted phenomena, both those
of mental disorders and those which amount to consciousness, require an
integrative approach rather than one specific explanatory framework. As
I show in Chapter 3, every kind of scientific explanation utilized in cog-
nitive science has its limitations, which proves especially troublesome in
the case of explaining consciousness. For example, the covering-law model
relies on the subsumption of the explanandum phenomenon under a law
of nature, but no such laws are acknowledged in psychology or biology.
Causal-mechanistic explanations in neuroscience tend to ignore the level
of first-person experience and mentality in general. It also seems clear that
we need some reliable phenomenological approach to capture all of the
important aspects of consciousness, including ones accessible from the first-
person perspective. Phenomenology is a genuine science of the mental, but
it is also limited because it cannot account for causal relations relevant to
consciousness. However, phenomenological analysis of first-person experi-
ence may provide information about the structural properties of experience
(including abnormal experiential states), which allows for making hypothe-
ses about underlying processes. If that is the case, then the role of phenome-
nology in explaining consciousness is more than descriptive; it may provide
hypotheses, but, more important, it may provide constraints on empirical
research in other fields. The character of these constraints is discussed in
subsequent chapters. Naturally, the idea of reciprocal constraints between
phenomenology and empirical research in cognitive science relates to natu-
ralization. In the next chapter I introduce the idea of naturalizing phenom-
enology as well as various constraining relations proposed in the debate.
1.5 Conclusion
In this chapter, I introduced a basic concept of phenomenology that is
amenable to integration with a naturalistic framework. My concept of
phenomenology is based on Husserlian phenomenological philosophy, in
particular on his idea of phenomenological psychology, but it downplays
the transcendental character of his philosophy (in the next chapter, I elab-
orate more on why downplaying transcendentalism is necessary for the
project of naturalization). I take phenomenology primarily as a method
of investigation and theory of consciousness. However, as I argue, a phe-
nomenological theory of consciousness is not about its qualitative charac-
ter, and phenomenological method is not introspection. Phenomenology
investigates the structure of consciousness, for example, various forms of
intentionality, and the genesis of conscious experience, including passive
and pre-reflective processes of constitution. The concept of phenomenol-
ogy I endorse is not anti-naturalistic, although it is in opposition to reduc-
tive naturalistic explanations of consciousness. As I have argued, Husserl’s
view of naturalism is twofold. On the one hand, he explicitly opposes the
naturalistic reductive approach to consciousness. On the other hand, he
acknowledges the importance of empirical psychology but calls for its revi-
sion and integration with phenomenology. I discussed Husserl’s project of
phenomenological psychology, which was thought to provide empirical
psychology with phenomenological concepts of consciousness, including
the concepts of intentionality and intentional functions. Finally, I argued
that integrating phenomenology with cognitive science, also known as
naturalizing phenomenology, is important for studies of consciousness.
I concluded that phenomenology can deliver informative descriptions of
first-person experience and analyses of the structure of experience, and in
doing so, it may provide constraints on other fields. This possibility opens
the way to integrating phenomenology with cognitive science.
integrating
phenomenology with
cognitive science =
naturalizing
phenomenology?
Notes
1 Kantian examples of transcendental conditions of experience are time and
space as forms of all sensory intuitions.
2 Another important aspect of the phenomenological theory of consciousness,
and one which I am not addressing in detail in this book, is embodiment
(e.g., Husserl, 1989; for an overview of Husserl’s theory of embodiment, see,
e.g., Taipale, 2014). Husserl’s and Merleau-Ponty’s conceptions of embodi-
ment are key contributions to cognitive science, in particular to the field of
embodied cognition (see, e.g., Gallagher, 2005). Although I agree with the
embodied approach to cognition, I am not going to discuss it in detail for
two reasons. First, it is simply too vast a topic to include in a book mainly
concerned with methodology and the philosophy of science. Second, I take
it for granted, because most of the approaches to cognition that I discuss in
this book acknowledge, more or less, that cognition is embodied and situated
in the environment. Naturally, there are more issues in Husserlian phenom-
enology of potential interest to cognitive science, to mention only a few—
intersubjectivity (Husserl, 1960; for an overview of Husserl’s conception of
intersubjectivity, see, e.g., Zahavi, 2001), memory and image consciousness
(Husserl, 2005), and affectivity (for an overview, see, e.g., Lotz, 2007).
3 According to Płotka (2021), the clear distinction between transcendental phe-
nomenology and phenomenological psychology is problematic because, as he
argues, Husserl operates with clear-cut methods.
4 Note that Merleau-Ponty emphasizes the difference between introspection
and phenomenology in a similar manner. In Phenomenology of Perception he
writes, “this phenomenal field is not an ‘inner world,’ the ‘phenomenon’ is not
a ‘state of consciousness,’ or a ‘mental fact,’ and the experience of phenomena
is not an act of introspection or an intuition in Bergson’s sense” (Merleau-
Ponty, 2002, p. 66). For him, the idea of internal perception expresses a return
to the conception of the “immediate data of consciousness” and therefore is
“a hopeless enterprise.”
5 For a critique of IIT’s axiomatic method and explanation of consciousness,
see Pokropski (2018, 2019).
6 Notice that Husserl was even more critical of philosophy than natural science.
As he writes, “I am not saying that philosophy is an imperfect science; I am say-
ing quite simply that it is still not a science, that it has yet to begin as science”
(Husserl, 2002, p. 250). Thus, the motivation behind Husserl’s project of phe-
nomenological philosophy is to create a truly scientific and rigorous philosophy.
7 It is also important to notice when and to whom Husserl is addressing his
critique, namely, to such figures as Wundt and others, who applied a com-
bination of psychophysical and physiological methods in the early days of
experimental psychology (Kockelmans, 1987).
References
Balduzzi, D., & Tononi, G. (2009). Qualia: The geometry of integrated infor-
mation. PLOS Computational Biology, 5(8), 1–24. doi:10.1371/journal.
pcbi.1000462
Campbell, J. (2002). Rationality, meaning and the analysis of delusion. Philosophy,
Psychiatry and Psychology, 8, 89–100.
Chalmers, D. J. (1996). The conscious mind: In search of a fundamental theory.
Oxford University Press.
Chalmers, D. J. (2010). Facing up to the problem of consciousness. In The char-
acter of consciousness (pp. 3–27). Oxford University Press.
Colombo, M., & Heinz, A. (2019). Explanatory integration, computational phe-
notypes, and dimensional psychiatry: The case of alcohol use disorder. Theory
& Psychology, 29(5), 697–718.
741–765. doi:10.2307/2024640
Dennett, D. C. (1991). Consciousness explained. Little, Brown.
de Warren, N. (2009). Husserl and the promise of time: Subjectivity in transcen-
dental phenomenology. Cambridge University Press.
Engel, G. (1977). The need for a new medical model: A challenge for biomedicine.
Science, 196(4286), 129–136.
Gallagher, S. (2005). How the body shapes the mind. Oxford University Press.
Gallagher, S., & Zahavi, D. (2012). The phenomenological mind (2nd ed.).
Routledge.
Høffding, S., & Martiny, K. (2016). Framing a phenomenological interview: What,
why and how. Phenomenology and the Cognitive Sciences, 15(4), 539–564.
Husserl, E. (1960). Cartesian meditations: An introduction to phenomenology (D.
Cairns, Trans.). Kluwer.
Husserl, E. (1971). “Phenomenology” Edmund Husserl’s article for the
Encyclopaedia Britannica (1927): New complete translation by Richard E.
Palmer. Journal of the British Society for Phenomenology, 2(2), 77–90.
Husserl, E. (1973). Experience and judgment: Investigations in a genealogy of
logic (J. S. Churchill & K. Ameriks, Trans.). Routledge & Kegan Paul.
Nijhoff.
Husserl, E. (1980). Phenomenology and the foundations of the sciences: Third
book of Ideas pertaining to a pure phenomenology and to a phenomenological
philosophy (T. E. Klein & W. E. Pohl, Trans.). Martinus Nijhoff.
Husserl, E. (1982). Ideas pertaining to a pure phenomenology and to a phenom-
enological philosophy; First book: General introduction to a pure phenomenol-
ogy (F. Kersten, Trans.). Springer.
Husserl, E. (1989a). Ideas pertaining to a pure phenomenology and to a phenom-
enological philosophy; Second book: Studies in the phenomenology of consti-
tution (R. Rojcewicz & A. Schuwer, Trans.). Springer.
Husserl, E. (1989b). The crisis of European sciences and transcendental phenom-
enology: An introduction to phenomenological philosophy (D. Carr, Trans.).
Northwestern University Press.
Husserl, E. (1991). On the phenomenology of the consciousness of internal time
(1893–1917) (J. B. Brough, Trans.). Springer.
Husserl, E. (1997). Thing and space: Lectures of 1907 (R. Rojcewicz, Trans.). Kluwer.
Husserl, E. (1999). The idea of phenomenology (L. Hardy, Trans.). Martinus
Nijhoff.
Husserl, E. (2001a). Analyses concerning passive and active synthesis: Lectures
on transcendental logic (A. J. Steinbock, Trans.). Springer.
Husserl, E. (2001b). Logical investigations (Vol. 2; J. N. Findlay, Trans.).
Routledge.
Husserl, E. (2002). Philosophy as rigorous science (M. Brainard, Trans.). New
Yearbook for Phenomenology and Phenomenological Philosophy, 2, 249–295.
Husserl, E. (2005). Phantasy, image consciousness, and memory (1898–1925)
(Vol. 11; J. B. Brough, Trans.). Springer Science & Business Media.
Irvine, E. (2013). Consciousness as a scientific concept: A philosophy of science
perspective. Springer.
Kałwak, W., Reuter, M., Łukowska, M., Majchrowicz, B., & Wierzchoń,
M. (2018). Guidelines for quantitative and qualitative studies of sen-
sory substitution experience. Adaptive Behavior, 26(3), 111–127. doi.
org/10.1177/1059712318771690
Kendler, K. S., Zachar P., & Craver, C. (2011). What kinds of things are psychiat-
ric disorders? Psychological Medicine, 41, 1143–1150.
Kockelmans, J. J. (1987). Husserl’s original view on phenomenological psychol-
ogy. In J. J. Kockelmans (Ed.), Phenomenological psychology: The Dutch
school (pp. 3–29). Martinus Nijhoff Publishers.
Kortooms, T. (2002). Phenomenology of time. Springer.
Krueger, J., & Overgaard, S. (2012). Seeing subjectivity: Defending a perceptual
account of other minds. Consciousness and Subjectivity, 47, 297–319.
Landgrebe, L. (1981). The phenomenology of Edmund Husserl. Cornell
University Press.
Levine, J. (1983). Materialism and qualia: The explanatory gap. Pacific
Philosophical Quarterly, 64(4), 354–361.
Levine, J. (2001). Purple haze: The puzzle of consciousness. Oxford University
Press.
Lotz, C. (2007). From affectivity to subjectivity: Husserl’s phenomenology revis-
ited. Palgrave Macmillan.
Merleau-Ponty, M. (2005). Phenomenology of perception (C. Smith, Trans.).
Routledge.
Moran, D. (2005). Edmund Husserl: Founder of phenomenology. Polity Press.
naturalized. Royal Institute of Philosophy Supplement, 72, 89–115.
Moskalewicz, M. (2016). Lived time disturbances of drug addiction therapy
newcomers: A qualitative, field phenomenology case study at Monar-Markot
Center in Poland. International Journal of Mental Health and Addiction, 14(6),
1023–1038.
Moskalewicz, M., Schwartz, M. A., & Gozé, T. (2018). Phenomenology of intuitive
judgment: Praecox-feeling in the diagnosis of schizophrenia. Avant, IX(2), 1–12.
Nagel, T. (1974). What is it like to be a bat? The Philosophical Review, 83(4),
435–450.
Oizumi, M., Albantakis, L., & Tononi, G. (2014). From the phenomenology to
the mechanisms of consciousness: Integrated information theory 3.0. PLOS
Computational Biology, 10(5), 1–25. doi:10.1371/journal.pcbi.1003588
Parnas, J., Møller, P., Kircher, T., Thalbitzer, J., Jansson, L., Handest, P., & Zahavi,
D. (2005). EASE: Examination of anomalous self-experience. Psychopathology,
38(5), 236–258.
Petitmengin, C., Remillieux, A., & Valenzuela-Moguillansky, C. (2019).
Discovering the structures of lived experience: Towards a micro-phenomeno-
logical analysis method. Phenomenology and the Cognitive Sciences, 18(4),
691–730.
Piccinini, G. (2009). First-Person data, publicity & self-measurement.
Philosophers’ Imprint, 9(9), 1–16.
Płotka, W. (2021). A commentary on Pokropski’s functionalist reading of Husserlian
phenomenology. Frontiers in Psychology, 12. doi:10.3389/fpsyg.2021.614842
Pokropski, M. (2015). Affectivity and time: Towards a phenomenology of
embodied time-consciousness. Studies in Logic, Grammar and Rhetoric, 41(1),
161–172.
Pokropski, M. (2018). Commentary: From the phenomenology to the mechanisms
of consciousness: Integrated information theory 3.0. Frontiers in Psychology,
9. doi:10.3389/fpsyg.2018.00101
Pokropski, M. (2019). Phenomenology and mechanisms of conscious-
ness: Considering the theoretical integration of phenomenology with
the mechanistic framework. Theory & Psychology, 29(5), 601–619. doi.
org/10.1177/0959354319868769.
Ratcliffe, M. (2014). Experiences of depression: A study in phenomenology.
Roy, J-M., Petitot, J., Pachoud, B., & Varela, F. J. (1999). Beyond the gap: An intro-
duction to naturalizing phenomenology. In J. Petitot, F. J. Varela, B. Pachoud,
& J.-M. Roy (Eds.), Naturalizing phenomenology: Issues in contemporary phe-
nomenology and cognitive science (pp. 1–80). Stanford University Press.
Rupert, R. D. (2009). Cognitive systems and the extended mind. Oxford
University Press.
Sass, L. A. (2014). Explanation and description in phenomenological psychopa-
thology. Journal of Psychopathology, 20, 366–376.
Schacter, D. L. (1996). Searching for memory: The brain, the mind, and the past.
Basic Books.
Schwitzgebel, E. (2019, Winter). Introspection. In E. N. Zalta (Ed.), The Stanford
encyclopedia of philosophy. Metaphysics Research Lab. https://plato.stanford.
edu/entries/introspection/
Smith, J. (2010). Seeing other people. Philosophy and Phenomenological
Research, 81(3), 731–748.
Taipale, J. (2014). Phenomenology and embodiment: Husserl and the constitu-
tion of subjectivity. Northwestern University Press.
Thinès, G. (1977). Phenomenology and the science of behaviour. George Allen
& Unwin.
Tononi, G. (2004). An information integration theory of consciousness. BMC
Neuroscience, 5, Article 42. doi:10.1186/1471-2202-5-42
Welton, D. (1982). Husserl’s genetic phenomenology of perception. Research in
Phenomenology, 12(1), 59–83.
Vincini, S. (2020). The epistemological contribution of the transcendental reduc-
tion. Husserl Studies, 37, 39–66. doi:10.1007/s10743-020-09265-8
Yoshimi, J. (2007). Mathematizing phenomenology. Phenomenology and the
Cognitive Sciences, 6(3), 271–291.
Zahavi, D. (2001). Husserl and transcendental intersubjectivity. Ohio University
Press.
Zahavi, D. (2003). Husserl’s phenomenology. Stanford University Press.
Zahavi, D. (2013). Naturalized phenomenology: A desideratum or a category
mistake?. Royal Institute of Philosophy Supplements, 72, 23–42.
Zahavi, D., & Kriegel, U. (2015). For-me-ness: What it is and what it is not. In D.
Dahlstrom, A. Elpidorou, & W. Hopp (Eds.), Philosophy of mind and phenom-
enology (pp. 36–53). Routledge.
2 Naturalizing Phenomenology
Reconsidered
= integrating with cognitive science
2.1 Introduction
In the broadest sense, naturalism acknowledges that we ought to explain
phenomena exclusively in terms of the natural sciences. But philosophy
also investigates phenomena that are within the scope of these sciences,
such as the mind. Thus, a question arises about the relation between
philosophical investigations and those conducted by the natural sciences.
Roughly speaking, we can approach this issue from two perspectives (see
Godfrey-Smith, 2003, pp. 149–153). The first belongs to the radical posi-
that’s
tion of reductive naturalism (e.g., Quine, 1969), which holds that philo-
stupid
sophical issues, specifically epistemological ones, ought to be given over why
to empirical psychology. Accordingly, epistemology ought to be replaced
with psychology, which will ultimately deliver a fully naturalistic account
of the origin of knowledge and belief formation. In this version of natu-
ralism, there is no such thing as a genuine philosophical question. Every
question is to be answered by the natural sciences. A more recent ver-
sion of naturalistic reductionism with respect to the mind and cognition
is motivated by rapid developments in neuroscience (e.g., Bickle, 2003;
Churchland, 1986). Bickle’s take on the position, which he calls “ruthless
reductionism,” is essentially fundamentalistic, as he aims to explain men-
tal phenomena purely in neurobiological terms and considers neurobiol-
ogy a fundamental level of explanation.
The second belongs to a weaker version of naturalism, according to
which there are specifically philosophical issues, theories, and methods,
and thus holds that philosophy cannot be replaced by the natural sci-
ences—and even that the products of the natural sciences can be used as
a resource in philosophical considerations. As such, there is a one-way
relation between philosophy and the natural sciences: scientific research
can help us answer some of our philosophical questions, for example,
questions concerning the mind, but that relation does not work in the
opposite direction—scientific endeavors do not benefit from philosophy,
with the possible exception of the philosophy of science. That being said,
naturalism may be understood as a claim about a continuity between
DOI: 10.4324/9781003035367-4
Naturalizing Phenomenology Reconsidered 43
philosophy and science. In which case, the relation is mutual and nonre-
ductive; philosophy and science coexist, interact, and can cooperate with
each other. There is no consensus, however, as to how this continuity
ought to be understood, that is, what the logical, conceptual, and meth-
odological grounds are for this promising cooperation. Even so, there are
proposals for naturalizing phenomenology that occupy this position (e.g.,
Varela, 1996).
My approach to naturalism and the naturalization of phenomenology1
is close to the second nonreductive approach but differs on two important
points. First, I am skeptical of there being specifically philosophical issues
reserved exclusively for philosophers, with the possible exception of meta-
philosophical issues. I think that there are many phenomena, including
cognitive and conscious phenomena, which may be approached from vari-
ous perspectives and studied with various methods. But I do believe that
next to the perspectives of the special sciences, there is a specific philo-
sophical point of view or, to put it differently, philosophical understanding
of the phenomenon in question. Although addressed in detail in Chapter
3, this picture of philosophical understanding can be briefly character-
ized here in two points: (1) it investigates the constitutive conditions of
a studied phenomenon, rather than its enabling conditions (see Wheeler,
2005, 2013), and (2) it offers a meta-perspective on scientific disciplines
or, as Wilfrid Sellars puts it, an “eye on the whole,” that is, “knowing
one’s way around with respect to the subject-matters of all the special
disciplines” (Sellars, 2007, p. 370). The second point communicates that
the most interesting aspect of the relation between scientific endeavors
and philosophy is not how the natural sciences can help in philosophical
investigations, although I am sure they can, but how philosophical investi-
gations can contribute to scientific, naturalistic explanations. With respect
to the naturalization of phenomenology, that would be recognizing a way
in which a phenomenological understanding of mental phenomena could
contribute to naturalistic explanations of them.
Before developing my position further in this chapter, I review natu-
ralistic approaches to phenomenology and evaluate the idea of continu-
ity between phenomenology and natural science expressed in terms of
constraints. First, I introduce the very idea of the naturalization of phe-
nomenology and, drawing on Jean-Michel Roy, Jean Petitot, Bernard
Pachoud, and Francisco J. Varela, the editors of the seminal volume titled
Naturalizing Phenomenology (Petitot et al., 1999), mathematization as a
method of naturalization. Next, I consider several proposals for natural-
izing phenomenology, including front-loaded phenomenology, the formal-
ization of phenomenological descriptions, semantic bridging using applied
ontology, and neuro- and micro-phenomenology. The main focus of my
approach to the debate is the notion of constraints, which characterizes
the relation between phenomenology and cognitive sciences. However,
I argue against the types of constraints that have been proposed in the
debate; in particular, I argue that these positions offer either conceptual
constraints, which are too weak, or stronger isomorphic or homeomor-
phic constraints, which are, however, implausible. This critical evaluation
serves as the ground for developing, in subsequent chapters, an alternative
approach to naturalization based on the idea of mechanistic integration
of cognitive science and proposing a different type of constraint that phe-
nomenology could deliver to research fields in cognitive science.
2.2 Three Views on the Naturalization of Phenomenology

For Husserl, the naturalization of phenomenology was equivalent to
a reductive approach along the lines of the naturalistic attitude in sci-
ence. As we saw in the first chapter, he was highly critical of attempts to
explain consciousness in a purely naturalistic manner and argued that
naturalism is a position flawed with naivety, reliant on scientific realism
according to which the world is simply “out there” and the objective of
science is to accurately describe it. However, as I argued, Husserl was
not simply anti-naturalistic, and he was open to cooperation between
phenomenology and scientific disciplines, especially between phenom-
enology and empirical psychology, which the new discipline of phenom-
enological psychology was thought to straddle. Typically, it is Husserlian
phenomenology, thought of as the most developed and methodologically
aware philosophical account of experience, that is considered as a candi-
date for naturalization, although there have also been attempts to natu-
ralize the phenomenology of Martin Heidegger (e.g., Wheeler, 2005) and
Maurice Merleau-Ponty (e.g., Gallagher, 2005). These phenomenologies
greatly contributed to the field of embodied and extended cognition in
particular. Heideggerian analyses of Dasein were especially important,
including the notion of ready-to-hand, which were successfully integrated
into the understanding of an embodied agent in the world and utilized in
nonrepresentational accounts of embodied coping (e.g., Chemero, 2009).
In the contemporary debate, the naturalization of phenomenology is
typically understood nonreductively, much like the earlier-mentioned sec-
ond view of naturalism; that is, phenomenology and science cooperate,
and there is some sort of continuity between them (for an overview of
the naturalizing philosophy debate, see Gallagher & Schmicking, 2010;
Petitot et al., 1999). In Naturalizing Phenomenology, the editors clarify
in the introduction that “by ‘naturalized’ [they] mean integrated into an
explanatory framework where every acceptable property is made con-
tinuous with the properties admitted by the natural sciences” (Roy et al.,
1999, pp. 1–2). The main issue in the discussion is, therefore, the nature
of that continuity and the mutually imposed constraints.
In the naturalizing phenomenology debate, three general positions can
be distinguished with respect to the possibility of such an enterprise. The
first one holds that naturalizing phenomenology is a dead-end, and there
are several arguments to that end. Dermot Moran (2013), for instance,
stresses Husserl’s critique of naturalism and argues that phenomenology
is primarily a transcendental philosophy investigating the very founda-
tions of reality. As Moran (2013) writes,
phenomenology cannot be naturalized because it tells the story of the

genesis and structure of the reality that we experience but in so doing
reveals subjective stances and attitudes which themselves can never
be wholly brought into view, cannot be objectified.
(p. 90)
This, so to speak, orthodox approach to phenomenology sees in it a funda-

mental science grounding and therefore preceding all other scientific disciplines.
Zahavi (2013), for whom Husserl was a “staunch anti-naturalist” (p. 30) and
for whom phenomenology is primarily a theoretical philosophical and not
empirical enterprise, argues along similar lines. The biggest advantage of the
phenomenological approach is the same as the critical limitation of empirical
sciences; namely, phenomenology can provide a transcendental clarification of
consciousness and explain its role in the constitution of reality.
It is a fact that Husserl was highly critical of naturalistic tendencies
in philosophy. It is also a fact that for Husserl phenomenology was a
transcendental philosophy and that he pursued foundationalism; that is,
he believed that philosophy is independent of science and that it ought
to serve as a foundation for science. At first glance, naturalization is thus
against Husserl’s own position. However, as I argued in the preceding
chapter, Husserl was mainly opposed to naturalistic tendencies in the
philosophy of his day, which aimed to reduce epistemological questions
to biological or psychological ones. Similarly, Husserl strongly criticized
psychologism, that is, the attempt to explain logical laws in psychological
terms, but he was not opposed to psychology as such, which he conceived
of as an important discipline, although lacking proper method. Husserl
was not an anti-naturalist in the sense of denouncing the naturalistic atti-
tude of natural science as inappropriate. His claim was rather that we
should critically rethink the naturalistic attitude to show its naivety, to
disclose its origin, and to indicate its relation to the phenomenological
attitude. Furthermore, his recognition of experimental psychology and
the idea of phenomenological psychology show that Husserl considered
his project something that could be integrated with empirical research.
It seems clear that the integration of phenomenology with the natural-
istic framework cannot concern transcendental phenomenology, which
is grounded in pure intuition and cannot be falsified by empirical sci-
ences. Thus, the second approach to naturalization seeks a weaker notion
of phenomenology that is amenable to such integration. In Chapter 1, I
argued for a concept of phenomenology which downplays its transcen-
dental character and takes it primarily as a method of description and
analysis of structures of experience. This basic concept of phenomenol-
ogy relies on Husserl’s (1977) idea of phenomenological psychology,
according to which the eidetic reduction allows one to study structures
of consciousness but does not bracket the natural attitude (Kockelmans,
1987). In a similar manner, Jack Reynolds (2017) defends the concept of
minimal phenomenology, which
does not preserve a self-sufficient domain that is distinct from empiri-

cal science, either methodologically or substantively. This minimal
phenomenology is shown to be compatible with liberal naturalism
in the ontological register (but incompatible with strong forms of
scientific naturalism that aim for reductive explanations across the
board) and it is also compatible with weak forms of methodologi-
cal naturalism, understood as advocating ‘results continuity’ with the
relevant empirical sciences.
(p. 20)
Although minimal phenomenology gives up transcendentalism, it is, as

Reynolds argues, quasi-transcendental because it emphasizes the irreduc-
ibility of the first-person perspective and the necessity of its phenomeno-
logical study.
Zahavi (2010) does not completely rule out the possibility of natural-
ization but proposes a more moderate approach. According to his view,
“to naturalize phenomenology might simply be a question of letting phe-
nomenology engage in fruitful exchange and collaboration with empiri-
cal science” (p. 8). The naturalization of phenomenology is important, as
phenomenology investigates the same phenomena that empirical science
tries to explain. There is growing consensus that a scientific theory of
consciousness has to account for subjectivity and intentionality and that
phenomenology can support science in this pursuit. Phenomenology gives
justice to the true complexity of consciousness and proposes accounts of
subjectivity and intentionality, as well as temporality, embodiment, and
intersubjectivity. All these aspects of consciousness are critical to explana-
tions and can be informative for empirical studies of conscious phenom-
ena. That influence, however, is not one-directional; it is not just the case
that phenomenology can guide empirical science. According to Zahavi
(2013), “empirical science can present phenomenology with concrete
findings that it cannot simply ignore, but must be able to accommodate;
evidence that might force it to refine or revise its own analyses” (p. 35).
This two-way relation between phenomenology and empirical research is
sometimes called the relation of “reciprocal constraints” (Varela, 1996)
or “mutual enlightenment” (Gallagher, 1997).
The third approach to naturalizing phenomenology puts emphasis on
the very notion of nature and calls for its phenomenological redefini-
tion, which will ultimately lead to a transformation of natural science
(e.g., Gallagher, 2018; Thompson, 2007; Zahavi, 2013). The point of
departure is a Husserlian critique of natural science, which erroneously
adopts the concept of nature as “ready-made” without including the
constitutive role of subjectivity. For Husserl, thinking about nature as
something objective, something that is simply “out there” and waiting
to be described and measured by scientists, is simply naive. Proponents
of phenomenologizing nature argue that this novel concept of nature
depends on bridging physics and biology, and biology to subjectivity. To
achieve this, some adopt the idea of self-organizing biological systems
(e.g., Thompson, 2007) and argue that this redefinition of nature relies
on an enactive and ecological understanding of the environment in which
an embodied cognitive agent is situated. Enactivism rejects neurologi-
cal reductionism and brain-centrism and describes subjects as embodied,
embedded, and situated. Accordingly, a subject and an environment are
in a mutual relation of co-constitution, which can be grasped in phenom-
enological and enactive terms. As Shaun Gallagher (2018) puts it,
the terms “intentional arc,” “embodiment,” and “body schema,” like

the terms “affordance” and “situation,” are used in a way that dis-
rupts the conception of nature as something that science takes for
granted, and tries to get at a different conception of nature—a con-
ception of nature that allows for irreducible structures.
(p. 133)
To put it differently, the target reconceptualization of nature consists of

incorporating subjectivity, that is, a cognizer, an observer, and a conscious
subject, by applying subjectivity-related terms to explain natural pro-
cesses relevant to the subject. Importantly, the relations between nature
and subjectivity are not simple and one-way; thus, for instance, there can-
not be a causal relation between subjectivity and an underlying material
substrate. On the contrary, these relations are dynamic and reciprocal,
which means that nature and subjectivity co-constitute each other.
The problem with this proposal is that it operates with a pared-down
and anachronic idea of natural science. When Gallagher refers to the clas-
sic scientific view of nature, he thinks of a deterministic world governed
by universal laws of nature. That view is associated with the idea of inter-
theoretic reduction (Oppenheim & Putnam, 1958). Accordingly, higher
level scientific theories, such as sociological and psychological ones, are
reducible to lower level theories, like biology and neuroscience, and the
latter, in turn, are reducible to the fundamental level of physics. As we will
see in the next chapter, the idea of reductive unification is outdated, and
it seems that there are few scientists and philosophers who believe that
we will ever arrive at one fundamental natural science. And the related
deductive-nomological model of explanation has also lost its dominant
position in science (see Salmon, 1989). The contemporary landscape of
science is much more pluralistic than it was in the 1950s. Biology and
neuroscience are good examples of the transition in thinking about sci-
entific explanation. In psychology, the concept of a law of nature has no
application (e.g., Cummins, 2000). Nor do we find strict laws of nature
governing biology (e.g., Godfrey-Smith, 2014). Organisms do not behave
as mere physical objects or determined automata but may be seen as
complex mechanical systems with degrees of autonomy. Biology rejects
the nomological model of explanation in favor of the mechanistic model.
One could argue that such a model of explanation is committed to some
sort of brain-centrism in neuroscience, that is, an attempt to explain all
cognitive phenomena in neural- and brain-related terms. On the contrary,
an important aspect of mechanistic explanations is that they are mul-
tilevel; that is, they span several levels of organization of an organism
or cognitive agent. The working of a mechanism is also dependent on
the environment in which it is situated; thus, such explanations require
including the environmental context (Bechtel, 2008). The mechanistic
model of multilevel explanations has recently been adopted in neurosci-
ence, but it explicitly rejects explanatory fundamentalism, that is, attempts
to explain cognitive phenomena only on one neural level (Craver, 2007).
It seems therefore that the brain-centrism objection against naturalistic
cognitive science is invalid.
Commentators of Gallagher’s (2018) paper “Rethinking Nature:
Phenomenology and a Nonreductionist Cognitive Science” also notice
that we do not have to redefine nature to address such aspects of sub-
jectivity as affordances, situatedness, or dynamic relations with environ-
ments or other agents. Predictive coding, for example, is a theoretical
framework that has the resources to model affordances in terms of
embodied agents’ expectations and action possibilities (e.g., Hohwy,
2018; Hipólito, 2018). And dynamic and nonlinear dependencies can
be modeled using dynamical systems theory, and dynamical models of
behavior can be supplemented with mechanistic details (e.g., Kaplan &
Bechtel, 2011). Here the example is biological oscillators, that is, bio-
logical mechanisms generating cyclical behavior such as the mechanisms
of circadian rhythms (Bechtel & Abrahamsen, 2010). To conclude, sci-
ence no longer holds the partes extra partes conception of nature. Nor,
however, does it apprehend subjectivity as phenomenologists see it. That
being said, it seems that the natural sciences have the resources to study
selected aspects of subjectivity, and thus, the results of such studies are
relevant to phenomenology.
To sum up, the first and third views on naturalizing phenomenology seem
too radical. The first results in isolating phenomenology from science and
condemns both science and phenomenology to ignorance. The natural sci-
ences, whether we like it or not, provide us with the dominant view of the
world, including a scientific understanding of human beings and other cog-
nizing, and presumably conscious, organisms. Phenomenology cannot be
blind to scientific discoveries and explanations of phenomena that it is itself
interested in. It is phenomenology’s duty to engage in dialog with natural sci-
ence. Husserl (1989) often emphasizes the responsibility philosophers have,
as the “functionaries for mankind” (p. 17), to look after scientific reason, to
keep it from getting lost. The third view, calling for a redefinition of nature,
seems no better, although it does prompt a rethinking of the foundations
of the scientific view of the world. It gets bogged down, however, in meta-
physical considerations and fails to see that contemporary natural science,
especially cognitive science, is pluralistic and addresses issues relevant to the
embodied and situated nature of cognition. The best naturalization strat-
egy seems therefore the middle road—focusing on how phenomenology
and natural science can cooperate and exchange ideas in order to progress
our understanding of consciousness and cognitive capacities. This approach
places methodology and philosophy of science before metaphysics; it focuses
on strategies of theoretical integration and the nature of mutual constraints.
2.3 Mathematization
The naturalization of phenomenology is closely related to the idea of
mathematizing phenomenological descriptions, which was one of the
main issues in Naturalizing Phenomenology. As the editors write,
it is our general contention indeed that phenomenological descrip-

tions of any kind can only be naturalized, in the sense of being inte-
grated into the general framework of the natural sciences, if they can
be mathematized. We see mathematization as a key instrument for
naturalization, being in fact consonant on this point with Husserl
himself although drawing opposite conclusions from it.
(Roy et al., 1999, p. 42)
The general idea relies on translating phenomenological descriptions

into mathematical formalisms. The advantage of it is that mathema-
tized descriptions are accessible to other researchers and, in principle,
could be compared with other descriptions of the same experience.
Mathematization could also ensure continuity between phenomenology
and natural science understood as a continuity of representations trans-
lated to the universal language of mathematics.
As I argued in Chapter 1, Husserl’s attitude toward the application
of mathematics in phenomenology was problematic. On the one hand,
he blamed mathematized natural science for contributing to the crisis
of reason, and he emphasized differences between the mathematical and
phenomenological approaches. As he writes in the first book of Ideas,
let us make clear to ourselves the most universal peculiarities of

mathematical disciplines as contrasted with those of an eidetic theory
of mental processes and let us therefore make clear what those aims
and methods really are which, as we have suggested, are essentially
inappropriate to the sphere of mental processes.
(Husserl, 1982, p. 161)
To put it differently, according to Husserl, phenomenology is not a

formal discipline, like mathematics, but material, that is, it deals with
experiential objects and acts of consciousness; thus, application of math-
ematical methods to consciousness is inappropriate. On the other hand,
Husserl recognizes that mathematical methods are responsible for the
success of modern science. He also sometimes thinks of applying phe-
nomenology in studies of consciousness in analogy to the application
of mathematics in the natural sciences (Husserl, 1980, p. 38), and fre-
quently uses quasi-mathematical notations in his writings. For example,
in Logical Investigations (Husserl, 2001b), he uses symbols, such as
“P(o),” “P(P(o)),” and the like, to analyze the hierarchical structure of
mental acts of perception, recollection, and so on. In lectures and manu-
scripts on time-consciousness, Husserl (1991) visualizes the flow of con-
sciousness in the form of quasi-mathematical diagrams. In the lectures
titled Thing and Space (Husserl, 1997), he gives a quasi-dynamical analy-
sis of spatial orientation; that is, he analyzes the parameters of visual
field transformations in accordance with bodily movement. In his manu-
scripts on phantasy and image consciousness, he uses symbolic notation
to express eidetic laws of reproduction of perceptual acts (Husserl, 2005,
p. 373). It seems therefore that Husserl’s opinion on the relation between
mathematics and phenomenology is incoherent. In the opinions of Roy,
Petitot, Pachoud, and Varela, Husserl’s distanced position to mathemat-
ics was a consequence of the limitations of the mathematics of his day.
In particular, they point to the recent development of dynamical systems
theory and its applications in studying temporal aspects of consciousness.
Dynamical systems theory (DST) is a mathematical theory about the
behavior of dynamical systems, that is, the evolution of such systems’
behavior in time (see the next chapter for an extended discussion of
dynamical explanations). DST’s formal apparatus typically consists of dif-
ferential equations that include variables representing key features of the
target system and parameters, which stand for external conditions relevant
to the behavior. A selected set of equations describes the target system’s
behavior, which is typically represented by a trajectory in an n-dimensional
state space (n depends on the number of variables). Today, DST is consid-
ered an important method of analysis of cognitive systems’ behavior (e.g.,
Beer, 2000; Kelso, 1995; van Gelder & Port, 1995), but in Husserl’s times,
application of DST was limited to physics. Interestingly, Husserl seems to
have anticipated that dynamical analysis could be useful for studying cog-
nition. According to Jeffrey Yoshimi (2007), Husserl shares fundamental
assumptions with DST. First, DST accounts for the possible states of a
system’s behavior represented in the state space. Similarly, Husserl explores
possibilities of experiential states or plausible modifications of the struc-
ture of the experiential manifold. Second, Husserl also uses variables to
characterize this manifold of experience. In the Thing and Space lectures
(Husserl, 1997, pp. 168–170), he uses variables to describe changes in the
experience of spatial orientation with respect to the movement of different
body parts. The complete list of variables constitutes a kinesthetic system
of the visual sphere. Third, Husserl studies the “behavior of the system”
by modifying variables, for example, making them constant or arbitrarily
varying them, which is a similar method of analysis to the one applied in
contemporary DST. Fourth, the DST state space has geometrical, specifi-
cally topological properties, and Husserl also seems to apply such proper-
ties to the experiential manifold when he speaks about the manifold of
visual experience in Thing and Space. Fifth, the evolution of a dynami-
cal system’s behavior in time is ordered by a set of rules, and, similarly,
Husserl (e.g., 1997, 2001a, 2001b) talks about changes in experience as
characterized by a lawlike regularity, although we have to remember that
he explicitly distinguished phenomenological laws from the formal laws
of mathematics and the laws of nature.
Tim van Gelder (1999), and Varela (1999) likewise, argues for a
dynamical approach to cognition and conceives DST as a mathemati-
cal platform able to integrate phenomenology with cognitive science.
According to van Gelder (1999), “phenomenology and cognitive science
should be regarded not only as compatible, but as mutually constraining
and enriching approaches to the study of mind” (p. 246). Van Gelder
discusses a dynamic model of auditory pattern recognition, that is, the
capacity to recognize temporal objects such as a melody, and considers
how Husserlian phenomenology of time-consciousness can contribute to
this. Accordingly, dynamical models express properties of the process of
recognition, which can be captured on the phenomenological level. In
contrast to computational models, dynamic models exhibit a simultane-
ous unfolding of a pattern and the evolving state of the system, and thus,
dynamical models fit much better with the nature of our experience
than computational models do. Other things being equal, this gives
us good reason to prefer dynamical models over computational mod-
els; and if we accept this, then we have accepted that phenomenology
can substantially constrain cognitive science.
(van Gelder 1999, p. 258)
But what kind of constraints between the phenomenal and the dynamic
are we talking about? Gallagher (1997) reads van Gelder’s proposal as
one of isomorphic constraints, that is, that there is a one-to-one mapping
between the phenomenal and neural levels. As Gallagher (1997) remarks,
“the reasonableness of this strategy will depend, I think, on whether one
takes phenomenology as a strong constraint or a weak one,” and accord-
ing to him, van Gelder appears “to take phenomenology as a strong con-
straint, in the following manner: for van Gelder and Dreyfus the causal
mechanism to be identified on the sub-personal level needs to be isomor-
phic with details explicated in the phenomenological account” (p. 202).
The problem with such a reading is, however, that van Gelder does not
explicitly use the concept of isomorphic constraints, nor does he apply a
strong conception of similarity between the phenomenological and neural
levels. Van Gelder does not clearly articulate what kinds of constraining
relations he has in mind. His claim is very general—that a phenomenology
of time-consciousness can constrain models of underlying mechanisms, for
example, that they should exhibit the dynamics of a network rather than
the sequential order of computational models. I do not feel convinced by
van Gelder, whose proposal of constraints seems very weak and unclear.
It is important, therefore, to consider in detail what types of constraints
phenomenology can place on explanations in cognitive science.
2.4 A Question of Constraints

In this section, I endeavor to clarify what weak and strong constraints are.
I discuss three types of constraining relations between phenomenology
and cognitive science proposed in the naturalization debate. I discuss the
relatively weak notion of conceptual constraints, as represented by front-
loaded phenomenology, formalization of phenomenological description,
and application of formal ontologies as semantic bridges between phe-
nomenology and other scientific domains, according to which phenom-
enological concepts used in phenomenological descriptions and analyses
constrain empirical research, providing a characterization of first-person
phenomena. From that perspective, it is also plausible that phenomeno-
logical conceptions support research hypotheses, which may then be
tested empirically in a direct or indirect fashion. Next, I elaborate further
on the mathematization of phenomenology, including Varela’s dynamical
approach, in order to discuss stronger conceptions of constraints, namely,
isomorphism, generative passages, and homeomorphism. Stronger con-
straints are thought to actually tell us something about the relation
between the experiential level and the level of the underlying neural sub-
strate. As I argue, all the discussed conceptions of constraints have seri-
ous deficiencies, requiring us to develop another approach.
2.4.1 Conceptual Constraints
2.4.1.1 Front-Loaded Phenomenology
Front-loaded phenomenology is one way of incorporating phenomeno-
logical theory into the research practices of naturalistic cognitive science
in a nonreductive manner. It requires neither training research subjects nor
acknowledging a complete phenomenological theory; it simply proposes the
use of phenomenological concepts and distinctions from phenomenologi-
cal literature in the early stages of research, namely, in experimental design.
“Phenomenology comes into the picture by contributing to the experimental
design, by providing clear phenomenological distinctions, which also inform
part of the analytic framework for interpreting the results” (Gallagher &
Brøsted Sørensen, 2006, p. 126). As an example of such a phenomenologi-
cally informed experiment, proponents of this approach refer to research on
the sense of agency (Chaminade & Decety, 2002) and the sense of ownership
(the so-called alien-hand experiment first done by Nielsen, 1963). Roughly
speaking, front-loaded phenomenology favors experimental practice over
methodological considerations. If phenomenological concepts and distinc-
tions taken from the works of Edmund Husserl, Maurice Merleau-Ponty,
and other phenomenologists seem to contribute to our understanding of
experience, then let us use them in experimental designs and see what the
results are. However, it is not entirely clear what constraining role phenom-
enological concepts play in the whole scientific process.2
One possibility is that phenomenological terminology, next to scientific or
folk psychological concepts, contributes to the body of knowledge underly-
ing an experimental design. This is certainly plausible, but such a formulation
of conceptual constraints does not give phenomenology a strong position. It
seems unclear why we should choose the phenomenological tradition instead
of any other approach to first-person phenomena or any other method of
describing and analyzing subjective experience, including folk psychology.
Another possibility is that phenomenological insights could support
the formulation of general heuristics for how to approach the issue in
question and specific testable hypotheses. The former, again, does not
give priority to phenomenology, because proposing an explanatory heu-
ristic, such as decomposition, can be done on the grounds of any scientific
or folk theory, and it is not clear why the phenomenological approach
should be preferred. A solution to this which I develop in Chapter 4
relies on showing that phenomenological decomposition of experience
provides specific functional constraints on functional and mechanistic
models. Testable hypotheses are more plausible. Gallagher and Brøsted
Sørensen (2006) claim that experiments can “test and verify the phenom-
enological description” (p. 131) and that
it is quite possible that experimenting with phenomenology will lead

to a productive mutual enlightenment, where progress in the cogni-
tive sciences will motivate a more finely detailed phenomenological
description developed under the regime of phenomenological reduc-
tion, and a more detailed phenomenology will contribute to defining
an empirical research program.
(pp. 131–132)
There is, however, one problem with this application of front-loaded phe-
nomenology. In the examples of phenomenologically informed experimen-
tal designs that Gallagher considers, that is, the alien-hand experiments,
phenomenology seems to provide a description of the explanandum phe-
nomenon rather than testable hypotheses. In both the sense-of-agency
experiment and the alien-hand experiment, phenomenological concepts
serve to identify and describe the target phenomena, the explanations of
which are searched for on the neurobiological level. If that is the case,
then phenomenological insights are not testable hypotheses and so must
play a different role in the research process; that is, they contribute to the
description of the target phenomenon.
There is also the further issue of the testability of phenomenological
hypotheses, which is controversial as Husserlian phenomenology is often
considered an a priori transcendental science. As I mentioned earlier, for
some contemporary Husserlian scholars, phenomenological transcenden-
talism is a sufficient argument for rejecting the possibility of the natu-
ralization of phenomenology (e.g., Moran, 2013). In response, one can
argue for a softer version of phenomenology that distances itself from
Husserlian transcendentalism. The idea of phenomenological psychology
shows that that is possible. However, even given such a “weak” version of
phenomenology, arguing for the testability of phenomenological insights
through experiments requires recognition of the nature of phenomeno-
logical constraints and the relation between phenomenological claims
and experimental results. Only then would “mutual enlightenment” be
a plausible option.
2.4.1.2 Formalization of Phenomenological Description

Conceptual constraints can take the form of concepts and semantic dis-
tinctions, but they can also be formalized in order to achieve greater
universality and research applicability. One method of phenomenologi-
cal description formalization was proposed by Eduard Marbach (2010).
This method is primarily designed to represent the structure of experi-
ence using symbolic notation, not its content. The idea of symbolic nota-
tion denoting mental activities appears marginally in Husserl’s Logical
Investigations and in some of his manuscripts on phantasy, image con-
sciousness, and memory (Husserl, 2005). According to Marbach (2010),
formalization is thought to clarify phenomenological description and
make it accessible to a broader community of researchers. As he argues,
formalized phenomenological descriptions provide “conceptually-based
constraints for the empirical work by making explicit in advance what is
only implicit in the pre-reflective natural consciousness of participants”
(Marbach, 2010, p. 79). Researchers do not have to infer inner states
on the basis of participant’s behavior and collected data. Marbach’s
approach makes a description of the explanandum standardized and
formal. He also believes that analyzing the “equations” of lived experi-
ences can reveal their pre-reflective structure, and as such play the role
of a heuristic guiding experimental research. The second advantage of
this approach, according to Marbach (2010), is that replications of the
experimental situation are readily available (p. 80), although it is not
entirely clear how his phenomenological notation supports that.
Marbach notation is complex and captures different kinds of men-
tal activities, intentional objects, modes of representation, and relations
between mental activities. For example,
a. mental activities are designated by triplets such as PER—perception,

REM—remembering, IMA—imaging, and so on;
b. intentional objects by lower case letters: x, y, z; (PER)x—perception
with intentional correlate x;
c. brackets designate relations, for instance “( )” designates intentional
correlation in occurring mental activity; “[ ]” stands for intentional
implication (i.e., an occurring mental act implies a non-actual activ-
ity); and
d. strokes such as “foundation stroke ____” (a mental activity is
grounded in another activity), or “belief stroke |-”, when preceding a
bracket designates that an activity is supported by a belief state.
A simple example of a formalized description looks like this:

(REP p |- [PER]) |- x.
This “equation” is to be read as follows: “actually representing x,
believed to be real, by means of representing (reproducing) a perceiving
of x believed to have occurred in the past” (Marbach, 2010, p. 67). The
whole notation is obviously much richer and tries to capture the diversity
and complexity of mental life. Ultimately this method is thought to pro-
vide a formalized description of a subject’s lived experience.
2.4.1.3 Applied Ontology and Semantic Bridging

Another example of formalized conceptual constraints can be found
in the field of phenomenological psychiatry and psychopathology.
Phenomenology and psychiatry have a long tradition of cooperation,
and today there is growing consensus that the application of phenom-
enological methods in psychiatry has significant value (e.g., Fernandez,
2019; Fuchs, 2010; Stanghellini et al., 2019). Phenomenology gives jus-
tice to the complexity and diversity of first-person experience of symp-
toms of mental maladies; thus, most phenomenological approaches to
psychopathology rely on the application of phenomenological concepts
and methods, including recently developed phenomenological interviews
and questionnaires that capture patient phenomenology (e.g., Høffding
& Martiny, 2016; Parnas et al., 2005). Importantly, phenomenological
descriptions of symptoms, including abnormal experiences, can be funda-
mental for diagnosis and treatment and can be grounds for formulating
hypotheses about abnormal experiences.
Rasmus Larsen and Janna Hastings (2018, 2020) recently proposed
to use the tool of applied formal ontology in order to embed a phenom-
enological conception of emotions and affective life in the process of
psychiatric diagnostics.3 The idea behind it comes from applying formal
ontologies to scientific research, as in the case of, for instance, gene ontol-
ogy (Ashburner et al., 2000). Formal ontology is a method of formal-
ization; it consists of entities and relations. More important, it can also
integrate knowledge across different domains of research, for example,
knowledge about biological processes, on the one hand, and chemical
ones, on the other, as in the case of gene ontology. Applied ontology may
be considered a “semantic bridge” between these domains.
According to Larsen and Hastings, phenomenology holds crucial knowl-
edge concerning the domain of mental functioning that, if standardized, could
be applied to mental health research. As they write, “we aim to initiate a stan-
dardised formalisation of central entities and relationships in patient phenom-
enology, applicable across the sciences and disciplines in mental health research
and practice” (Larsen & Hastings, 2020, p. 202). They propose to modify
Mental Functioning Ontology (Hastings et al., 2012) by adding terms and
entities to it related to phenomenological structures, such as body awareness
and time awareness. To start building the ontology of mental health symp-
toms, we can find experiential structures relevant to mental maladies in phe-
nomenologically informed clinical tools such as, for example, the Examination
of Anomalous Self-Experience scale (EASE) (Parnas et al., 2005) and the mea-
sure of Identity and Eating Disorders. For instance, an instantiation of body
awareness in abnormal experience is the EASE category “body estrangement”
or “psychophysical split.” Moreover, a complete formal ontology of mental
health symptoms could be used in computational science and, in turn, could be
used in psychiatric assessment tools and patient tracking systems.
This proposal goes beyond using phenomenology only to describe men-
tal phenomena (in this case, mental maladies) and aims to guide research,
diagnosis, and treatment. In some sense, this approach corresponds
to front-loaded phenomenology; that is, applied ontology facilitates
front-loading phenomenology by providing standardized semantics. An
advantage of this approach is the broader picture of integration through
“semantic bridging” between different domains of study. It follows that
phenomenology may figure as part of a larger integrative framework of
applied ontology; for example, it can be linked with psychiatric and neu-
robiological knowledge. That said, Larsen and Hastings remark that the
success of applied ontology greatly depends on the practitioners, who, for
example, collect data from interviews. It also depends on phenomenology
scholars and phenomenological psychiatrists, who analyze the data in
order to distinguish relevant entities and relations.
2.4.2 Isomorphism
Isomorphism and the following notions are examples of stronger con-
straints that try to define the relation between the explanandum phe-
nomenon of experience and underlying neural substrate. The idea is not
completely new. Thinking about the relation between psychological phe-
nomena and neural states in terms of isomorphism appeared in Gestalt
psychology already in the late 1920s and was defended by Wolfgang
Köhler, who coined the notion of psychophysical isomorphism. Today,
isomorphic relations between phenomenology and cognitive science
appear in two forms, called analytical isomorphism and phenomenal iso-
morphism, respectively. According to the first, “analytical isomorphism”
(Lutz, 2002), also called “neural-perceptual isomorphism” (Roy et al.,
1999; Thompson et al., 1999), there is some sort of a similarity between
the phenomenal content of a given experience and neurobiological data.
Moreover, that similarity is explanatory. Accordingly,
Φ looks like Ψ ⇒ Φ explains Ψ (Roy et al., 1999, p. 67), where Φ stands
for neurobiological patterns and Ψ for phenomenal structures.
An example of such an isomorphism is found in vision studies, especially
in studies of perceptual completion (Thompson et al., 1999). Consider,
for example, the Craik–O’Brien–Cornsweet effect (see Cornsweet, 1970):
two uniform regions are seen as having a different brightness, whereas, in
fact, they have the same level of brightness; the only difference in lumi-
nance is at the edge separating the two regions. The proposed expla-
nation relies on identifying a specific set of neurons, called the “bridge
locus,” the activation of which is both necessary and sufficient for the
occurrence of the effect. To put it differently, the identified set of neurons
constitute the “immediate substrate” for the explanandum effect. The
isomorphism approach goes one step further and assumes that there is a
one-to-one correspondence between the perceived distribution of bright-
ness and neural activation of the bridge locus. If that is the case, then the
prediction is that we should detect a step difference in the neural activa-
tion, which would correspond to the difference in perceived brightness.
In short, the explanation is that “the brain takes the local edge infor-
mation and uses it to fill in the two adjacent regions so that the region
with the luminance peak (left) becomes brighter than the region with
the luminance trough (right)” (Thompson et al., 1999, p. 167). To sum
up, the proposed explanation consists of two claims: the first identifies
the bridge locus (neural substrate), and the second claim is based on an
isomorphic premise that maps structural features of the phenomenal level
onto the neural level.
It seems, however, that an isomorphic relation between phenomenol-
ogy and cognitive science cannot lead to the naturalization of phenom-
enology. Gallagher (1997) convincingly argues that neural-perceptual
isomorphism is implausible because of essential differences between the
phenomenological and neural levels, such as different temporal scales on
the level of experience and the underlying neural level. The distributed
character of neural processing is also essentially different from the gestalt-
like character of perception at the phenomenological level. These differ-
ences make one-to-one mapping implausible and suggest that the relation
between the phenomenological and the neurobiological has to be weaker
than isomorphic. In general, there is no good reason to assume that an
isomorphic relation holds between the explanandum phenomenon, such
as the behavior of a system, and underlying mechanisms (explanans).
Much as, for example, there is no reason to assume that there is an iso-
morphic relation between the behavior of a car driving smoothly and the
structure and organization of its engine, brakes, and gears, although there
is certainly a causal relation between these levels.
But let us assume for the sake of argument that such an isomorphic
relation could be established. Would it be explanatory? There are good
reasons to think that it would not be. First, there are many isomorphic
objects, including mathematical ones, such as isomorphic sets or geo-
metrical figures, but we would not say that one explains the other. The
isomorphism between such objects only means that their structures
can be mapped in a one-to-one fashion. Even if we considered a more
folk notion of isomorphism used in psychology, that is, isomorphism
as structural similarity, such a relation would still not be explanatory
as such. Similarity alone does not explain anything. For example, con-
sider a human-shaped shadow on a wall and a person standing in front
of that wall. The shadow is isomorphic; that is, it is similar in shape
to the person’s body. But imagine that the shadow stretches out in one
direction. Now, does the isomorphic relation to the shape of the person’s
body explain the shape of the shadow? Certainly, it contributes to the
explanation; it tells us something about why the shadow has the shape
that it does. But to have a full explanation of this kind of phenomenon,
we need to supplement our knowledge of similarities with causal and
nomological knowledge, including information about the position of the
light source and the law of rectilinear propagation of light. Only in this
way can we explain why there is a shadow on the wall and why it has a
stretched, person-like shape. By analogy, to explain a type of mental state,
assumed to be isomorphic with neurobiological processes, we also need
causal explanations of how the properties of the phenomenon we want
to explain are produced.
The second version of isomorphic relation, called “phenomenal iso-
morphism” by some (Lutz, 2002), is a more global approach, which takes
phenomenological descriptions and argues that they can be explained
by relating them to global and distributed neural activations (notice
that no single “bridge locus” is postulated). Here, the division of labor
is clear. Phenomenology delivers descriptions of experiential structures,
and cognitive (neuro)science delivers a scientific explanation in terms of
large-scale neural synchronization. Antoine Lutz discusses a recent ver-
sion of the Mooney face test as an example of this isomorphic approach.
In the experiment, subjects were presented upright or upside-down
Mooney figures (partially open figures the recognition of which requires
closure) and asked to press a button as soon as they perceived a face-
shape in the presented stimuli. Subjects’ electrical brain activity was
recorded using EEG. Local and global synchronizations of neural activ-
ity were observed; in particular, a large-scale pattern of synchrony was
detected when subjects perceived a face in the stimuli (upright condition).
The synchrony pattern is, as the authors of the study hypothesized, a
neural correlate of the perception of face gestalt. The isomorphic relation
is thought to lie between the perceptual awareness of the stimuli and the
underlying mechanisms of large-scale neural synchronization. It is not
clear, however, how one could account for one-to-one mapping between
these levels. Furthermore, the experiment largely reduces the phenom-
enal level to behavioral responses to stimuli. The subject’s role is only to
report, in the form of a motoric response, the occurrence of a perceptual
state; thus, they cannot provide any informative insights into the struc-
ture of lived experience, which, as one could argue, is isomorphic with
neural activations.
According to Lutz, the phenomenal isomorphism approach faces seri-
ous limitations. First, it rules out the possibility that one experiential state
can affect another experiential state, which undermines the phenome-
nological notion of experience, that is, that one experiential state can
(motivationally) affect another; for example, a perceptual state can moti-
vate a belief state. Reducing the subject’s activity to reporting “what”
they perceptually experience makes it impossible to investigate “how”
the experience emerged, that is, the dynamics of the studied experience
and the dynamic relations between different mental states. Importantly,
modifications in the dynamics of experience may be present on the neu-
robiological level, but without a proper first-person methodology, they
cannot be identified and explained. Second, this approach simplifies
the cognitive system to context-free input–output relations; that is, the
subject passively receives the stimuli and reacts with a simple motoric
response. Thus, this approach cannot do justice to the emergence of more
complex experiential states, which are context-dependent. Third, Lutz
also raises the issue, although it is a common assumption in many experi-
mental designs: that the experiential state preceding the stimulation is
neutral and does not interfere with the state under study, which makes it
impossible to investigate the real dynamics and emergence of the expe-
rience. The neurophenomenological method was designed to overcome
these limitations.
To sum up, constraints understood as isomorphic relations are either
implausible or too limited. Processes at the phenomenological level
and those on the neurobiological level differ so greatly that there is no
good reason to assume that there is an isomorphic (one-to-one) relation
between them. Furthermore, even if such a relation could be established,
it would not be explanatory, although it might help in formulating genu-
ine scientific explanations. Lutz also indicates three limitations of experi-
mental protocol based on phenomenal isomorphism. Overcoming these
limitations was one of the motivations behind developing neurophenom-
enology and the “generative passages” account of constraints.
2.4.3 Neurophenomenology and Generative Passages

Another proposal for stronger constraints comes from neurophenom-
enology, which was first introduced as a methodological remedy to the
hard problem of consciousness (Varela, 1996).4 The idea behind the proj-
ect was to develop methodological tools that could shed new light on
the “explanatory gap”, that is, the relation between first-person lived
experience and underlying neural processes (Chalmers, 1996). Varela,
taking inspiration from Husserlian phenomenology (e.g., phenomeno-
logical reduction), acknowledged that essential features of experience
are accessible to the subject and may be studied. Importantly, neurophe-
nomenology is interested in generic structures of experience and not in
descriptions of individual experiences and their qualitative character.
Neurophenomenology’s ambition of solving the hard problem is
impressive, but looking at it more realistically, neurophenomenology
aims to correlate disciplined first-person descriptions of experience with
neuroimaging data (EEG). In the paper in which Varela introduces the
idea of neurophenomenology, he formulates the working hypothesis of
neurophenomenology, which is that “phenomenological accounts of the
structure of experience and their counterparts in cognitive science relate
to each other through reciprocal constraints” (Varela, 1996, p. 343). As
he further argues, “disciplined first-person accounts should be an inte-
gral element of the validation of a neurobiological proposal, and not
merely coincidental or heuristic information” (Varela, 1996, p. 344).
Neurophenomenology is, therefore, a nonreductionist approach to the
naturalization of phenomenology that sees phenomenology as having an
important role to play as part of empirical cognitive science—namely,
delivering reliable first-person descriptions with identified generic struc-
tures correlated with neural dynamics and validation of empirical theo-
ries of consciousness.
Proponents of neurophenomenology characterize its methodology in
three points:
1. (NPh1) First-person data from the careful examination of experience

with specific first-person methods.
2. (NPh2) Formal models and analytical tools from dynamical systems
theory, grounded on an embodied-enactive approach to cognition.
3. (NPh3) Neurophysiological data from measurements of large-scale,
integrative processes in the brain (Lutz & Thompson, 2003, p. 34).
Neurophenomenology recognizes the problem of introspective reports

being an unreliable source of information about one’s inner life. One
option is to elaborate more sophisticated methods of carrying out phe-
nomenological interviews or to train subjects to be more aware of and
cautious about their experiences and language of description. In neu-
rophenomenology, descriptions are produced by experimental subjects
trained in phenomenological method, which is a simplified version of
Husserlian phenomenological reduction (NPh1). The vocabulary used to
describe studied experiences is elaborated together with the experimenter,
who guides the subject. In short, subjects are trained (1) to suspend their
commonsensical and theoretical beliefs about their experiences and men-
tal states in general and (2) to imaginatively vary their own experience
in order to apprehend and describe invariants, which are used to identify
the structural properties of experiences. The objective of the phenom-
enological method used in neurophenomenology is precisely to describe
this invariant structure of experiences. The third-person methodological
component (NPh3) adopted in neurophenomenology is electroencepha-
lography (EEG) of subjects’ brain activity as recorded in experimental
trials. However, the key to successfully establishing relations between
first-person descriptions and third-person data is the correct applica-
tion of dynamical systems theory (NPh2) as a mathematical framework
to analyze and describe the dynamics of both relata. The mathematical
framework is the middle ground on which the phenomenological and
neurobiological can meet.
According to Varela (1996, p. 343), this kind of phenomenological
structural description can provide constraints on empirical observations.
But as he argues, “one thing is clear: the specific nature of the mutual
constraints is far from a simple empirical correspondence or a categorical
isomorphism” (Varela, 1999, p. 305). Proponents of neurophenomenol-
ogy characterize the circulation between the phenomenological level and
the neurobiological level as “generative passages” and argue that it over-
comes the limitations of isomorphism (Lutz, 2002; Roy et al., 1999). In
the generative passages approach,
the mutual constraints not only share logical and epistemic account-
ability, but are further required to be operationally generative, that is
to say, to be in a position to generate in a principled manner eidetic
descriptions that can directly link to explicit processes of biological
emergence. For this to happen at least both sides of the wavy line
must be joined at a level of description sufficiently abstract that it
rightly belongs to both sides at the same time. In other words, the
neurobiological and the phenomenological accounts have to achieve
a level of formal and mathematical precision to make this passage
possible.
(Roy et al., 1999, p. 68)
To put it differently, generative passages relies on relating, via the formal

language of dynamical systems theory, phenomenological descriptions of
experiential structures with neurophysiological data (EEG) of distributed
neural activations, both local and global. Lutz (2002) himself admits
that the conception of “generative passages” requires further elabora-
tion (p. 162), and it can be characterized as a form of “three-way iso-
morphism” between the phenomenal, neural, and the dynamic, that is, a
one-to-one mapping between these three levels, or it can be downgraded
to homomorphism, that is, a weaker relation of many-to-one that reduces
the structural features of experience when projected onto the dynamical
and neural levels.
It is not easy to see how the idea of “generative passages” works in
practice, since there are only a few experiments that have been done in
accordance with this framework. An example is the study conducted by
Lutz and colleagues (2002) on the degrees of perceptual readiness, that
is, a subjective feeling of preparation to perceive the emerging stimulus.
The experimental setup corresponds to the Mooney face experiments, in
which subjects were asked to report when they perceived a visual stimu-
lus. However, the objective of the experiment was to allow subjects to
describe and categorize their perceptual experience rather than simply
report the fact that a perceptual state occurred. During the training phase,
the subjects were repeatedly asked to focus their gaze on a background
static dot pattern, fuse two black squares (displayed at the bottom) with
a subtle eye movement, and remain in that eye position. After 7 seconds,
the random dot pattern shifted to an autostereogram designed to induce
a three-dimensional illusory geometric shape that the subjects were able
to perceive due to their positioning. The subjects were then asked to press
a button as soon as they perceived the shape. Importantly, after each trial,
the experimenter asked subjects open questions about their inner expe-
rience, directing their attention towards how the experience unfolded
rather than toward perceptual content. The subjects were repeatedly
exposed to the task until they developed experiential categories captur-
ing different states of attentional preparation categorized as stable readi-
ness, fragmented readiness, spontaneous unreadiness, and self-induced
unreadiness. In the experiment, the subjects used these categories to label
particular experiences in experimental trials, during which EEG signals
were recorded.
In accordance with the neurophenomenological program, Lutz et al. (2002)
were interested in large-scale oscillatory patterns of neural behavior rather
than in signal averaging. They adopted stability of phase difference between
multiple pairs of electrodes as a measure of long-range synchronization of
neural populations. The number of electrodes with above-chance phase
alignment was averaged across time windows (with time bracket duration
corresponding to particular brain wave frequencies) and compared between
baseline (prior to the 7-s fixation) and attentional preparation (just before a
dot pattern change to an autostereogram) periods. Higher (lower) alignment
in the latter period called phase-locking (phase-scattering) was an indicator
of increased (decreased) global neural synchronization. Two major findings
included (1) large-scale patterns of increased synchronization observed in the
frontal brain region in the case of increased attentional preparation and (2)
lower attentional readiness corresponding to a greater degree of phase scat-
tering in the posterior brain regions (see Olivares et al., 2015).
Lutz et al. (2002) claimed that, in their study, “mutual constraints were
tested and instantiated in the implementation of [phenomenal clusters] to
guide the analyses of neurophysiological data” (p. 1590). However, it is
hard to understand in what sense these mutual constraints were established.
Distinguished phenomenal clusters (first-person descriptions) neither led
researchers to modify their analysis pipeline nor directed their attention
toward particular properties of the signal, nor did they restrict analysis
options. They were simply juxtaposed with dynamic neural synchroniza-
tion patterns, as any experiential category could be—either predefined or
developed with another procedure. Undeniably, first-to-third-person data
matching was more reliable in this case, but it does not satisfy the criteria
for “coupling” in the DST sense (for further discussion, see Chapter 5).
2.4.4 Neurophenomenology and Homeomorphism

The key methodological components of neurophenomenology were
taken up and extended by Claire Petitmengin et al. (2007):
Neurophenomenology argues that unfolding this twofold dynamic is

what allows the articulation of neurological activities and subjective
experiences. Therefore, the development of this program presupposes
the design of suitable methods for:
(1) Analyzing the global dynamics of cerebral activity associated

with a given cognitive process.
(2) Collecting rigorous and precise descriptions of the dynamics of
the corresponding subjective experience.
(3) Establishing correlations between these two dynamic structures
and refining their process of reciprocal determination.
(p. 747)
According to Petitmengin, we can learn to investigate our cognitive pro-

cesses involved in experience, which are otherwise usually not experi-
enced in everyday activities. To do so, one has to focus on particular
occurrences of a cognitive process and shift attention from “what” is
experienced to “how” it is experienced. In this way, the very process of
the emergence of experience can be grasped and analyzed. This process
is supported by the interviewer in the so-called elicitation interview,
which was adopted by Petitmengin and used, for example, in a study
with epileptic patients (Petitmengin et al., 2006, 2007). The elicitation
interview, originally based on the explication interview proposed by
Pierre Vermersch (1994), focuses on an in-depth exploration of a par-
ticular experience, but its main area of interest lies in “how” the experi-
ence unfolds. During the procedure, the interviewer guides an iterative
process of detailed description of the evoked past experience of interest,
redirecting the interviewee’s attention from the content of the experience
toward the process itself. To do so, the interviewer asks open, “content-
empty” questions and remains vigilant to detect inaccurate descriptions
contaminated by the subject’s presuppositions. The dynamical approach
is mentioned in explicating the process, as it provides a mathematical
framework capable of doing integrative work between the phenomeno-
logical and neurobiological levels.
Petitmengin et al. (2007) argue for a homeomorphic relation, that is,
topological equivalence, between descriptions of first-person-experience
dynamics and the dynamics of neural processes. As they argue,
[a] correlation between a neuro-dynamic structure and a pheno-

dynamic structure consists of establishing:
(1) A temporal coincidence between the cerebral activity which is

recorded and the subjective experience which is described.
(2) A possible correspondence between the neuro-dynamic structure
(a succession of neuronal configurations) and the pheno-dynamic
structure (a succession of subjective events) that corresponds to
the former in time; this correspondence may take the form of an
homeomorphism.
(p. 755)
And furthermore, they add, “[T]he rough homeomorphism that we have

identified could be made more precise, thanks to finer comparisons of the
neuro-dynamic and pheno-dynamic structures” (Petitmengin et al., 2007,
p. 758). Putting aside what “rough” homeomorphism actually means,
it is doubtful that a neuro- and micro-phenomenological framework in
its current form could provide such topological identification. First, a
homeomorphic relation, as a relation of topological equivalence, is a rela-
tion between two n-dimensional spaces. It is thus implausible that there
is a homeomorphic relation between the simple diachronic structure of
experience (a one-dimensional succession of an experience’s phases) and
the dynamics of brain processes represented n-dimensionally (where n
stands for the number of variables defining the state space representing
neural network dynamics).
Furthermore, establishing a homeomorphic relation between the
dynamics of the phenomenological and neural levels requires elaborat-
ing a terminology capable of grasping dynamical features on both levels.
DST certainly provides such a language along with methods for both
quantitative and qualitative analysis of the target system’s behavior but
is seems that neurophenomenology has not delivered this dynamical
correlation.5 Proponents of homeomorphic constraints do use primary
categories from the DST dictionary, such as stable and unstable states,
attractors and repellents, and so on, to describe a system’s behavior.
However, such dynamic description (see van Gelder & Port, 1995) is too
abstract and metaphorical to serve as a dynamic model and to establish a
formal homeomorphic relation.
The next issue with homeomorphic relations is that two spaces are
homeomorphic if and only if there is a function between them that is
a bijection (one-to-one), that is continuous, and that is inverse. Let us
notice further that the idea of a homeomorphic relation between pheno-
and neuro-dynamic structures depends on the doubtful assumption that
phenomenological descriptions of “experiential landscape” operate on
the same level of granularity as descriptions of the state space of neural
dynamics. There are several reasons to doubt that and the very idea that
a one-to-one relation between these levels is even possible. First, it seems
that there is a great number of ways we could decompose our experience,
and none of the experiential components can be acknowledged as ele-
mentary. The micro-phenomenological structure of experience may con-
sist of several levels of abstraction, and what the final phenomenological
cluster will look like is a matter of pragmatic decision driven by explana-
tory interests. There are no good arguments to think that we will find an
equivalent structure with the same levels of composition on the correlated
neural level. Second, neurophenomenology itself acknowledges that there
are different time scales in which mental phenomena occur: “the scale of
the intrinsic cellular rhythms from 10 to 100 msec (1/10 scale), the scale
of large scale integration on the order of 100–300 msec (1 scale), and the
scale of long term processes over a few seconds (10 scale)” (le van Quyen,
2003, pp. 71–72). Occurrent conscious experiences appear on scale 1 and
scale 10, whereas neural processes are on scale 1/10. It is highly unlikely
that there is a one-to-one correspondence between these levels.
Finally, notice that, paradoxically, establishing a homeomorphic rela-
tion between the phenomenological and neural levels opens the door to
reductionism. Generally speaking, a homeomorphic relation between
space A and space B means that they share topological relations. That
implies that we can infer the topological properties of A from B and vice
versa (homeomorphism is symmetrical). However, if that is the case, if
some of the properties of phenomenological dynamics can be deduced
from neural dynamic patterns, then neurophenomenology has a problem.
Namely, it is reductive in the same way that any account of cognition that
aims to infer properties of the mental from properties of the physical is.
This, however, would contradict the explicit nonreductive commitments
made by proponents of neurophenomenology, including Varela, Lutz,
and Petitmengin.
2.5 Conclusion
Although the naturalization of phenomenology has been the subject of
debate for over three decades, it is still a controversial topic. On the one
hand, some phenomenologists (e.g., Moran, 2013) reject the idea of natu-
ralizing phenomenology as implausible and at odds with Husserl’s project
of phenomenological philosophy, and others (e.g., Thompson, 2007) call
for a phenomenological redefinition of nature. On the other hand, more
pragmatically oriented cognitive scientists and philosophers who merge
different traditions (e.g., Gallagher & Brøsted Sørensen, 2006; Varela,
1996) argue that naturalization is not only possible but also vital both for
empirical studies of consciousness and phenomenology. The latter mid-
dle road to naturalization, endorsed by positions that I discussed in this
chapter, is a nonreductive integration of phenomenology with cognitive
science. The key issue in this project is finding a methodological bridge to
make such naturalization possible, in particular, considering the mutual
constraints between phenomenology and cognitive science.
As I argued, although conceptual constraints introduce phenomenol-
ogy to research practice, they only constrain explanations in a weak
sense. Constraints understood as “generative passages” or isomorphic
and homeomorphic relations between the phenomenal and the neurobio-
logical appear implausible because they assume some sort of similarity or
continuity between these levels (Roy et al., 1999). There is no satisfac-
tory empirical research supporting there being any such relations, and the
assumption backing it appears unjustified on its own.
I do not see constraining relations as necessarily involved in some
sort of similarity or continuity. There are two reasons for this. First, the
idea of continuity between phenomenology and empirical science seems
amenable to reductionism; that is, if the phenomenological level can be
derived from the neurobiological level, then it may be reducible in prin-
ciple. In my view, naturalizing phenomenology is not a reductive project
but an integrative one; that is, it seeks to answer how scientific and philo-
sophical views on mental phenomena and their representations “hang
together.” Second, I generally agree with the view that science is charac-
terized by disunity (Dupré, 1995; Fodor, 1974) and explanatory plural-
ism (Mitchell, 2002) rather than unity (Oppenheim & Putnam, 1958).
According to the former view, science consists of heterogenic structures
and disciplines that are incongruent and thus irreducible to one another.
That does not mean, however, that heterogenic scientific endeavors
cannot inform or constraint each other. As I argue in the next chapter,
drawing on Lindley Darden and Nancy Maull’s (1977) conception of
interfield theories and from Craver’s proposal of mechanistic integration
of cognitive science (2007), different fields of research can constrain one
another and together contribute to mechanistic multilevel explanations.
These issues shed new light on novel ways of naturalizing phenomenol-
ogy, including alternative types of constraints phenomenology can pro-
vide and thus contribute to explanatory mechanistic models. Thus, it is
of great importance to understand what naturalistic explanation is, what
models of naturalistic explanation are applied in cognitive science, how
different fields of research may constrain each other, and, last but not
least, whether phenomenology can provide such constraints. This is a
clear weakness of the naturalization debate—it hardly addresses the vari-
ety of available explanatory models or the conditions of their integration,
nor does it consider whether phenomenology offers a sui generis model
of explanation and how it might relate to other models. I address these
issues in the following chapter.
Notes
1 For an interesting discussion of the relation between phenomenology and
naturalism, see Reynolds (2017).
2 An earlier version of the discussion of front-loaded phenomenology appeared
in “Phenomenology and Mechanisms of Consciousness: Considering the
Theoretical Integration of Phenomenology with the Mechanistic Framework”
(Pokropski, 2019).
3 The application of formal ontologies in science, for example, computer science and
the life sciences, is a growing field of research (see, e.g., Smith & Ceusters, 2010).
An important contribution to this field was Ingarden’s phenomenological ontology
(e.g., Ingarden, 1965a, 1965b; see also Chrudzimski, 2004; Płotka, 2020).
4 For an earlier version of the discussion neurophenomenological method, see
Pokropski (2019).
5 See Chapter 3 for a more detailed discussion of dynamic explanation and Chapter
5 for a critique of the application of DST in neuro- and micro-phenomenology.
References
Ashburner, M., Ball, C. A., Blake, J. A., Botstein, D., Butler, H., Cherry, J. M., &
Sherlock, G. (2000). Gene ontology: Tool for the unification of biology. Nature
Genetics, 25(1), 25–29. doi:10.1038/75556
neuroscience. Routledge.
Bechtel, W., & Abrahamsen, A. (2010). Dynamic mechanistic explanation:
Computational modeling of circadian rhythms as an exemplar for cognitive
science. Studies in History and Philosophy of Science Part A, 41(3), 321–333.
Beer, R. D. (2000). Dynamical approaches to cognitive science. Trends in Cognitive
Sciences, 4, 91–99.
Bickle, J. (2003). Philosophy of neuroscience: A ruthlessly reductive approach.
Kluwer Academic Publishers.
Chaminade, T., & Decety, J. (2002). Leader or follower? Involvement of the infe-
rior parietal lobule in agency. Neuroreport, 13(1528), 1975–1978.
Chemero, A. (2009). Radical embodied cognitive science. MIT Press.
Chrudzimski, A. (2004). Roman Ingarden: Ontology from a phenomenological
point of view. Reports on Philosophy, 22, 121–142.
Churchland, P. S. (1986). Neurophilosophy: Toward a unified science of the
mind-brain. MIT Press.
Cornsweet, T. N. (1970). Visual perception. Academic Press.
Cummins, R. (2000). “How does it work” versus “what are the laws?”: Two
conceptions of psychological explanation. In F. Keil & R. A. Wilson (Eds.),
Explanation and cognition (pp. 117–145). MIT Press.
Darden, L., & Maull, N. (1977). Interfield theories. Philosophy of Science, 44(1),
43–64.
Dupré, J. (1995). The disorder of things: Metaphysical foundations of the dis-
unity of science. Harvard University Press.
Fernandez, A. V. (2019). Phenomenology and dimensional approaches to psychi-
atric research and classification. Philosophy, Psychiatry, & Psychology, 26(1),
65–75.
Fodor, J. (1974). Special sciences (or: The disunity of science as a working hypoth-
esis). Synthese, 28, 97–115.
Fuchs, T. (2010). Phenomenology and psychopathology. In S. Gallagher & D.
Schmicking (Eds.), Handbook of phenomenology and cognitive science (pp.
546–573). Springer.
Gallagher, S. (1997). Mutual enlightenment: Recent phenomenology in cognitive
science. Journal of Consciousness Studies, 4(3), 195–214.
Gallagher, S. (2005). How the body shapes the mind. Clarendon Press.
Gallagher, S. (2018). Rethinking nature: Phenomenology and a non-reductionist
cognitive science. Australasian Philosophical Review, 2(2), 125–137. doi:10.1
080/24740500.2018.1552074
Gallagher, S., & Brøsted Sørensen, J. (2006). Experimenting with phenom-
enology. Consciousness and Cognition, 15(1), 119–134. doi.org/10.1016/j.
concog.2005.03.002
Gallagher, S., & Schmicking, D. (Eds.). (2010). Handbook of phenomenology and
cognitive science. Springer.
Godfrey-Smith, P. (2003). Theory and reality: An introduction to the philosophy
of science. University of Chicago Press.
Godfrey-Smith, P. (2014). Philosophy of biology. Princeton University Press.
Hastings, J., Ceusters, W., Jensen, M., Mulligan, K., & Smith, B. (2012).
Representing mental functioning: Ontologies for mental health and disease. In
Proceedings of the third international conference on biomedical ontology (pp.
1–5). University of Graz.
Hipólito, I. (2018). Perception is not always and everywhere inferential.
Australasian Philosophical Review, 2(2), 184–188.
Hohwy, J. (2018). Phenomenology and cognitive science: Don’t fear the reduc-
tionist bogey-man. Australasian Philosophical Review, 2(2), 138–144.
Høffding, S., & Martiny, K. (2016). Framing a phenomenological interview: What,
why and how. Phenomenology and the Cognitive Sciences, 15(4), 539–564.
Nijhoff.
Husserl, E. (1989). The crisis of European sciences and transcendental phenom-
Husserl, E. (1997). Thing and space: Lectures of 1907 (R. Rojcewicz, Trans.).
Kluwer.
Routledge.
Ingarden, R. (1965a). Der Streit um die Existenz der Welt, Vol. 1: Formalontologie,
Pt. 1: Form und Wesen. Niemeyer.
Ingarden, R. (1965b). Der Streit um die Existenz der Welt, Vol. 2: Formalontologie,
Pt. 2: Welt und Bewußtsein. Niemeyer.
Kaplan, D. M., & Bechtel, W. (2011). Dynamical models: An alternative or comple-
ment to mechanistic explanations? Topics in Cognitive Science, 3(2), 438–444.
Kelso, J. A. S. (1995). Dynamic patterns: The self-organization of brain and
behavior. MIT Press.
Kockelmans, J. J. (1987). HusserI’s original view on phenomenological psychol-
Larsen, R. R., & Hastings, J. (2018). From affective science to psychiatric disor-
der: Ontology as a semantic bridge. Frontiers in Psychiatry, 9. doi.org/10.3389/
fpsyt.2018.00487
Larsen, R. R., & Hastings, J. (2020). Mapping the patient’s experience: An
applied ontological framework for phenomenological psychopathology.
Phenomenology and Mind, 18, 200–219.
le van Quyen, M. (2003). Disentangling the dynamic core: A research program
for a neurodynamics at the large-scale. Biological Research, 36(1), 67–88.
1(2), 133–167.
Lutz, A., Lachaux, J. P., Martinerie, J., & Varela, F. J. (2002). Guiding the study
of brain dynamics by using first-person data: Synchrony patterns correlate
with ongoing conscious states during a simple visual task. Proceedings of the
National Academy of Sciences, 99(3), 1586–1591.
Lutz, A., & Thompson, E. (2003). Neurophenomenology integrating subjective
experience and brain dynamics in the neuroscience of consciousness. Journal of
Consciousness Studies, 10(9–10), 31–52.
Marbach, E. (2010). Towards a formalism for expressing structures of conscious-
ness. In S. Gallagher & D. Schmicking (Eds.), Handbook of phenomenology
and cognitive science (pp. 57–81). Springer.
Mitchell, S. D. (2002). Integrative pluralism. Biology and Philosophy, 17(1),
55–70.
Nielsen, T. I. (1963). Volition: A new experimental approach. Scandinavian
Journal of Psychology, 4, 225–230.
Olivares, F. A., Vargas, E., Fuentes, C., Martínez-Pernía, D., & Canales-Johnson,
A. (2015). Neurophenomenology revisited: Second-person methods for the
study of human consciousness. Frontiers in Psychiatry, 6. doi.org/10.3389/
fpsyg.2015.00673
Oppenheim, P., & Putnam, H. (1958). Unity of science as a working hypothesis.
Minnesota Studies in the Philosophy of Science, 2, 3–36.
Parnas, J., Møller, P., Kircher, T., Thalbitzer, J., Jansson, L., Handest, P., & Zahavi,
D. (2005). EASE: Examination of anomalous self-experience. Psychopathology,
38(5), 236–258.
Petitmengin, C., Baulac, M., & Navarro, V. (2006). Seizure anticipation: Are neu-
rophenomenological approaches able to detect preictal symptoms? Epilepsy &
Behavior, 9(2), 298–306.
Petitot, J., Varela, F. J., Pachoud, B., & Roy, J.-M. (Eds.). (1999). Naturalizing
phenomenology: Issues in contemporary phenomenology and cognitive sci-
ence. Stanford University Press.
Płotka, W. (2020). A controversy over the existence of fictional objects: Husserl
and Ingarden on imagination and fiction. Journal of the British Society for
Phenomenology, 51(1), 33–54. doi:10.1080/00071773.2019.1629553
Pokropski, M. (2019). Phenomenology and mechanisms of conscious-
ness: Considering the theoretical integration of phenomenology with
the mechanistic framework. Theory & Psychology, 29(5), 601–619. doi.
org/10.1177/0959354319868769
Roy, J.-M., Petitot J., Pachoud, B., & Varela F. J. (1999). Beyond the gap: An intro-
Quine, W. V. (1969). Epistemology naturalized. In Ontological relativity and
other essays (pp. 69–90). Columbia University Press.
Salmon, W. C. (1989). Four decades of scientific explanation. University of
Pittsburgh Press.
Sellars, W. (2007). Philosophy and the scientific image of man. In K. Scharp
& R. B. Brandom (Eds.), In the space of reasons: Selected essays of Wilfrid
Sellars. Harvard University Press.
Smith, B., & Ceusters, W. (2010). Ontological realism: A methodology for coor-
dinated evolution of scientific ontologies. Applied Ontology, 5(3), 139–188.
Stanghellini, G., Broome, M., Fernandez, A. V., & Raballo, A. (Eds.). (2019). The
Oxford handbook of phenomenological psychopathology. Oxford University
Press.
Thompson, E., Noë, A., & Pessoa, L. (1999). Perceptual completion: A case study
in phenomenology and cognitive science. In J. Petitot, F. J. Varela, B. Pachoud,
Thompson, E. (2007). Mind in life: Biology, phenomenology, and the sciences of
mind. Harvard University Press.
Wheeler, M. (2005). Reconstructing the cognitEive world: The next step. MIT
Press.
Wheeler, M. (2013). Science friction: Phenomenology, naturalism and cognitive
science. Royal Institute of Philosophy Supplement, 72, 135–167.
van Gelder, T. (1999). Wooden iron? Husserlian phenomenology meets cognitive
science. In J. Petitot, F. J. Varela, B. Pachoud, & J.-M. Roy (Eds.), Naturalizing
phenomenology: Issues in contemporary phenomenology and cognitive science
(pp. 245–265). Stanford University Press.
van Gelder, T., & Port, R. F. (1995). It’s about time: An overview of the dynamical
approach to cognition. In R. F. Port & T. van Gelder (Eds.), Mind as motion:
Explorations in the dynamics of cognition (pp. 1–43). MIT Press.
Varela, F. J. (1999). The specious present: A neurophenomenology of time con-
sciousness. In J. Petitot, F. J. Varela, B. Pachoud, & J.-M. Roy (Eds.), Naturalizing
Vermersch, P. (1994). L’entretien d’explicitation. Editions ESF.
Yoshimi, J. (2007). Mathematizing phenomenology. Phenomenology and the
Zahavi, D. (2010). Naturalized phenomenology. In S. Gallagher & D. Schmicking
(Eds.), Handbook of phenomenology and cognitive science (pp. 3–19). Springer.
Zahavi, D. (2013). Naturalized phenomenology: A desideratum or a category
mistake? Royal Institute of Philosophy Supplements, 72, 23–42.
3 Models of Explanation in Cognitive
Science
3.1 Introduction
Science is thought to describe, categorize, predict, and explain phenomena
that occur in the world. Between them, however, explanation is conceived
as the core of scientific endeavor (Nagel, 1961). Despite the distinct role
of explanation in science, just what explanation is and what it means to
explain something remains notoriously ambiguous. In fact, one of the
great debates in 20th-century philosophy of science concerned the nature
of scientific explanation. In this chapter, I introduce a few important con-
cepts, distinctions, and types of explanation that are relevant to the topic
of this book. I address neither all of the available conceptions of explana-
tion nor all of the related issues, as doing so would take us well beyond
the scope of this work (for an overview, see, e.g., Salmon, 1989).
The chapter is structured as follows: I introduce the explanatory mod-
els applied in cognitive (neuro)science and the project of mechanistic
integration of cognitive science and consider how phenomenology fits
into this picture. First, I present the theoretical background of the sci-
entific explanation debate and discuss the key models of explanation,
including the deductive-nomological model, personal explanations, func-
tional explanations, dynamical explanations, and, last but not least, the
mechanistic model of explanation. The mechanistic model of explanation
is especially important because it provides an integrative framework for
cognitive science, according to which different research fields contribute
to a multilevel mechanistic explanation by providing constraints on the
space of mechanisms. At the end, I consider what kind of explanations,
if any, are put forward by phenomenology and whether phenomenology
could be integrated with other explanatory models and feature as part of
a multilevel explanation. I argue that phenomenology does not provide
a genuine model of explanation but rather that it provides a constitutive
understanding of mental phenomena in question. Despite this, I hypoth-
esize that phenomenological constitutive understanding can provide con-
straints on mechanistic explanations, namely, functional and dynamical
constraints.
DOI: 10.4324/9781003035367-5
Models of Explanation in Cognitive Science 73
3.2 Scientific Explanation—Background
It is generally agreed upon that an explanation should deliver us a better
understanding of the explanandum phenomenon, that is, the phenom-
enon we want to explain. More specifically, an explanation consists of a
phenomenon to be explained, assumptions that amount to the explanans,
and a dependence relation that shows how the former and the latter are
related (Bangu, 2017, pp. 105–106). A classic example of this depen-
dence relation is the logical relation endorsed in the deductive-nomo-
logical (D-N) model, although causal explanations replace it with causal
dependence. This is the general idea of explanation. But phenomena may
be of a different nature, complexity, and regularity. Depending on these
aspects, explanations can vary in form and scope. Physical phenomena
that occur regularly, such as tides and eclipses, are typically explained
according to the D-N model (Hempel & Oppenheim, 1948), that is, by
subsuming their occurrences under a scientific law. The scientific law can
either take the form of a universal law of nature or a lawlike statistical
generalization, as it is in the deductive-statistical (D-S) and inductive-
statistical (I-S) models (Hempel, 1965). A given explanation can have dif-
ferent levels of certainty depending on the kind of law used and its form
of inference. D-N explanations, in which the explanans includes univer-
sal laws of nature, are certain. When the lawlike statement is a statistical
generalization, the explanation is not certain but probable.
We expect a different kind of explanation for singular phenomena,
such as the extinction of the dinosaurs. These explanations rather indi-
cate a cause, a combination of causes, or a causal mechanism than some
universal law of nature (e.g., Scriven, 1975). There are also phenomena
that are not strictly speaking natural and depend on the sociocultural con-
text (e.g., Watkins, 1957). For example, an explanation of the Bolshevik
Revolution would refer to various historical, economical, and social
causes and factors. Biological phenomena, such as respiration, require
yet another type of explanation, namely, mechanical explanation, which
relies on describing the causal mechanism responsible for the target phe-
nomenon. Finally, there are mental phenomena concerning which there is
still debate over what model of explanation is optimal. On the one hand,
there are psychological phenomena to which different models of explana-
tion apply, including causal mechanistic, functional, and statistical mod-
els. For example, the cognitive function of memory can be explained in
a mechanistic manner by establishing levels of organization with respect
to the underlying neural mechanisms (e.g., for the mechanistic model of
spatial memory in rats, see Craver, 2007). But explanations of memory
can also abstract away from the details of its physical realization and
focus on its functional organization (e.g., Baddeley, 2007). Another set
of examples of mental phenomena that are addressed from different
perspectives are mental maladies, such as schizophrenia, explanations
of which have been proposed on the grounds of belief-desire psychol-
ogy (Campbell, 2002), phenomenology (Sass, 2014), and neurobiology
(Andreasen, 1999). On the other hand, there is a class of mental phenom-
ena that relates to the issue of consciousness—to first-person experience.
It is argued that these subjective phenomena are the “hard problem” of
cognitive science, and there is no consensus with respect to which model
of explanation is the best for tackling them (Chalmers, 1996). Different
proposals for explaining consciousness vary from purely theoretical and
top-down (e.g., Rosenthal, 2005) to bottom-up and empirical (e.g., Prinz,
2012).
According to Salmon (1989, pp. 117–121), we can roughly distinguish
three conceptions of scientific explanation: modal, ontic, and epistemic.
Modal explanations tell us what the modal status of an explanandum
phenomenon is in terms of nomological necessity and possibility; for
example, they communicate that in a given situation, an event had to hap-
pen because it instantiated a law of nature. Ontic explanations assume
that events occur in a world that is full of regularities. These regulari-
ties can be of different sorts; usually, they are causal, deterministic, and
expressed in universal laws of nature. Such explanations rely on fitting
the explanandum event into these causal patterns. Finally, the epistemic
conception conceives explanations as arguments that consist of different
sorts of scientific representations, such as lawlike sentences, graphs, mod-
els, and so on. According to this conception, explanations are the product
of scientific work.
Addressing the epistemic conception, notice that the type of explana-
tion called for also depends on the questions we ask about the target
phenomenon, and so our epistemic interests (e.g., Grobler & Wiśniewski,
2005). It is thought that the question of “why” is what scientific explana-
tion ought to answer. But there are different senses in which we ask “why,”
and explanatory answers may differ in form. Why does water boil? Why
did the Bolshevik Revolution take place? Why do humans have lungs?
Why do some people gamble? These questions, despite all asking “why,”
require different explanatory strategies to formulate an answer. We can
also ask “what.” The “what” question is related to causal explanation
(e.g., Salmon, 1984; Scriven, 1975). What caused the extinction of the
dinosaurs? What was the cause of that car accident? What is the cause of
Creutzfeldt–Jakob disease? Answering these questions requires indicat-
ing a cause or a collection of causes relevant to the explanandum phe-
nomenon. Establishing causal relevance is key to finding the explanation.
Not all causal explanations require knowing the actual causal mecha-
nism responsible for the target phenomenon. Thus, this type of explana-
tion is sometimes called a “bare causal explanation” (Glennan, 2017, pp.
224–228) and is distinguished from causal mechanical explanations. A
causal-mechanical explanation answers the question of “how,” that is,
how a specific phenomenon was or is being produced. Answering such
a question usually requires discovering the necessary conditions for the
phenomenon to occur and describing the underlying mechanism along
with its internal organization. For example, an explanation of how blood
circulates in the human organism requires describing a multilevel mecha-
nism, including components such as the heart, veins, arteries, blood, and
so on, which generate and sustain the circulation. The mechanistic model
of explanation is the most common explanatory strategy in the life sci-
ences (for an overview, see Craver & Darden, 2013) and neuroscience
(Craver, 2007).
By accepting the scientific standpoint, we agree that all natural phe-
nomena have natural causes, but it does not follow that all explanations
have to be causal or causal-mechanical. To put it differently, there are sci-
entific explanations that are not causal (see Glennan, 2017, pp. 230–236).
Examples of this sort of explanation are some dynamical explanations
(e.g., Chemero, 2000; van Gelder, 1998) or what Elliott Sober (1983)
calls “equilibrium explanations.” Consider a physical system the state
of which changes in time, say, an iron ball rolling on a curved surface.
Obviously, this is a causal process, but “what the equilibrium explanation
does that makes it non-causal is that it describes the features of the space
that constrain the causal process, rather than describes the causal process
itself” (Glennan, 2017, p. 231). In general, an equilibrium or dynamical
explanation shows how a target system moves through its state space
and reaches its equilibrium together with what the key parameters are
for describing the process. In the following sections, I address selected
explanatory models in more detail.
3.3 Models of Explanation in Cognitive Science

3.3.1 Deductive-Nomological Model of Explanation
The deductive-nomological model of explanation (Hempel & Oppenheim,
1948), also known as the D-N model, was dominant in the 1950s (see
Salmon, 1989, pp. 11–32). Since then, it has lost its influence, but it is still
a significant point of reference for contemporary discussions of explana-
tions. The D-N model is also related to the idea of unification in science
as related to the conception of inter-theoretic reduction. But, as I argue
later (Section 3.4.1), unification and the strong conception of nomologi-
cal reduction are under severe criticism in today’s cognitive science, and
what we are seeing is the revival of explanatory pluralism and the idea of
theoretical integration.
The D-N model states that scientific explanation rests on deductive
argument and relies on deducing the explanandum phenomenon (usually
an empirical phenomenon that occurs regularly in the world and that can
be described in observational terms) from premises (explanans), among
which at least one is a general law of nature (other premises concern,
for instance, initial or precipitating conditions). Laws of nature are dis-
tinguished in this model from mere generalizations. They are character-
ized by their universality (they have unlimited scope and are true for all
objects of some specific kind), support counterfactuals (tell us what may
happen if such and such were the case), and have modal import (tell us
what is necessary or possible; Salmon, 1989). Examples of general laws
of nature include Newton’s laws of motion and the limit set by the speed
of light.
The D-N model is the most common form of explanation in physics.
The model can be modified in such a way that instead of universal laws
it uses statistical generalizations. We can refer to this modified type of
explanation as either deductive-statistical or inductive-statistical depend-
ing on the type of explanandum phenomenon. For example, we can use
the inductive-statistical (I-S) model to explain that some individual has
the flu by reference to the fact that they had contact with a person who
had the flu if we accept a premise corresponding to the statistical general-
ization that contact with someone who has the flu is highly likely to result
in infection. In this case, however, the explanandum is not nomologically
necessary as it is in deductive inferences but is rather only nomologically
expected in virtue of its statistical premise. In short, the structure of the
D-N model of explanation looks as follows:
1. The conclusion is the explanandum (a target phenomenon that is

either a singular event or a regular pattern).
2. The premises are the explanans (statements that include at least one
universal law of nature or statistical generalization).
3. The premises should be true and have empirical content.
This neat and formal model of explanation fits physical science best, but
even then there are counterexamples that it cannot handle and that lead
to paradoxes. The most famous counterexamples come from Salmon
(1989, pp. 46–50). Three of them are especially interesting in the context
of this chapter. First, the barometer case—assuming that there is a gen-
eral law stating that each time the reading of a barometer falls a storm is
coming, we can infer from noticing a drop in the reading of a barometer
that a storm is coming. Such reasoning fits perfectly well to the D-N
model. However, a drop in the reading of a barometer is not causally
relevant to storms and thus cannot explain them, although it can serve as
a good predictor. Both of these events have a common cause—a drop in
atmospheric pressure—and a full explanation of a storm should include
generalizations about such conditions.
Second, the flagpole example: a vertical flagpole stands on flat ground
and casts a shadow. Knowing the length of the flagpole, the position of
the sun, and the law of rectilinear propagation of light (premises included
in the explanans), we can deduce, in accordance with the D-N model,
the shadow’s length (explanandum). But we can reverse this reasoning
and from knowledge of the length of the shadow, the position of the
sun, and the previously mentioned law, we can infer the length of the
flagpole. If that is the case, then the length of the flagpole is explained
by the shadow it casts. That conclusion seems counterintuitive, as we are
inclined to think that the flagpole causes the shadow and not vice versa.
Explanations should not run in both directions in that sense, they should
not be symmetrical, precisely because that leads to absurd consequences.
The third example concerns eclipses. We can explain the past occurrence
of a lunar eclipse by deducing it from the current positions of the earth,
sun, and moon and the laws of celestial mechanics. And we can produce an
analogous inference deducing future occurrences of eclipses. Here we touch
on a similar problem we did with symmetry earlier. The D-N model does
not recognize temporal restrictions and so does not distinguish between ret-
rospective and predictive explanations. To put it differently, if we knew the
relevant laws of nature before the explanandum phenomenon happened,
then we could predict its occurrence, and the logical form of such a predic-
tive inference would be the same as the retrospective deduction. But expla-
nations are different from predictions. For example, consider the symptoms
of a disease. Symptoms can be used as predictors for the disease that causes
them. And the disease can serve as an argument in the prediction of symp-
toms. But only the disease explains the symptoms, not vice versa.
The D-N model seems least applicable to biology, in which, as it is
argued (e.g., Bechtel, 2008; Godfrey-Smith, 2014), there are no strict
laws. This does not mean, however, that there are no patterns in biology.
For instance, Mendel’s first law was considered an example of a biologi-
cal (genetical) regularity. The “law” captures patterns in gene inheritance,
which are not incidental. But there are many exceptions to it, such as
Down syndrome in humans, so the “law” is not universal. Hence, biology
is able to capture regularities occurring in organisms and their behav-
ior, yet its laws are essentially different from the universal laws of phys-
ics. According to Peter Godfrey-Smith (2014), “rather than a two-way
distinction between laws and accidental regularities, biological patterns
show different amounts of what can be called resilience or stability” (p.
14). Such biological patterns are better explained according to the mech-
anistic model rather than the D-N model.
It is also hard to find universal laws in psychology. There are a few
candidates for lawlike statements, such as Weber’s psychophysical law
(ΔI/I = k), where I stands for the intensity of a stimulus, ΔI is the mini-
mum detectable increment over the stimulus intensity, and k is a constant
value (Bechtel & Wright, 2009). But such nomic regularities, according
to William Bechtel and Cory D. Wright, should themselves be subject to
explanation. Similarly, Robert Cummins (2000) argues that regularities
in psychology are called effects and that they are something that scien-
tists want to explain rather than use as explanans. Generally speaking,
psychological explanations typically refer to psychological states, such
as beliefs and desires, and thus fall under personal explanations or use
concepts of functions, functional modules, and cognitive capacities as in
the functional type of explanation. That being said, under the influence
of rapidly developing neuroscience, explanations of mental phenomena
have been more recently formulated in terms of neural mechanisms.
3.3.2 Personal Explanations
The distinction between the personal and sub-personal level of explana-
tion was proposed by Daniel Dennett (1969) in his book Content and
Consciousness. According to Dennett, explaining human behavior can
take two forms: one personal which relies on concepts of mental states,
such as pain, and one sub-personal which refers to mental states’ func-
tional organization and physical realization. Dennett’s objective was to
show that sub-personal explanations can be understood as a type of psy-
chological explanation. The idea of personal explanations was adapted
by John McDowell and the so-called Pittsburgh school (Drayson, 2014).
McDowell (1986) extended the scope of personal explanations to all
sorts of mental states with propositional content such as beliefs, desires,
and intentions that are thought to be reasons for action. Accordingly, we
can explain human behavior by indicating those reasons which under-
lie the studied behavior. In his later works, Dennett (1981) developed
this personal type of explanation and connected it with the idea of an
explanatory strategy called the “intentional stance.” Adopting the inten-
tional stance allows to us treat studied subjects as rational agents and
to explain and predict their behavior in terms of different sorts of inten-
tional states, including beliefs, in an analogous manner to the “functional
stance,” in which the target system is described in functional terms. At
this stage, Dennett’s explanatory strategies were instrumental; that is,
the intentional strategy was a useful tool for explaining and predicting
the behavior of rational agents, but it did not tell us anything about the
true nature of mental states themselves. Later (e.g., Dennett, 1991b), he
shifted his position toward realism and characterized mental states in
dispositional terms. According to Zoe Drayson (2014), the Pittsburgian
version of personal-level explanation differs from the Dennettian version
in several important respects. The most important one concerns the nor-
mative character of personal explanation related to the assumption that
humans are rational agents, which was stressed by McDowell and his
colleagues. Thus, according to the Pittsburgian picture, personal explana-
tions are a different type of explanation than sub-personal ones, which
rely on causal constraints and natural laws.
Personal explanations are often considered to be at the top of the hier-
archy of explanatory levels, above the functional and physical levels. It
is often argued that the personal level of explanation, which is taken to
be distinct and autonomous from other levels, does not require refer-
ence to explanatory facts from other levels. Autonomy is also expressed
therein, given that personal explanations can only appeal to psychologi-
cal states, which means that they ultimately run out; that is, the moment
we employ a nonpsychological state, we leave the personal level and
move to the sub-personal (Dennett, 1969). According to others (e.g.,
Bermúdez, 2000) personal explanations, which are also sometimes called
horizontal explanations, are compatible with the vertical type of explana-
tions common in cognitive science. A vertical explanation shows how a
certain mental state or cognitive capacity is realized by a mechanism or
a set of mechanisms on a sub-personal level. Proponents of the explana-
tory autonomy of personal explanations hold that vertical sub-personal
explanation reveals only enabling conditions; that is, it only shows how
the personal level is physically possible.
However, there are good arguments suggesting that sub-personal
facts can play a more significant role, which in consequence breaches
the autonomy thesis (Bermúdez, 2000; Colombo, 2013). The arguments
usually refer to abnormal or irrational behaviors, such as those in psy-
chopathological cases, which contradict the rationality constraint crucial
for personal explanations. Cases such as blindsight, unilateral neglect,
or addiction show that there are mental phenomena for which personal
explanations are incomplete or just impossible. Thus, it is argued that the
sub-personal can also be constitutive for explanations of mental phenom-
ena. It is important to emphasize that the concept of constitutive rele-
vancy is understood here in accordance with what McDowell (1994) calls
constitutive understanding or constitutive explanation, which is different
from mechanistic constitutive explanation (Craver, 2007). According to
McDowell, constitutive explanation relies on conceptual distinctions and
logical relations and provides us with an understanding of how and why
people behave in a certain way, in particular what rational motives and
norms make a specific mental phenomenon what it is. So a constitutive
explanation is a personal level explanation with a rationality constraint. In
contrast, an enabling explanation deals with causal relations and empiri-
cal facts, which makes certain mental phenomena possible but does not
explain their nature. McDowell separates these types of understanding,
but Colombo (2013) argues that at least “some facts about subpersonal
states and events are constitutively relevant to some personal-level phe-
nomena, and therefore can, and sometimes should, inform personal-level
explanations” (pp. 567–568). For example, addiction can be considered
a phenomenon the explanation of which requires understanding a neural
mechanism—for example, the specific activity of dopamine in the reward
circuit. Rewards as conceptualized on the personal level, such as acquir-
ing money, are different from a reward for the brain, which is a flood
of dopamine in the case of addiction. Understanding these sub-personal
facts about the functioning of an addicted brain not only indicates the
enabling conditions but, more important, gives us an understanding of
the whole phenomenon of addiction and thus helps us to understand and
predict the behavior of an addicted person. It also helps us distinguish
genuine addiction, such as a gambling addiction, from other behaviors
that, on the personal level, might look similar but do not share this neural
characteristic. If that is the case, then the conceptual apparatus employed
in personal level explanations can be revised in the light of sub-personal
facts provided by empirical cognitive science.
To conclude, roughly speaking, personal level explanations are a dis-
tinct type of explanation of human behavior in terms of beliefs, desires,
and intentions (BDI). Their autonomy is, however, doubtful, as there are
cases of first-person phenomena that require information from the sub-
personal level in order to explain. In those cases, sub-personal facts are
not merely enabling or causal but also constitutive for our conceptualiza-
tion and understanding of the target phenomenon.
3.3.2.1 Heterophenomenology Explained
Much of the debate about personal-level explanations concerns the reli-
ability of first-person evidence. Although many psychological studies
were built on introspective reports, such as the classic studies on prob-
lem-solving conducted by Allen Newell and Herbert A. Simon (1972),
introspection itself is still a controversial topic. On the one hand, there
is no consensus on whether the source of introspective knowledge is sin-
gular or plural, that is, whether it consists of multiple processes (e.g.,
Komorowska-Mach, 2019; Schwitzgebel, 2012). Furthermore, it is often
argued that introspection is an unreliable source of evidence because sub-
jects are often unaware of their subjective responses and give meaning
to their experiences with antecedently held theories (Nisbett & Wilson,
1977). It is also argued that introspective reports are unverifiable and
have no truth value (Dennett, 1991a). On the other hand, it is difficult
to imagine a study of consciousness without any kind of first-person
insights. Thus, some researchers (e.g., Dennett, 1991a, 2003; Hurlburt
& Schwitzgebel, 2007; Jack & Roepstorff, 2002; Piccinini, 2003, 2010)
argue for a more developed and rigorous methodology for validating
and analyzing first-person data generated on the basis introspection. The
results of such methods could serve as evidence supporting first-person
explanations and could contribute to a broader explanatory framework
that integrates different research fields, including those using the third-
person methods of neuroscience.
Talking about first-person data raises the issue of phenomenology. For
Dennett, first-person phenomenology is not explanatory because men-
tal states, such as pain, are unanalyzable qualities (e.g., 1969) and sole
introspection, which Dennett (1991a) erroneously equates with Husserl’s
phenomenological method, is an unreliable source of evidence. I share
Dennett’s skepticism about qualia, but his reading of Husserlian phenom-
enology is wrong. As I argued in Chapter 2, phenomenological method
differs significantly from introspection and is not interested in studying
the ineffable qualities of particular experiences but in general structures
of experience (for other arguments against Dennett’s reading of phenom-
enology, see, e.g., Gallagher & Zahavi, 2012; Zahavi, 2007). According
to Dennett, the only reliable phenomenology is heterophenomenology
(1991a), which is a third-person phenomenology or, to be more precise,
an analysis of the studied subject’s verbal reports in terms of BDI. In
brief, the method consists of two steps. First, the experimenter records
the verbal utterances of the studied subject. These recordings are the
raw data that are then interpreted in terms of beliefs. The experimenter
adopts an “intentional stance” (Dennett, 1981) and ascribes beliefs to the
subject about their experiences. As Dennett (2003) writes, “these are the
primary interpreted data, the pretheoretical data, the quod erat explica-
tum (as organized into heterophenomenological worlds), for a science of
consciousness” (p. 21). Next, the experimenter “composes a catalogue of
what the subject believes to be true about his or her conscious experience”
(Dennett, 2003, p. 20, emphasis in the original). Importantly, the experi-
menter does not assume that the subject is right but only acknowledges
that it is what the subject believes. For example, in simple perceptual
tasks, when a subject says that they saw a light, a heterophenomenologist
cannot acknowledge that the subject actually saw a light; the hetero-
phenomenologist is only allowed to infer that the subject believes that
they saw the light. So heterophenomenological method is interested in
the subject’s beliefs about first-person experience, not in the experience
itself. The explanatory question that a heterophenomenologist can try to
answer is why a subject in an experimental situation has a specific set of
beliefs about his or her mental life.
Dennett (2005) claims that heterophenomenology is a widely accepted
approach to first-person evidence, and “it has been practiced, with vary-
ing degrees of punctiliousness about its presuppositions and prohibitions,
for a hundred years or so, in the various branches of experimental psy-
chology, psychophysics, neurophysiology, and today’s cognitive neurosci-
ence” (p. 36). Even if that is the case, it is so not without controversy.
The very idea that first-person evidence is private and incorrigible is
already doubtful. On the one hand, one can argue from phenomeno-
logical positions that our mental lives are to some extent observable and
public thanks to bodily expressions (e.g., J. Smith, 2010). On the other
hand, one may concede that first-person evidence is private without con-
ceding that it should be removed from science. It just means that we
need to develop a methodology for how to process first-person data and
extract information from it. Heterophenomenology sets aside our mental
lives, our conscious experience, and focuses on beliefs about one’s mental
states. Why should we call it “phenomenology” then? Phenomenology, as
well as any kind of first-person methodology, studies experience or, to put
it more abstractly, mental states, not just beliefs.
Dennett argues that heterophenomenology suspends judgments about
the truth value of subjects’ beliefs in a manner similar to Husserl’s method
of bracketing (see Chapter 1). The difference is, however, that Husserl’s
method is performed by a phenomenologist on their own beliefs and
judgments. In Dennett’s proposal, the method of bracketing is performed
by a researcher and concerns the studied subject’s beliefs. Thus, hetero-
phenomenology relies on a naïve attitude that tacitly assumes that the
observer’s beliefs about the subject are true. Recognition of the observ-
er’s biases touches upon the issue of “primary interpreted data,” which
according to Dennett (2003) is “pretheoretical” (p. 21). It seems that
Dennett uses this term because the interpreted data precedes a theory
of consciousness. However, if the data is interpreted in terms of BDI, it
cannot be completely pretheoretical, since belief is a folk-theory concept.
Belief talk presupposes that thinking has a propositional character, and
that is clearly a theoretical statement. As Piccinini (2010) points out, a
heterophenomenologist who applies the concept of belief is committed
to at least two problematic assumptions: first, that first-person reports
are always mediated by beliefs, that is, they are not directly caused by
perceptual or emotional experiential states, and, second, that first-person
reports are adequate representations of the contents of beliefs. The for-
mer assumption seems implausible since it is possible that some cases
of first-person reports, especially some first-person behaviors, may fol-
low directly from an experience; for instance, a painful experience can
directly cause a verbal expression. The latter assumption breaks down in
the light of an experiential subject’s confabulations or cognitive distor-
tions. Obviously, we can try to control the process to rule out confabula-
tions and other errors in reports, but why not, as Piccinini argues, cut
out the middle level corresponding to beliefs and think about reports as
direct expressions of mental states. After all mental states are what we
want to investigate and explain.
Taking these arguments into account, Piccinini proposes an improved
version of heterophenomenology. The improvements are based on his ear-
lier proposal of a third-person methodology for first-person data in which
subjects are considered self-measuring instruments generating first-person
behaviors (Piccinini, 2003, 2010). The aim of the method is to validate
first-person data and extract information about the subject’s mental states.
This version of heterophenomenology features several important revisions.
First, he extends the notion of first-person data to all first-person behav-
iors, not just utterances. That is completely fine with Dennett (2003), who
himself suggests such possible extensions. However, in contrast to Dennett,
Piccinini undermines the mediating role of beliefs in producing first-person
behavior, which is caused directly by occurrent mental states. Furthermore,
he argues against agnosticism concerning the subject’s report. For him,
we should embrace our everyday attitude toward others’ expressions and
beliefs and evaluate the credibility of first-person data. If the studied subject
is nothing more than a self-measuring instrument, then they can be treated
similarly to a telescope, the output of which is validated to check whether it
shows what it is supposed to show. Evaluation of the subject’s first-person
report is based on commonsense folk-psychology, and one should express
the platitudes one refers to. This is a different strategy for how to deal with
the observer’s tacit assumptions and theory ladenness than the one applied
in Husserlian phenomenology and Dennett’s heterophenomenology, which
was to bracket such beliefs as much as possible. Piccinini argues for making
them public in order to control these aspects in an intersubjective fashion.
Self-measurement methodology also drops the incorrigibility claim:
Like other measurements, self-measurements are neither incorrigible

nor always reliable. It is the job of the external observers to calibrate
the instrument, set up the experiment, use the instrument, and inter-
pret the data so as to rule out confounding factors and establish what
phenomena can be inferred from the data.
(Piccinini, 2010, p. 100)
3.3.3 Functional Explanations
In philosophy of mind and cognitive science, functionalism can be
approached not only as a metaphysical conception of mental states, which
argues mental states are functional states individuated by causal relations
to other mental states and inputs and outputs (e.g., Block, 1980), but also
as an explanatory strategy called functional analysis. Functional explana-
tions were initially recognized as a way of explaining biological as well
as social phenomena (e.g., Nagel, 1961; Salmon, 1989) in teleological
terms such as goal, purpose, and function. It is important to notice that
we may operate with two different notions of function: a systemic and an
etiological one (e.g., Godfrey-Smith, 2014, pp. 59–65). The systemic con-
ception of function understands it as the causal role of some component,
which it plays in the studied system in such a way that it contributes to
the capacity we want to explain. For example, door keys contribute to
opening doors, but under certain conditions, they can serve as a bottle
opener. So, in the systemic conception of function, the ascription of a
function to a system’s part may vary depending on one’s explanatory
interests. The etiological conception of function, which is more common
in biology, focuses on the question of why the component is in the target
system (why is it there?) and seeks answers in the effect that this part
produces. This notion of function relates to the proper function of the
system’s component in question. So, with etiological notion of function,
we can explain, for example, why cats have claws, referring to the effect
claws have on catching prey and climbing trees.
Carl Hempel (1965) was one of first to describe the pattern definitive
of this functional explanatory strategy. In short, we take an explanandum
that is a relatively persistent feature of an item, say, an organ in an organ-
ism, and we show that in certain conditions the item’s feature obtains
and contributes to the system’s proper functioning. Hempel adopts the
etiological understanding of function. For example, we can explain why
there is a heart in an organism by appealing to the heart’s disposition
to pump blood in its vascular system, which, in turn, contributes to the
organism’s well-being. But we cannot explain the heart’s disposition of
emitting sounds that way, because its emitting sounds does not affect the
organism’s functioning. Hempel’s structure of functional explanations
has one important problem, namely, the problem of equivalent realiza-
tions. We cannot explain the presence of a heart in an organism by the
fact that it pumps blood because there are other possible mechanisms
that could realize the same pumping function. If that is the case, then the
functional strategy has no explanatory power and can only have heuris-
tic value. Attempting to overcome this issue relies on showing that the
analyzed item was purposefully placed in the system in order to realize
this disposition. In the case of natural systems, one can argue (e.g., Nagel,
1961) that, when we consider a specific organism, the target disposition
of an organ cannot be realized by anything else because the organ is a
result of evolution. Thus, that particular organ is evolutionarily optimal
for realizing the target disposition.
However, in the seminal paper “Functional Analysis,” Cummins (1975)
argues that the problem comes from assumptions underlying Hempel’s
functional model and confusing a teleological approach with a functional
one. According to Cummins, we cannot explain why a specific item is
part of a system functionally—what we can explain is the behavior of the
containing system. The explanandum is a system’s disposition (Cummins
calls it a capacity), which occurs with lawlike regularity (so-called dis-
positional regularity). The functional analytic strategy does not subsume
the disposition’s occurrences under a law but decomposes the target dis-
position into a number of sub-dispositions, the manifestations of which
amount to the occurrence of the analyzed disposition. Cummins adopts
the systemic notion of function because the ascription of a function to
an item depends on the contribution of the function to the explanandum
disposition.
Cummins argues for distinguishing teleological thinking from func-
tional thinking. The reason behind identifying the functional approach
with teleological reasoning relies on explicit or implicit reference to the
Aristotelian idea of the final cause, that is, the idea that some future state
can causally affect the present. For example, one could argue that the
present behavior of an animal, for instance hunting for prey, is deter-
mined by a future state (eating the prey). But functional explanations of
such cases do not have to lead to thinking in terms of the future causing
present behavior. Apart from Cummins’s functional analysis, the theory
of “consequence-etiology” (Wright, 1976) shows that we can separate
the teleological approach from the functional one and explain the occur-
rence of an animal’s behavior not through it pursuing some goal in the
future but through that behavior having been causally effective in achiev-
ing goals in the past. So, for example, hunting for prey is goal-directed
behavior that is caused not by the future goal (eating prey) but by past
hunting behavior that was effective in keeping the hunting organism well
nourished. Accordingly, the real cause of the explanandum behavior is
its efficacy in the past, which is a feedback process spanning a long time
scale. Importantly, the past in such explanations can mean, in this case,
the evolutionary history of the organism’s species.
In contemporary philosophy of science and philosophy of mind, there
are several types of functional analysis. Piccinini and Craver (2011)
distinguish three types of such analysis: functional analysis by internal
states, so-called boxology, and task analysis. Computational function-
alism’s preferred form of explanation is functional analysis by internal
states, since it explains the target capacity using a set of internal states
and their relations to each other and to inputs and outputs. The key idea
behind this version of functionalism is that mental states can be under-
stood as Turing machine states (Putnam, 1960) and the mind as a sort of
computer program. Internal states can be described in representational
terms and their manipulation as computations. The form of analysis
called boxology, related to graphical representations of cognitive mod-
ules, can be found in 20th-century cognitive psychology and was utilized
in Jerry Fodor’s conception of the mind’s modularity (Fodor, 1983). The
conception encapsulates cognitive functions and subfunctions in special-
ized modules that have inputs and outputs; they are domain-specific (they
process only information from a specific domain), informationally encap-
sulated (external processes do not have access to the information being
processed in a module), and mandatory in their operations.
Finally, there is the already introduced functional analysis as task anal-
ysis proposed by Cummins (1975, 2000). According to Cummins (2000),
“functional analysis consists in analyzing a disposition into a number
of less problematic dispositions such that programmed manifestation of
these analyzing dispositions amounts to a manifestation of the analyzed
disposition” (p. 125). This method is quite universal and can be applied
to artificial as well as natural systems, including cognitive systems. An
example of an artificial system is a production line, which is composed of
a number of simple tasks. An example of a natural system is a situation in
which cats hunt small birds, where the disposition to hunt should include
sub-dispositions such as recognizing birdlike objects, stealing, running,
claw-catching, and so on. Each of these sub-dispositions can be, in turn,
divided into yet simpler dispositions. Research on working memory is an
example of such functional decomposition applied to cognitive functions.
For instance, in the model proposed by Alan Baddeley (2007), working
memory is decomposed into four subfunctions: the central executive, the
visuospatial sketchpad, the phonological loop, and the episodic buffer.
This type of functional analysis can be roughly divided into three steps.
First, we identify the explanandum phenomenon, which in the case of
cognitive psychology is one of the target system’s cognitive capacities.
Then we define the capacity in dispositional terms; that is, the system’s
cognitive capacity φ is a complex dispositional property D, which means
that the system exhibits a lawlike regularity in its behavior. According
to Cummins, the best way to explain this kind of dispositional prop-
erty is decomposition of the disposition D into a set of sub-dispositions
d1, d2, …, dn. The final step is representing the system’s target capacity
decomposed in functional analysis in the form of a functional design. The
functional design represents the system’s functional architecture, which,
according to Cummins, explains the target phenomenon.
Cummins not only shows how the functional explanatory strategy
should proceed but also argues that the nature of psychological explana-
tion is functional. His line of thought is as follows: there are no general
laws of nature in psychology, thus those explanations cannot take the
form of subsumption under a law; regularities discovered by psychology
are not explanatory laws, but effects and capacities, which themselves
require explanation; an explanation of capacity consists in showing the
underlying functional structure. As Cummins concludes, “capacities
and their associated incidental effects are to be explained by appeal to
a combination of functional analysis and realization, and the currently
influential explanatory frameworks in psychology are all frameworks
for generating this sort of explanation” (Cummins, 2000). The frame-
works mentioned by Cummins include explanatory frameworks such as
BDI, computationalism, connectionism, the evolutionary approach, and
neuroscience.
Martin Roth and Cummins (2017) argue that functional analysis is a
distinct and autonomous explanatory strategy that delivers a functional
design of the target system’s capacity. Such a design does not entail struc-
tural decomposition; that is, it does not say anything about how specific
functions might map onto the components of a possible physical mecha-
nism. The autonomy of psychological explanations is defended in a simi-
lar vein (e.g., Barrett, 2014; Weiskopf, 2011). On the contrary, Piccinini
and Craver (2011) argue that functional explanations are neither autono-
mous nor distinct because they lack explanatory power. They can be per-
ceived as incomplete mechanistic explanations. I will return to the issue
of autonomy at the end of this chapter.
Cummins’s approach faces several important problems. One issue con-
cerns his conception of function relative to the arbitrariness of functional
analysis. In Cummins-style functional analysis, the way one decomposes
the target capacity, how many functions and subfunctions one includes in
the description, how many levels of functional organization are included—
they all depend on the interests and knowledge of the researcher. In other
words, there is no ultimate functional decomposition of a system but a
number of decompositions depending on what we want to explain and
what we already know about the target system and the explanandum phe-
nomenon. The problem results from Cummins’s (1975) systemic under-
standing of function: something has a function in the target system if it
contributes to the system’s performance (capacity) that we want to explain.
As Ruth Millikan (2002) notes, in Cummins’s functional explanations, we
do not ask why a system has a certain function but how it functions. This
approach uses a notion of function that is detached from the system’s (evo-
lutionary) history. Thus, Millikan proposes a different conception of func-
tion ascription to living systems called proper function, which is related to
the system’s evolutionary history. A proper function is a function of an ani-
mal trait, an organ, or animal behavior that is a result of natural selection.
The other problem with applying Cummins-style task analysis in
explanations of our mental lives is that dispositions occur under some
conditions. For example, salt has a disposition to dissolve in an envi-
ronment containing water, or a glass window has a disposition to
break when struck by something hard. To put it differently, a disposi-
tion always obtains when certain conditions are satisfied. This allows
one to see them as lawlike regularities dependent on causal conditions.
However, this is not always the case for mental dispositions, which often
are not exhibited with such regularity and for which conditions could be
difficult to define. Consider, for example, someone’s disposition to feel
touched when watching dramas. That behavior depends in large part on
the mood, place, companions, and, of course, the movie. But even if one
felt emotionally touched, their reaction can take various forms—one can
hide one’s emotions, someone else may cry, and so on. It is important,
therefore, to be aware that thinking in dispositional terms, and thus, in
accordance with a functional approach, has limitations in explaining our
mental lives. However, as I argue in Chapter 4, phenomenology shares
some important similarities with functional explanations.
3.3.4 Dynamical Explanations
Another major form of explanation corresponds to dynamical systems
theory (DST). It consists of a mathematical theory that enables modeling
and analyzing the behavior of dynamical systems. A dynamical system is,
generally speaking, any state-determined system the behavior of which
evolves in time. They may vary in nature and composition, from physical
systems through biological ones, up to the collective behavior of popula-
tions. Thus, DST finds application across the sciences, including physics,
biology, ecology, sociology, and recently cognitive science. DST calculus
is formed by differential equations or discrete equations in the case of
discrete-time systems. These equations include variables that represent key
features of the target system’s change over time and parameters that stand
for other conditions relevant to the behavior. A set of these equations con-
stitute a formal dynamical model of the target system. The target system’s
behavior can be represented as a trajectory in n-dimensional state space,
where n depends on the number of variables describing the system’s states.
DST offers methods for both quantitative and qualitative analysis of
the behavior of a dynamical system. Quantitative methods require formal
models, that is, equations that allow for simulating the target system’s
behavior. Qualitative analysis is a more general approach that focuses on
the overall behavior of the system and reveals the features of its long-term
dynamics, such as its stable and unstable states or attractors (regions in
the state space towards which the system evolves). The long-run behavior
of the system may be visualized with the use of diagrams, such as phase
portraits, bifurcation diagrams, or parameter charts (see Beer, 2000).
Besides quantitative analysis (mathematical model) and qualitative anal-
ysis (analysis of behavior on the basis of phase portraits, etc.), van Gelder
and Port (1995) discuss a more general and conceptual approach called
“dynamic description,” which relies on the use of DST categories, such as
stable state, attractor, and bifurcation, to describe the system’s behavior.
It is important to remember that, in this case, dynamical categories are
used in a highly abstract fashion unless a proper dynamical analysis is
produced. A dynamic description may be the first step in building a com-
plete dynamical explanation.
Although the dynamic approach in science, especially in physics, has a
long history (see, e.g., Abraham & Shaw, 1992), in cognitive science it is
relatively new. It is argued that dynamical systems theory can be used to
explain cognitive systems (e.g., Chemero, 2000; Kelso, 1995; van Gelder
& Port, 1995). In his seminal paper, “What Might Cognition Be, If Not
Computation?” van Gelder (1995) provocatively proposes to think about
cognition in terms of dynamical systems rather than computational ones.
Such an approach assumes that cognition is an ongoing process that
evolves in time and spans the brain, body, and environment. Van Gelder
rejects the computer metaphor and offers a new one, namely, the Watt’s
governor. It is a simple but elegant device designed in the 18th century
by James Watt for steam engines in order to automatically control the
throttle and maintain a constant flywheel speed (see Figure 3.1) despite
changing steam pressure. It consists of a spindle with two arms attached
to the flywheel. Metal balls are attached at the ends of these arms. Faster
spindle rotation moves the arms upward and slower rotation downward.
Importantly, the arms are linked to the engine’s throttle valve; when the
speed of the flywheel increases, the arms go up, which closes the valve
reducing the flow of steam and speed; when the wheel’s speed decreases,
the arms go down and the valve opens enabling a greater flow of steam.
The simple yet brilliant design allows for controlling the speed of the
Figure 3.1 Watt’s centrifugal governor.
flywheel, which was crucial for using steam engines, for example, in the
cotton industry.
According to van Gelder, the behavior of the device can be described
with a differential equation:
d 2 g d
n cos sin sin r
2
dt 2 l dt
where θ is the angle of the arms, t is time, n is a gearing constant, ω is the

engine’s speed, l is the arm’s length, g is the gravitational constant, and r is
a friction constant. The mathematical model of the governor allows one
to predict its behavior. Indeed, proponents of the dynamical approach
argue that we do not need anything else to explain its working.
In cognitive science, the dynamical approach to cognition is expressed
through the dynamical hypothesis, which, according to van Gelder (1998),
has two major components. The nature hypothesis is a claim about

the nature of cognitive agents themselves; it specifies what they are
(i.e., dynamical systems). The knowledge hypothesis is a claim about
cognitive science: namely, that we can and should understand cogni-
tion dynamically. Obviously, these are closely related; the best evi-
dence for the former would be the truth of the latter.
(p. 619)
For Anthony Chemero (2000), the epistemological part of the dynami-
cal hypothesis can be defended with the dynamical stance, analogous to
Dennett’s (1981) explanatory strategy involving the intentional stance. In
the intentional stance, a target system is treated as a rational agent having
beliefs, desires, and other intentional mental states in order to explain and
predict its behavior. A successful prediction and plausible explanation are
the best evidence that the intentional stance is the correct explanatory
strategy for this type of system. The dynamical stance is also an epistemo-
logical attitude, in which we treat the target system as a dynamical system
and explain and predict its behavior using the mathematical and concep-
tual tools of dynamical systems theory. For instance, in the dynamical
stance we say that the system bifurcates; that is, it passed through a phase
transition from one stable state to another. Importantly, the dynamical
stance suspends claims about the true nature of the system; thus, it is
independent of the ontological part of the dynamical hypothesis.
Explanations of cognitive phenomena formulated in the dynamical
stance capture their temporal structure, that is, how the system’s behavior
changes over time. For some proponents of the dynamical approach (e.g.,
Chemero, 2011), this predictive power with respect to the target sys-
tem’s behavior eliminates the need for other stances, such as a representa-
tional stance. Thus, dynamism is sometimes introduced as committed to
anti-representationalism and opposed to classic computational cognitive
science, such as Fodor’s language of thought. Cognition emerges from
dynamical interaction between an organism and an environment rather
than being reduced to processing symbolic representations. Moreover, it
is argued that the concept of representation is not necessary to explain
and predict a system’s behavior at all. Even if the target system has some
sort of internal representations, referring to them in an explanation is not
mandatory. What is explanatory is the target system’s phase portrait and
ultimately its mathematical model, which describes the system’s behavior.
Consider the Watt’s governor. For defenders of the dynamical stance, the
device is nonrepresentational, and its behavior can be explained in non-
representational dynamical terms. Once we capture the key parameters
and variables for the governor’s functioning, we are able to explain it.
However, it is also possible to think of the governor as a representational
and computational device (Bechtel, 1998); for instance, we can treat the
angle between the spindle and rotating arms as representing the engine’s
speed. We can certainly do so, but are we accordingly explaining some-
thing new? Chemero’s (2000) answer is clear and negative: “despite the
fact that a representational gloss is possible, once one has the dynamical
explanation, the representational gloss does not predict anything about
the system’s behavior that could not be predicted by the dynamical expla-
nation alone” (p. 638).
So far so good, but the challenge for the dynamical model of explana-
tion is establishing what kind of phenomena we can explain using this
strategy. In cognitive science, the dynamical approach has been applied to
explanations of various phenomena (see, e.g., van Gelder & Port, 1995)
but mostly to different sorts of motor and sensorimotor behavior, for
example, the famous dynamical model of oscillatory finger movement
known as the Haken–Kelso–Bunz (HKB) model (Haken et al., 1985) or
the dynamical explanation of A-not-B error in infant reaching (L. Smith
& Thelen, 2003). Some argue that the dynamical approach to cognition is
incapable of explaining higher cognitive functions. For example, accord-
ing to Chris Eliasmith (2001), dynamic descriptions are too low-dimen-
sional; that is, the state space in question consists of too few dimensions,
and thus, they are not flexible enough to capture the complexity of higher
cognitive processes. Of course, there are certain limits in dynamical mod-
eling of complex and multifaceted phenomena, but the objection seems
exaggerated. An interesting example of a more complex (16 dimensions)
dynamical model of cognitive capacity is the dynamical model of percep-
tual object categorization proposed by Randall Beer (2000, 2003). It is a
dynamical model of an artificial agent that discriminates between objects.
The agent has a circular body and seven distance sensors in order to per-
ceive objects, which are dropped from the top of the environment. The
agent is taught to catch circular-shaped objects and to avoid diamond-
shaped ones (see Figure 3.2).
The sensorimotor activity of the agent is controlled by an artificial
14-neuron continuous-time recurrent neural network (seven sensory neu-
rons, two motor neurons, five interneurons). The behavior of the whole
system is described by 16 equations:
si si Ii x, y; i 1, , 7
7 12
i si si
j 1

w ji g s j w s i 8,,12
j 8
ji j j
12
i si si w s i 13,14

j 8
ji j
x 5 s13 13 s14 14
y 3
where x is the horizontal position of the object; y is the vertical position

of the object; si is the state of each neuron (14 neurons in total); w, τ, θ,
and σ are neural parameters; and I is an external input vector (Beer, 2003,
p. 214). The system can be decomposed into two coupled subsystems:
Figure 3.2 M
odel of the perceptual categorization agent. (I) Simulated agent in
the environment. (II) The agent's neural network brain including seven
sensory neurons, five interneurons, and two motor neurons. Adapted
from “The Dynamics of Active Categorical Perception in an Evolved
Model Agent,” by R. D. Beer, 2003, Adaptive Behavior, 11(4), p. 213.
Copyright 2003 International Society for Adaptive Behavior.
first, the agent’s “neural network brain” and body (14 equations) and,
second, the environment (Equation 15 and 16). The coupling is math-
ematically expressed by putting variables S13 and S14 (responsible for two
motor neurons) into the second to last equation and variables x and y
(the object’s horizontal and vertical position relative to the agent) into
Equations 1 through 7.
Beer’s model shows two important things. First, dynamical systems can
be decomposed. As Beer (2003) writes,
we have many different choices as to how to decompose the dynam-

ics of the agent–environment system and how to visualize it. Clearly,
whatever decomposition we choose should emphasize the factors
underlying movement, since the agent’s decision is expressed in its
motion over time and it is this behavior that we want to explain.
Thus, we will decompose the agent–environment dynamics into: (1)
the effect that the relative positions of the object and the agent have
on the agent’s motion; (2) the effect that the agent’s motion has on
the relative positions of the object and the agent.
(p. 228)
Second, the dynamical approach has the potential to model complex cog-
nitive capacities but, as I argue, that is so because this model goes beyond
its own limitations and includes other perspectives to some extent, namely
the mechanistic one.
There is ongoing debate concerning the nature of dynamic descriptions
and their explanatory power. Some argue that dynamical explanations
are a subtype of D-N explanations (e.g., Raja et al., 2017; Walmsley,
2008); that is, they seek to formulate lawlike generalizations in the lan-
guage of differential calculus which explains the target system’s particu-
lar behavior. For example, the HKB model (Haken et al., 1985) consists
of a coordination law expressed in a single equation. The model allows
one to predict the system’s behavior and supports counterfactuals; that
is, it allows one to say what would happen with the system if certain
parameters or variables changed. Moreover, developments of the HKB
model show that it can be generalized to other oscillatory activities
(Kelso, 1995; Oullier et al., 2008; Port, 2003); thus, it can be conceived
of as a general law for coupled oscillatory motion. There are, however, at
least two serious problems with such a nomological reading of dynamical
models. First, as Joel Walmsley (2008) argues, if dynamical explanations
are D-N explanations, then they are deductive; that is, the explanandum
phenomenon or prediction is deduced from the lawlike formula and data
about the current state of the system. But if that is the case, then the
explanatory model becomes reductive—all the facts about the studied
phenomenon can be deduced. Second, the dynamical approach and D-N
explanations are committed to so-called predictivism; that is, they share
the argument that prediction equals explanation. The flagpole example,
discussed in the context of the D-N model of explanation, shows that
that is not necessarily the case—that we can have relatively good predic-
tive power without having a genuine explanation.
There is yet another issue with dynamical explanations in cognitive sci-
ence. The dynamical approach is not interested in the system’s structural
composition but in its overall behavior and its evolution in time. Thus,
for some researchers, dynamical analysis merely delivers a description
of the explanandum phenomenon, the explanatory power of which is
weak, even if it proves to be a reliable basis for accurate predictions.
To establish explanatory power, the dynamical model has to be supple-
mented with a model of the underlying mechanism, describing its parts,
internal organization, and causal interactions (e.g., Bechtel, 1998; Kaplan
& Craver, 2011). For example, according to Craver (2006), the Hodgkin
and Huxley (HH) model of action potential (Hodgkin & Huxley, 1939)
is not explanatory but merely describes the behavior of neurons. The
HH model can, however, still be heuristically useful and help in find-
ing details about actual mechanisms responsible for neuron behavior, for
example, ion channels. Generally speaking, dynamical explanations can
be an important source of information guiding the study of the target
system and the building of a model of its parts and internal organization.
The constraints delivered by the dynamical approach concern the global
behavior of a target system, but, importantly, they can also character-
ize the behavior of mechanisms’ components (Zednik, 2011). A good
example of such a dynamical-mechanistic model is Beer’s model noted
earlier. In it, equations characterize the dynamics of the system’s compo-
nents, which are interconnected neurons responsible for different func-
tions organized in three layers. Separate equations describe the changing
position of the object. The overall dynamics generated by the whole set
of equations is therefore a derivative of component behavior. Dynamical
and mechanistic explanations can, therefore, be treated as complemen-
tary rather than exclusive explanatory strategies.
3.3.5 Mechanistic Explanations
Mechanistic explanations are in many ways as influential as those pro-
duced by the D-N model. Looking into the not-so-distant past, we may
note Rene Descartes or Julien Offray de La Mettrie as precursors of
mechanism (for a historical overview, see, e.g., Miłkowski, 2017; Roux,
2017). Descartes initiated a line of thinking about living organisms in
terms of mechanisms. A more recent predecessor of the contemporary
(neo)mechanistic model of explanation is the causal-mechanical model
of explanation developed by Salmon (1989), although his main area of
interest was physical science. According to Salmon, to explain a phe-
nomenon is not to show that it could be inferred from a law of nature or
lawlike generalizations, but to show how the phenomenon is produced
by causal processes. The (neo)mechanistic model of explanation1 (e.g.,
Bechtel, 2008; Craver, 2007; Glennan, 2017) that I discuss in this section
is a development of causal-mechanical explanation, which may also be
referred to as a constitutive or componential explanation. This type of
mechanistic explanation focuses on describing the mechanism, that is, the
components, activities, and organization that underlie or constitute the
phenomenon to be explained.
Whereas D-N explanations answer the question of “why,” mechanistic
ones are dedicated to answering the question of “how”; for example,
how is the target system’s behavior being produced, or how is a specific
function being realized in the target system? The mechanistic approach
emphasizes the particularity of the world rather than its generality, unlike
the D-N model. It focuses on real entities, particular mechanisms diverse
in form and organization, situated in an external context. Mechanistic
explanations are, contrary to D-N explanations, asymmetrical; that is,
the mechanism explains the target phenomenon and not vice versa. This
does not mean, however, that there is only one direction of determina-
tion, namely bottom-up determination from the parts of the mechanism
to the produced phenomenon. As I will show, there are cases of top-down
relations in which the mechanism’s behavior at a higher level can affect
its parts at a lower level.
In the following sections, I stress a few important features of mechanis-
tic explanations, namely that they are causal, contextual, multilevel, and
nonreductionistic in the strong sense. Mechanistic explanations are causal
because they focus on the causal mechanisms underlying an explanandum
phenomenon. They are contextual, as they situate a given mechanism in
the context of their environment, emphasizing that the context is impor-
tant for understanding the mechanism’s functioning. They are multilevel
because they address different levels of a mechanism’s organization and
their interlevel relations. Addressing different levels often requires inte-
gration of multiple research fields. Finally, mechanistic explanations are,
as I argue, nonreductionist in a strong sense because they acknowledge
that, in some cases, the behavior of the whole surpasses the behavior
of its components and that to describe a mechanism on multiple levels
we need to integrate different research fields without reducing one to
another.
3.3.5.1 Mechanisms
There is a common misunderstanding with respect to mechanisms that
needs to be addressed at the outset: a mechanism is not a simple aggre-
gate of parts. An aggregate is a collection of parts in which the order of
the parts is irrelevant to the whole. Think of a pile of stones as an exam-
ple. In an aggregate, we can interchange parts with one another without
any effect on the behavior of the whole. What matters for mechanisms is
just that: the organization of their parts and their activities. Consider the
following definition:
A mechanism is a structure performing a function in virtue of its

component parts, component operations and their organization. The
orchestrated functioning of the mechanism is responsible for one or
more phenomena.
(Bechtel & Abrahamsen, 2005, p. 423)
First, notice that a mechanism typically performs a function, that is, it

plays a role in a larger mechanism or system. This situates the mechanism
in a context, which is as just as important to the explanation as the mech-
anism’s component parts are. Second, as I argued at the beginning of this
chapter, there are different types of explanandum phenomena—some of
them are regular, but others are not. In the life sciences, where mechanistic
explanations are often used, the explanandum phenomenon is typically
some kind of biological function of an organism, or an organism’s behav-
ior. In cognitive science, a higher level phenomenon is usually a cognitive
agent’s behavior or some abstract cognitive function. The notion of func-
tion can vary depending on the target system. In the case of artificial sys-
tems intentionally designed to work in a certain way, Cummins’s concept
of function does the job. In the case of natural systems, it is necessary to
include evolutionary history; thus, Millikan’s notion of proper function
is preferred (Millikan, 2002). For example, the heart’s proper function
is pumping blood in the cardiovascular system. However, mechanistic
explanations are not limited to those cases. They may also plausibly be
applied to functions that are neither designed nor proper but that are, for
instance, side effects or disfunctions. For example, a heart pumping blood
emits sounds, which is not the heart’s proper function; nevertheless, that
fact can be explained mechanistically. Or consider the mechanisms of
various diseases, which are not separate mechanisms evolved to produce
the disease but are rather related to disfunctions and other factors that,
when combined, result in the symptoms of the disease.
Third, a mechanism’s component parts can be understood only in rela-
tions to each other, that is, their organization and their effects. To charac-
terize the relation between a mechanism and a phenomenon, proponents
of mechanism usually use terms such as “underlies,” “produces,” “is
responsible for,” or “exhibits,” but, as we will see below, it is not easy to
clarify the nature of interlevel relations in mechanisms. Roughly speak-
ing, the nature of this relation between a phenomenon and a mechanism
may take various forms. Stuart Glennan (2017, p. 109), for instance,
distinguishes between nonconstitutive and constitutive phenomena. The
former are produced by a mechanism, precisely by its parts and their
activities. The latter can only be explained in terms of the activity of the
whole mechanism. To put it differently, the latter phenomena are not pro-
duced by the mechanism, but the mechanism underlies them. In general,
the relation between a mechanism and a phenomenon can be depicted as
in Figure 3.3.
Looking at the lower level, we can see why mechanistic explanation is
sometimes called “explanation by decomposition” (e.g., Craver, 2007),
and decomposition is certainly one of the key heuristic strategies applied
in this framework. To understand a mechanism’s internal organization, it
has to be decomposed into parts and the activities they perform. The for-
mer type of decomposition is called structural decomposition and the lat-
ter functional decomposition. In short, structural decomposition divides
the system into specific task-related parts. Functional decomposition
“assumes that one activity of a whole system is the product of a set of
subordinate functions performed in the system” (Bechtel & Richardson,
Figure 3.3 M
echanistic explanation. Adapted from Explaining the Brain:
Mechanisms and the Mosaic Unity of Neuroscience (p. 7), by C. F.
Craver, 2007, Clarendon Press. Copyright 2007 by Carl F. Craver.
2010, p. 23). At first glance, mechanistic decompositions may seem simi-

lar to Fodor’s (1983) idea of modularity, that is, mapping subfunctions
onto modules. In fact, these approaches are importantly different. For
example, one of the key features of Fodorian modules is their being infor-
mationally encapsulated, whereas mechanistic parts are not encapsulated
in that sense (Bechtel, 2009a). Although the parts of a mechanism are
distinguished by the operations they perform, they can affect each other
in many ways.
Introducing decomposition as an explanatory strategy, proponents of
mechanism (e.g., Bechtel & Richardson, 2010) usually follow Simon’s
(1962) and William Wimsatt’s (1986) division into several levels of
decomposability: a system can be decomposable, nearly decomposable,
minimally decomposable, or non-decomposable. Decomposable systems
are simple aggregative systems in which the sum of a part’s activities
equals the activity of the whole. Decomposable systems have a clear hier-
archical organization, and the activity of the parts can be explained inde-
pendently of the whole. Also, the organization of the parts is irrelevant
in principle. Nearly decomposable systems are more complex and dif-
ficult to decompose since the components of such systems interact with
each other, and higher levels of the organization determine lower levels
to some degree. Thus, in nearly decomposable systems we have not only
bottom-up but also top-down relations. Minimally decomposable sys-
tems are systems in which components are not only organized and inter-
act with each other but in which top-down relations are more relevant for
the whole than bottom-up relations. Finally, there are non-decomposable
systems in which top-down determinations make decomposition into
parts impossible. Such systems have no hierarchy whatsoever. Mechanists
usually argue (e.g., Bechtel & Richardson, 2010) that cognitive systems
are nearly decomposable and thus amenable to mechanistic explanation.
The second explanatory strategy crucial to mechanistic thinking is
localization. “Localization is the identification of the different activities
proposed in the task decomposition with the behavior or capacities of
specific components” (Bechtel & Richardson, 2010, p. 24). There are
two types of localization: simple or direct localization and complex or
distributed localization. Simple localization identifies some specialized
activity with a single part of a system. Franz Joseph Gall’s phrenological
localization of the mind’s faculties in the brain is a historical example of
simple localization. Complex localization considers the target capacity a
collection of component operations that are localized in different parts
of the system. Simple localization is sometimes considered a default first
strategy applied at the beginning of a research process and is likely to be
falsified and modified into complex localization. Studies on the visual
perception system serve as an example of how the localization of cer-
tain cognitive capacities in the brain changed over an extended period of
research (see, e.g., Bechtel, 2008). Already in the second half of the 19th
century, Bartolomeo Panizza, who studied stroke patients, established the
simple localization of visual processing in the occipital lobe. By the end
of the 20th century, scientists identified 32 brain regions responsible for
specialized functions related to processing visual information distributed
across the occipital, temporal, parietal, and frontal cortex (Felleman &
van Essen, 1991). It is therefore important to remember that decomposi-
tion and localization are just heuristic strategies; they are epistemological
categories, not ontological ones. It might be the case, and often is, as stud-
ies of cognitive functions and their localization in the brain have shown,
that an initial assumption concerning decomposability and simple local-
ization needs to be modified into near decomposability and complex
localization. It does not follow, however, that the mechanistic approach
is incorrect; instead, it shows that mechanistic models can be refined as
well as supplemented and constrained by other fields of research.
As Bechtel (2009b) puts it, “mechanistic explanation in psychology
requires not just looking down (decomposing the mechanism), but also
looking around (recomposing the mechanism) and looking up (situating
the mechanism)” (p. 544). Decomposition and localization are necessary,
but they are not sufficient to deliver a full mechanistic explanation of
a cognitive phenomenon. It is equally important to study the organiza-
tion of the identified parts and the relations between the operations they
perform, what Bechtel calls recomposition. For instance, establishing the
direction of the flow of information, the sequence of operations, and the
character of the connections between the parts is of key importance. For
example, the connections between the regions responsible for visual pro-
cessing may vary, being feedforward, feedback, or collateral in nature.
For the purposes of this work, the third way of looking, namely, “look-
ing up” and situating the mechanism, is the most important. “A mecha-
nism is always contextually situated” (Bechtel, 2008, p. 17) and situating
the studied mechanism in the environment is critical for understanding
how it works. The environment is usually considered a source of stimuli
that the whole mechanism receives but that can also affect selected parts
of the mechanism. However, the environment is not just a collection of
stimuli; it is also full of regularities, in reaction to which a complex mech-
anism such as an organism can develop responses. Moreover, the envi-
ronment is also a collection of objects which are meaningful for organism
behavior. To acknowledge the complexity of the environment, on the one
hand, as well as the activity and motility of the perceptual agent, on the
other, it is best to think about the relation between them in terms of sen-
sorimotor dynamical patterns. In this respect, Bechtel refers (2009a, pp.
557–558) to the ecological psychology of J. J. Gibson, which describes
perception as an active, embodied, and situated process. Mechanistic and
ecological approaches to cognition do not have to be opposed to one
another; rather, they can be seen as complementary (e.g., Abramova et al.,
2017; Golonka & Wilson, 2019).
Although Bechtel (2009a) acknowledges the importance of environ-
ment in explaining a mechanism’s behavior, at the same time, in the case of
cognitive mechanisms, he holds that the mind/brain is the “locus of con-
trol.” Accordingly, an organism considered from a mechanistic perspec-
tive is situated in the environment, but at the same time, it is independent
to some extent in controlling its behavior. The brain is the localization
of the controlling functions. For Bechtel, localizing control in the brain
is the right strategy for explaining mental phenomena. But in the case of
social phenomena, it may be abandoned in favor of distributed processes
in which individual agents participate. In fact, there is quite a long tradi-
tion of explaining social phenomena in terms of social mechanisms (for
an overview, see, e.g., Ylikoski, 2017). Social mechanisms are also causal
mechanisms but composed of social agents, individual, collective, and
institutional, in which “the agent is so intertwined with entities outside
itself that the responsible system includes one or more cognitive agents
and their environment” (Bechtel, 2009a, p. 156).
Social mechanisms raise the question of the reliability of the locus of
control. Indeed, it is possible to reject the locus of control constraint and
develop a mechanistic position according to the extended mind hypoth-
esis (Clark & Chalmers, 1998) or distributed cognition (Hutchins, 1995).
Craver, for instance, claims that “many cognitive mechanisms draw upon
resources outside of the brain and outside of the body” (2007, p. 141).
For example, using something as a form of external memory can render
that thing an external resource critical for some cognitive activities. In
this sense, when a cognitive mechanism offloads some of its cognitive
work onto the environment, we may consider it an extended mechanism.
Carlos Zednik (2011) goes further and argues that not only mechanisms
may use external resources but that components of complex mechanisms
can also be distributed across the brain, body, and environment. An
example of such an extended mechanism has already been mentioned:
Beer’s (2003) perceptual agent model. Beer’s model is both dynamical
and mechanistic and describes, using dynamical systems theory calculus,
a moving perceptual agent whose behavior is tightly coupled with the
environment: the agent avoids one class of objects and moves toward
another class. From the point of view of the dynamical model, the system
as a whole may be seen as including objects in the environment, and
our explanatory task is to describe the dynamical relations between its
components. Generally speaking, according to Marcin Miłkowski et al.
(2018), the 4E approaches to cognition, that is, embodied, embedded,
extended, and enacted, as well as the distributed cognition framework,
can be understood as research traditions providing heuristics for building
mechanistic explanations. Such guiding heuristics help discover mecha-
nisms responsible for the studied phenomenon and describe their orga-
nization. For Miłkowski et al., mechanism also offers a framework for
integrating these approaches and using their advantages to build genuine
explanations they cannot offer when taken in isolation.
At this point, it is worth noting that when we say that a part of a mech-
anism is in the external environment, it does not mean that the external
environment is a spatial part of the mechanism. Being a part of a mecha-
nism is understood here as being relevant to its behavior. So instead of
understanding spatial containment as a relation between the mechanism
and its parts, mechanists turn to causal or constitutive relevance; that is,
something is part of a mechanism if it makes a difference for its function-
ing (Craver, 2007). In this respect, mechanists usually refer to the appli-
cation of the mutual manipulability (MM) account, descendant from the
interventionist conception of causality (Woodward, 2003):
According to that account, a part is a component in a mechanism if

one can change the behavior of the mechanism as a whole by inter-
vening to change the component and one can change the behavior of
the component by intervening to change the behavior of the mecha-
nism as a whole.
(Craver, 2007, p. 141)
The MM approach allows one to establish the “constitutive relevance” of

parts, that is, to determine which parts are relevant to the constitution of
the mechanism. The relevance here is limited by the explanandum, which
means that not every physical part of the mechanism will be included
in the explanatory model, including only those that are constitutively
relevant.2 In cognitive neuroscience, the relevance of a mechanism’s parts
can be established through lesion studies and stimulation experiments.
With lesions, double dissociations are especially informative because they
provide strong evidence for two activities being realized separately in the
brain. The stimulation strategy relies on applying chemical or electro-
magnetic stimuli to the brain and observing how they change behavior.
These experiments allow one to establish which parts of the system are
parts of the studied mechanism and what function the mechanism’s com-
ponents play.
To round out this short characterization of mechanisms, allow me to
clarify two common misconceptions. Aside from the oversimplification
of thinking about mechanisms in terms of aggregates, we also tend to
think about them as deterministic and passive. That tendency suggests
that a mechanism, if it is not dysfunctional or broken, regularly generates
the same output behavior, which, moreover, can be described with ceteris
paribus generalizations. But human behavior is not so regular; we can
predict it only with a degree of probability. In psychology, there are no
universal laws analogous to the universal laws of physics. Psychological
generalizations are statistical. And in the life sciences, organisms are not
considered deterministic automata. This is where we encounter stochastic
or nondeterministic mechanisms. Contrary to deterministic mechanisms,
nondeterministic mechanisms behave with a degree of probability; thus,
they can be characterized only by statistical generalizations (Glennan,
2017, pp. 132–135). The probabilistic characterization of such mecha-
nisms may apply to different levels of composition. It may concern
the produced phenomenon, which is stochastic in nature, or it may be
applied on a lower level and related to the activity of the particular parts
of the mechanism and their outcomes. There are many examples of non-
deterministic mechanisms, including biological ones, such as the mecha-
nisms involved in reproduction, which although functioning properly do
not guarantee offspring, or the social mechanisms underlying collective
behavior. This plausibly applies to many cognitive mechanisms. Examples
of stochastic models in cognitive neuroscience include Bayesian networks,
which implement probability on the level of neural computation.
The second misconception concerns a given mechanism’s passivity and
comes from thinking about mechanisms in terms of simple machines we
use every day such as coffee makers or washing machines. These simple
machines receive a specific input and produce a specific output; thus,
we think about them as passive or purely responsive. As Bechtel (2008,
pp. 201–238) argues, this tendency is also present in the science of the
mind and neuroscience of the brain. For example, for a long time, mem-
ory was thought to be a passive process of retrieving information rather
than a constructive activity. Recent research on false memories suggests
that there are processes that can modify stored content (e.g., Schacter,
2012). Thinking about cognitive mechanisms as responsive is also related
to experimental practice. It is much easier to design experiments and to
control key factors when we assume that a given mechanism is inactive
and simply awaits an input, not modifying itself and its responses. But, as
Bechtel (2008) argues, this is not how the mind/brain naturally functions,
and “if mechanistic explanation is to prove adequate for understanding
the mind and brain, mental mechanisms must be construed as active”
(p. 206). To do so, Bechtel introduces operations involving positive and
negative feedback, which are crucial for the capacity of biological sys-
tems to maintain themselves. The idea of negative feedback can be under-
stood through the example of the already introduced Watt’s governor.
The constant speed of the steam engine’s flywheel is controlled by the
device, which opens the throttle when the speed is too low and closes it
when it is too high. In biological systems, negative feedback is the basis
of homeostasis. Positive feedback is a process the results of which amplify
the same process in return. For example, the influx of sodium ions into a
neuron causes the opening of more ion channels and thus reinforces the
influx of sodium until it releases some action potential. Positive feedback
is crucial in biological self-construction processes. Acknowledging these
principles allows one to model simple mechanisms that are intrinsically
active and autonomous, that is, able to regulate their response in order to
maintain themselves (e.g., Ruiz-Mirazo & Moreno, 2004). Furthermore,
such basic autonomous systems can be developed into more complex
autonomous adaptive agents (Bechtel, 2008, pp. 223–225), that is, sys-
tems that transform their environment and modify their behavior in
order to promote their existence.
3.3.5.2 Models of Mechanisms
Following Bechtel (2008), who states that “mechanisms do not explain
themselves” (p. 18), I adopt the epistemic view of mechanistic explana-
tion.3 Accordingly, a mechanistic explanation is a product of scientists’
work and consists of scientific representations of an actual mechanism
and its behavior. These representations may take the form of a simplified
sketch or a complete model, a detailed scheme describing the mecha-
nism’s parts, activities, and their organization. The explanatory process
goes from a sketch, which still includes black boxes and empty spaces,
toward a complete scheme, which covers all the details relevant to the
explanation.
When talking about mechanistic models, it is important to mention
the distinction between how-possibly and how-actually models (Craver,
2007, pp. 112–114; Glennan, 2017, pp. 68–73). How-possibly mod-
els are built on the basis of hypotheses about mechanisms that might
produce the explanandum phenomenon. Such a model does not refer
to an actual mechanism but expresses one of the possible realizations
of the phenomenon. On the contrary, a how-actually model describes
an actual mechanism responsible for the target phenomenon. Between
these two, we can situate a great many how-plausibly or how-roughly
models, which are more or less accurate representations of the underly-
ing mechanism. It is important to remember that there can be multiple
realizations of a function; thus, there can be several how-actually models
of mechanisms responsible for the same function but, for instance, in dif-
ferent organisms. Note also that on the epistemic account of mechanistic
models even a complete model does not include, and should not have to
include, all of the mechanism’s parts and their activities. It consists only
of those components that are relevant to the explanandum phenomenon.
To put it differently, models are always a matter of abstraction and ide-
alization. The former procedure omits some aspects or parts of the target
system that are irrelevant to the explanation. The latter adds features or
changes parameters in order to provide a better understanding of the
target mechanism.
An equally important dimension of mechanistic explanation is the
notion of a phenomenal model:4
The distinction between a phenomenal and a mechanistic model is

the distinction between a model that describes the phenomenon and
one that explains it. A phenomenal model characterizes what the
phenomenon is—often no mean achievement—while the mechanistic
model shows how it comes about.
(Glennan, 2017, p. 67)
A phenomenal model not only defines the explanandum but also indi-
cates its key properties, which, in turn, can serve as heuristics in search-
ing for mechanisms. Indeed, a complete mechanistic explanation consists
of two complementary parts: a detailed model of the mechanism and a
phenomenal model of the explanandum.
It is argued that, in contrast to phenomenal models, mechanistic mod-
els are explanatory because they support interventionist counterfactu-
als; that is, they allow one to answer “what-if–things-had-been-different”
questions (Woodward, 2017). In short, a mechanistic explanation tells us
what would change in the target phenomenon if we modified the behavior
of some part of the mechanism, changed the environmental conditions,
or level of stimulation. The explanatory power of mechanistic models
is also tied to our ability to control and manipulate the system (Craver,
2006). If direct manipulation is impossible, then simulation plays an
analogous role. Some mechanists (e.g., Craver) argue that phenomenal
models do not satisfy these conditions and thus that they are not explan-
atory. Accordingly, one can argue that the previously mentioned HKB
dynamical model of oscillatory movement (Haken et al., 1985) is a purely
descriptive phenomenal model. Although it can serve as a predictive tool,
it is not explanatory because it does not describe the system’s compo-
nents to the degree that would allow one to control its behavior (it isn’t
necessarily the case that it fails to support counterfactuals; as I argued,
many dynamical models do support them). Others, however, argue that
the distinction between the descriptive and explanatory is rather concep-
tual and that these aspects of explanations should not be perceived as
opposites.
An example from neuroscience of the previously mentioned HH
model of action potential (Hodgkin & Huxley, 1939) shows how models
which are initially thought to be merely phenomenal are important in
the explanatory process. The original HH model is a dynamical model
and consists of equations describing the dynamics of neuronal behavior,
which, as it was hypothesized, was produced by specific components and
activities of neurons, such as the flow of ions across cell membranes.
According to Glennan (2017), the HH model cannot be categorized as
purely phenomenal or mechanistic but rather as a how-roughly model of
a mechanism that consists of two dimensions: a description of behavior
and actual or possible components producing the behavior. This early
model informed further research and the discovery of mechanistic com-
ponents such as ion channels. The discovery, in turn, supplemented the
model with further relevant details.
Allow me to also note that modeling neuronal behavior points to the
importance of thinking about neurons as active mechanisms rather than
something passive and simply responsive. Neurons do not remain silent
when not stimulated but exhibit basal activity and firing rate, which can
be modulated up or down. Thanks to positive feedback, the subsequent
behavior of a neuron depends on its previous behavior; for example, it
fires with a fixed rate until it receives inhibitory stimuli. Indeed, individ-
ual neurons as well as their assemblies can be modeled as coupled oscilla-
tors (Bechtel, 2008, pp. 229–233). For that purpose, again, the dynamical
approach is of key importance.
3.3.5.3 Levels of Mechanisms
One of the key features of mechanistic explanations is that they are mul-
tilevel. As I mentioned earlier, one of the results of applying heuristic
strategies such as decomposition to the target mechanism is division into
components and operations that can be grouped at different levels of
organization. For example, a mechanistic explanation of the phenom-
enon of spatial memory in mice (their learning and navigating in an envi-
ronment) consists of four levels: mouse behavior (navigating in a maze), a
neural structure encoding a spatial map in the hippocampus, a long-term
potentiation mechanism in neurons, and neurotransmitters along with
their receptors (activation of the NMDA receptor; Craver, 2007, p. 166).
The notion of level is notoriously ambiguous (for an overview of
different conceptualizations of it in science, see Craver, 2007, ch. 5).
First, we can think of levels of nature, that is, strata and distinctions
that actually exist in the world, for instance, as biological hierarchies
captured in taxonomies or as related to science as such. For example,
Paul Oppenheim and Hilary Putnam (1958) introduced six levels of
nature: elementary particles, atoms, molecules, cells, organisms, and
societies. Each level corresponds to a scientific discipline with its own
theory, laws, and explanations. These levels are, according to Oppenheim
and Putnam, not independent; that is, higher levels can be nomologically
reduced, via bridge laws, to lower levels. This picture, however, does not
fit mechanistic explanations. Consider the example of spatial memory
presented earlier. The levels of explanation do not constitute a nomo-
logical theory interrelated by bridge laws; rather, these levels represent
different research fields that are related because they refer to the same
explanandum phenomenon. Each of these fields delivers constraints and
thus contributes to the complete explanation. So how should we think
about levels in mechanisms?
The are several ways we can distinguish levels in mechanisms. First, we
can distinguish different spatial levels: mechanisms and parts of mecha-
nisms come in different sizes. Usually, bigger mechanisms are composed
of smaller parts, which, in turn, can have yet smaller parts. Certainly,
spatial levels play an important role in science. In neuroscience we can
talk about larger brain regions and areas, smaller cell assemblies, and
neural columns, ultimately reaching the level of individual neurons and
even smaller parts, such as neurotransmitters and ions. In mechanisms
entities of different sizes can causally interact. Furthermore, mechanisms
not only have parts of different sizes, but they also work in different
time scales. Temporal levels are critical for studying the brain, behavior,
and our subjective experience. Events in the brain usually happen on the
scale of milliseconds, whereas perceptual apprehensions and actions typi-
cally happen on the scale of seconds. Finally, there are causal levels. For
example, we can divide the processing of visual information into steps
and localize specific operations in cortical regions such as V1, V2, V3,
and so on. Accordingly, each region is dedicated to processing specific
information and participates in a causal chain of processes producing
visual experience. All these types of levels can contribute to the study
of the internal organization of the target mechanism and the relations
between its parts, but none of them gives us an answer to the more gen-
eral question concerning the nature of levels of mechanisms.
Craver (2007, pp. 188–195) proposes to think about levels of mecha-
nisms in terms of composition. As he argues, “levels of mechanisms are
levels of composition, but the composition relation is not, at base, spa-
tial or material. In levels of mechanisms, the relata are behaving mecha-
nisms at higher levels and their components at lower levels” (Carver,
2007, pp. 188–189). To put it differently, the higher level is the behavior
of the mechanism constituted or produced by the lower level, on which
we find organized parts constituting the mechanism (see Figure 3.4). In
an analogous manner, we can decompose this lower level, taking a part
Figure 3.4 L
evels of mechanisms. Adapted from Explaining the Brain: Mechanisms
and the Mosaic Unity of Neuroscience (p. 194), by C. F. Craver, 2007,
Clarendon Press. Copyright 2007 by Carl F. Craver.
from that level and showing how its behavior is generated by parts and
operations at yet a lower level. A consequence of this understanding of
levels is that mechanistic explanations span levels and must have at least
two: a higher level corresponding to the phenomenon and a lower level
corresponding to the underlying mechanism.
Mechanists agree that relations between parts at a given level, intra-
level relations, are causal. How should we understand interlevel rela-
tions? First, although mechanisms and their parts have sizes, the levels
of mechanisms are not spatial. Thus, the relation of containment is ruled
out. The levels of mechanisms are levels of composition, which means
that the parts of lower levels are the components of higher levels. But can
we say that there is any causal interaction between these levels? Craver
(2007, p. 195) rejects the notion of interlevel causation, as it would lead
to the “strained” idea that the behavior of the whole mechanism can
interact with its parts. In a similar manner, Bechtel (2008) writes that
“certainly there are relations between levels within a mechanism. This is
the reason researchers investigating mechanisms need to bridge between
levels. But there are good reasons not to characterize the relation between
a mechanism and its parts causally” (p. 153). For instance, that would be
problematic in the case of a mechanism involving the temporal succes-
sion of causes and effects, as it is thought that causes precede their effects.
But that is not necessarily the case in mechanisms generating behavior;
that is, the overall behavior is collateral to the parts’ activities at the
lower level. Interlevel relations are constitutive; to be precise, they are a
hybrid of constitution and causation which Bechtel (2008) calls “mecha-
nistically mediated effects” (p. 153). One might ask whether “mechanisti-
cally mediated effect” is not just another name for emergence; in many
ways, it is. However, Bechtel is careful to avoid using “emergence,” as
he is aware that for some philosophers it conveys a mysterious relation.
But there is nothing mysterious about the emergent behavior of mecha-
nisms unless emergence is understood as referring to the behavior of the
whole that goes beyond the activity of its parts. In accordance with such a
notion of emergence, the higher level phenomena have a certain degree of
independence: they require a different mode of investigation than study-
ing entities and processes at lower levels (Bechtel, 2008). It does not fol-
low, however, that such emergent behavior cannot be explained in terms
of mechanistic composition. More recently, Bechtel (2019) has used the
notion of constraints to characterize interlevel relations. Accordingly, the
higher levels of a mechanism’s organization constrain the behavior of its
components on lower levels. To put it differently, higher level behavior
reduces the degrees of freedom of lower level components, consequently
producing the target phenomenon. Bechtel calls the higher level control
mechanism, which is mainly engaged in alteration of lower level compo-
nents’ activities, the production mechanism. This is yet another proposal
for how to address interlevel relations without causality.
3.3.5.4 Limits of Mechanistic Explanation

The mechanistic model of explanation does have some limitations. It has
been argued, for example, that it fails to deliver an explanatory demar-
cation (it rejects other explanations, such as D-N or dynamical ones, as
nonexplanatory) and that the mechanistic model is committed to a rep-
resentational ideal of completeness, that is, that it acknowledges that a
full explanation consists of a complete and perfect model of a mechanism
(for a longer discussion, see, e.g., Halina, 2017). I think these arguments
are misguided. First, as I have shown, mechanism has a clear idea of
what is explanatory and what is not. Furthermore, as we will see in the
next section, mechanism is open to integration with other explanatory
frameworks, including the dynamical framework. Second, mechanistic
explanations do address compositional and organizational details of the
studied mechanism; otherwise, they would be nonexplanatory phenom-
enal models. But the neo-mechanistic model of explanation is a compo-
nential or constitutive explanation, which means that it does not describe
all of a mechanism’s components, their organization, and activities, but
only those that are relevant to the target phenomenon. Furthermore,
mechanistic models, like other scientific models, are a product obtained
through operations of idealization and abstraction. What follows is that
a model with more details does not necessarily have more explanatory
power and, therefore, that the ideal of completeness does not apply to
mechanism (Craver & Kaplan, 2020).
Another common argument against mechanistic explanations of cog-
nitive functions appeals to the multiple realizability thesis (MRT). If
a cognitive function can be physically realized in multiple ways, then
describing how a single mechanism realizes it does not explain the func-
tion, it only represents one amongst its many possible realizations. If that
is the case, then the functional approach is explanatory, not the mecha-
nistic one. There are three basic counterpoints to be made here. First,
mechanists reply that MRT does not imply that any kind of hardware
could realize any function; some hardware realizes some functions better
than others (Bechtel, 2008, p. 31). Second, MRT comes from thinking
about the mind in analogy to computers. But computers were intention-
ally designed to be universal in respect to the functions they can perform.
On the contrary, biological brains evolved to realize very specific tasks
relevant to a specific organism and its survival (e.g., Cosmides & Tooby,
1995). Finally, mechanists do not claim that every realization of the tar-
get function is explained by the same mechanism (Craver, 2007, p. 160).
Mechanists only claim that each specific realization of the target function
is explained by some mechanism. In fact, it is often the case, for instance,
in biology, that the same function is realized by different mechanisms. But
still, it is always a particular mechanism together with its evolutionary
background that explains the studied biological function. Multiple realiz-
ability is thus not a challenge to mechanistic explanations. Furthermore,
as we will see later, functional and mechanistic explanations should not
be perceived as conflicting but as complementary.
But if models of mechanisms are always limited to a specific realiza-
tion of the target phenomenon, then they are not generalizable and thus
explanatorily limited. A good explanation should be universal and apply
to different systems. One answer to this issue could be placing the model
in a wider evolutionary context. Accordingly, a mechanistic model is gen-
eralizable if it is supported by evolutionary theory, showing that this type
of mechanism appears in different organisms at different times. In this
sense, a mechanism is a regular pattern in the domain of living organisms.
A broader answer is given by Bechtel and Abrahamsen (2005), who argue
that mechanistic models may be generalizable to some extent. In their
view, generalizations of mechanistic explanations differ from generaliza-
tions in the D-N framework, in which generalization is achieved by the
application of universal laws. Mechanistic generalizations are made by
establishing similarity between different models and applying one model
in an investigation of another mechanism. For example, investigations of
the squid axon were particularly important for understanding the gen-
eral phenomenon of neural transmission occurring in different types of
neurons. Of course, the neurons in squid are not exactly the same as the
neurons in the human brain; nevertheless, they are similar enough that
studying the former shed new light on the latter. Such mechanistic gen-
eralizations are limited, but these limitations may point us to a further
important insight, namely, the significance of variations in studied mech-
anisms across different species. For instance, differences in the same type
of mechanisms may inform us about the studied organism evolutionary
history or indicate exaptation of the mechanism’s function.
Besides the general limitations of the mechanistic model of explana-
tion, we can consider its local limitations in the domain of cognitive
science. As I argued earlier, mechanists assume that cognitive systems
are nearly decomposable systems, and thus, they can be explained in a
mechanistic manner. Their critics argue, however, that living cognitive
systems are either minimally decomposable or non-decomposable. For
example, Evan Thompson (2007, pp. 417–447) argues that the brain is
a dynamical network of processes with strong top-down determinations
and is thus minimally decomposable if not non-decomposable altogether.
However, he also admits that sometimes it is useful for explanatory
purposes to characterize the brain as nearly decomposable. Indeed, as I
noted, decomposability is just a heuristic strategy, a sort of rule of thumb
guiding research, and thus may fail. It seems clear that decomposability
should be understood as an epistemic category, not an ontological one.
Both decomposition and localization are fallible heuristic strategies, but
it does not follow that they cannot be applied to cognitive phenomena.
In fact, research on memory and vision shows that they are very effective
strategies (see, e.g., Bechtel, 2008). On the other hand, Thompson is right
that the mind/brain is probably a dynamical network of processes with
top-down determinations. Moreover, there is growing consensus that
explaining the mind requires moving beyond the brain into the body and
environment. But the right conclusion to take from the dynamical and
extended nature of cognitive systems is not that mechanistic strategies
are wrong but that they are insufficient on their own and need to be sup-
plemented with other approaches. The dynamical approach is one plau-
sible candidate. As I showed earlier with the example of Beer’s perceptual
agent model, we can apply the mechanistic strategy of decomposition
to dynamical systems. And other interactions between mechanistic and
dynamical perspectives (e.g., Bechtel & Abrahamsen, 2010) show that
integrating them is a promising way to explain cognitive phenomena.
Another limitation of the mechanistic approach appears in the case
of complex and multifaceted mental phenomena, such as consciousness
or mental maladies. The mechanistic strategy seems limited in its ability
to describe the target phenomenon and analyze the description in order
to formulate explanatory hypotheses. It seems clear that in such cases
the mechanistic approach needs to be supplemented by another method-
ology which delivers both first-person and third-person descriptions. In
particular, without first-person insights and careful analysis of a subject’s
experience, we cannot capture the key symptoms of a mental malady,
its phenomenological structure, on the basis of which we can formulate
hypotheses and heuristics to guide treatment and empirical research. A
proper description of the phenomenon is also critical for subsequent
functional and structural decomposition, which, in turn, determines
the sketch of a how-possibly mechanistic model. These limitations are
typically overcome by various psychological methods including qualita-
tive phenomenological interviews and, in the case of mental disorders,
psychiatric interviews and symptomatology. For example, an integrative
approach to psychiatric disorders called Mechanistic Property Cluster
(MPC) aims to represent mental maladies as networks of multilevel
causal mechanisms (Kendler et al., 2011). Such a property cluster con-
sists of multiple dimensions including psychological and sociocultural
ones, requiring qualitative methods able to address anomalous experien-
tial states, as well as genetic, neural, and environmental factors. Various
research fields in which such aspects of a malady are studied constrain
each other and contribute to a cluster. However, there are critical voices
here. As Colombo and Heinz (2019) argue, the mechanistic framework
and phenomenological approaches are conceptually and methodologi-
cally too far from one other to make integration plausible; for exam-
ple, they acknowledge inconsistent assumptions about mental maladies:
mechanism argues the malady can be decomposed and localized whereas
phenomenology assumes it cannot. Thus, Colombo and Heinz propose
another approach to integration in psychiatry called the dimensional
framework, which is based on the notion of computational phenotypes,
that is, a measurable behavioral or psychological type defined in compu-
tational terms. Phenomenology plays an important role in this approach;
it contributes to the specification of the explanandum phenomenon and,
in particular, it helps specify what a subject aims to accomplish within an
ecological context. Phenomenological analyses can also deliver a “narra-
tive glue” that helps elucidate the relation between a patient’s experience
and computational phenotypes.
To conclude, mechanistic explanations have their specific limitations.
Some of them can be easily dismissed, whereas others indicate the need to
overcome them through integration with other approaches to cognition
in order to create a multilevel integrative explanation. In the following
section, I address the issue of integration.
3.4 Explanatory Integration of Cognitive Science

On the one hand, attempts to explain consciousness and related mul-
tifaceted mental phenomena have revealed the limitations of different
explanatory models. On the other, multidisciplinary projects in cognitive
(neuro)science have raised the issue of methodological integration. In this
section, I introduce the idea of theoretical integration inspired by the
conception of interfield theories and mechanistic multilevel explanations.
Mechanistic multilevel explanations present a reasonable compromise
that stresses the causal scientific explanation but at the same time does
justice to the real complexity of mental phenomena and the plurality of
scientific methods and vocabularies applied in cognitive science.
3.4.1 Theoretical Reduction, Unification, and Integration

The D-N model of explanation introduced the idea of theoretical reduc-
tion (Oppenheim & Putnam, 1958). One theory can be reduced to
another if the terms of the former may be expressed in terms of the latter.
So, for example, the vocabulary of psychological theory was thought to
be potentially reducible to the vocabulary of biological theory, which
would, in turn, be reduced to the theory at the fundamental level of phys-
ics. The theoretical tools providing for such a reduction were hypotheti-
cal bridge laws in virtue of which higher level predicates were reduced
to lower level predicates, or, to put it differently, statements of higher
level theory can be deductively derived from statements of lower level
theory and bridge laws. The idea of theoretical reduction promised to
unify science. It was believed that ultimately all higher level scientific dis-
ciplines would be reduced to one fundamental discipline of physics. The
allure of unifying science was strong in the decades of the 1950s and the
1960s when the D-N model of explanation was dominant. Today, many
contemporary philosophers of science consider such model of reductive
unification flawed and endorse a more pluralistic view of science (see,
e.g., Dupré, 1995; Fodor, 1974).
One classic objection to reductive unification along the lines of the
D-N model has already been mentioned: multiple realizability (Fodor,
1974). Accordingly, such a reduction is flawed because the kinds of a
reduced theory cannot be clearly identified with the multiple kinds of
the reducing theory. This is so because the kinds of the reduced theory
can be realized by a potentially infinite number of kinds of the reducing
theory. The second common objection focuses on the nomological char-
acter of the reduction. Both the reduced and reducing theories consist
of lawlike statements, with bridge laws serving as intermediaries. Such
nomological conditions makes reductive unification difficult to apply in
cognitive science. As I argued earlier, in psychological and cognitive theo-
ries, we hardly find lawlike statements, and the explanatory framework
is rather functional (Cummins, 2000) or mechanistic (Craver, 2007) than
nomological.
Rejection of a strong conception of inter-theoretical reductive unifi-
cation does not undermine other approaches to unification and weaker
forms of integration. For example, Sorin Bangu (2017) argues that theo-
retical unity is closely related to the very idea of scientific explanation and
understanding and that unificationism has to rethink this idea, specifically
the causal component of explanation, in order to provide ontologically
reductive unification. But unification does not have to necessarily endorse
a strong version of reductionism—attempts to unify cognitive science
seem to base on other strategies (Miłkowski & Hohol, 2020). The first
is the already mentioned functionalist strategy, which delivers a general
framework for understanding cognitive dispositions as functional capaci-
ties. Functionalism provides a general notion of function and an explana-
tory model that takes the form of functional architecture, which amounts
to the target capacity. In a similar manner, mechanism provides explana-
tions of cognitive capacities in terms of underlying mechanisms. Here
the sketch of a mechanism underlying the target capacity is the explana-
tion. The second strategy relies on postulating some kind of grand prin-
ciple that organizes cognitive processes. An example of such a unificatory
strategy is predictive coding theory, which refers to the free-energy prin-
ciple (Friston, 2010). This strategy may also converge with the first one
and produce an explanation guided by the principle while also providing
a partial description of the functional and mechanistic organization of
the target system (e.g., Clark, 2013; Hohwy, 2013).
There are some conceptions of explanatory pluralism and related ver-
sions of theoretical integration that provide alternatives to theoretical
unification. Roughly speaking, integration is different from the idea of
reductive unification, which is characterized by the search for one expla-
nation on a fundamental level. Instead of reducing one scientific level
to another, integration focuses on coordination between scientific disci-
plines or research fields in order to use their efforts in explaining some
phenomenon.
Pluralism in science can be understood in a number of ways. For
example, eliminative pluralism (e.g., Barker, 2019) considers eliminating
some scientific concepts and replacing them with a collection of separate
concepts. Competitive pluralism, derived from the Popperian account of
science, argues that competition between alternative explanations of the
same phenomenon is a rational strategy because it is the best way of
testing and ultimately selecting the most corroborated explanation. Imre
Lakatos’s (1980) research programs are examples of competing explana-
tions that do not necessarily lead to elimination. According to Sandra
Mitchell (2002), competitive pluralism is inefficient in the case of mul-
tifaceted phenomena, which can have multiple causes, not to mention
that their properties are investigated in different research fields. A better
approach to explaining such phenomena is the integrative or compatible
pluralism she defends. In integrative pluralism, the current plurality of
sciences is not a symptom of immaturity, but the unavoidable charac-
ter of scientific endeavor following from the operation of idealization.
Theoretical models are the result of idealization and abstraction, and, as
such, they do not always refer to the same idealized system. Competition
is not required; on the contrary, a plurality of models may be useful
because different models present specialized partial accounts of the target
phenomenon and indicate its different causes and aspects. Such models
are not competing accounts but contributions to one integrative expla-
nation. That said, the application of different theoretical models in an
explanation of a particular phenomenon requires their integration.
Another integrative approach to pluralism in science stresses the inter-
field character of explanation. The general idea of how the integration
of two or more fields of study should proceed comes from Darden and
Maull’s (1977) conception of interfield theories. A field of study consists
of a central problem, related facts, and an explanatory goal together with
research methods and specific vocabulary. A well-established research
field may have a theory, but having one is not a necessary condition.
Importantly, several research fields can contribute to an interfield theory
corresponding to a multifaceted explanandum. For Darden and Maull
(1977), a good reason “for proposing an interfield theory exists when
two fields share an interest in explaining different aspects of the same
phenomenon” (p. 49). Chromosome theory is an example, one which
links the fields of genetics and cytology. An interfield theory emerges
from interactions between research fields, and several types of relations
between fields may contribute to an interfield theory:
1. A field may provide a specification of the physical location of an

entity or process postulated in another field.
2. A field may provide the physical nature of an entity or process pos-
tulated in another field.
3. A field may investigate the structure of entities or processes, the
function of which is investigated in another field.
4. Fields may be linked causally, the entities postulated in one field
providing the causes of effects investigated in the other (Darden &
Maull, 1977, p. 49).
Notice that the relations between fields have nothing in common with the
derivational relation between theories endorsed by proponents of theo-
retical reduction. An interfield theory emerges not from the deduction of
one theory from another but from complementing and constraining rela-
tions between research fields. So, for instance, one field may supplement
a description of an entity postulated in another field with structural or
physical details. Investigation of the causal nature of an entity in one field
may constrain research conducted in another field.
The integrative story of different research fields investigating the
same complex and multifaceted phenomena fits very well with cogni-
tive (neuro)science. Cognitive phenomena are complex and have mul-
tiple aspects requiring study from different perspectives using various
methods and theoretical vocabularies. Thus, any attempt to offer a com-
plete and reductive explanation, that is, an explanation in terms of one
fundamental scientific discipline such as physics, seems like a dead-end.
Rather, the integrative strategy seems to be the best option at the current
stage of cognitive science’s development. It was argued recently that it
is the mechanistic model of multilevel explanation that can deliver the
framework required to integrate such a multidisciplinary endeavor (e.g.,
Craver, 2007; Miłkowski, 2016). As Miłkowski and Hohol (2020) argue,
“the mechanistic account of integration is compatible with nonextreme
versions of explanatory pluralism”. The mechanistic model of integration
requires, however, that various methodologies and theories are not fully
autonomous and that they can be mutually constrained.
3.4.2 Mechanistic Integration of Cognitive Science

Mechanism situates itself in opposition to the strong version of inter-
theoretical reductionism as well as to explanatory fundamentalism,
which seeks explanation at some fundamental level, for instance, the
neural (Churchland, 1986) or molecular level (Bickle, 2003). According
to Craver (2007), such fundamentalism lacks evidential support, nor is
it supported by scientific practice and existing explanations in neurosci-
ence. For example, behind the success of the discovery of the long-term
potentiation mechanism that underlies spatial memory was a resignation
from the reductionist program in favor of integrating research from dif-
ferent fields. In mechanistic integration,
the central idea is that neuroscience is unified not by the reduction of

all phenomena to a fundamental level, but rather by using results from
different fields to constrain a multilevel mechanistic explanation. The
goal of building a mechanistic explanation, rather than an explana-
tion simpliciter, provides an abstract framework or scaffold that is
elaborated as different fields add constraints on the explanation.
(Craver, 2007, p. 231, emphasis in the original)
To put it differently, mechanistic explanations oscillate between differ-

ent levels, from the behavior of an organism to processes at a molecular
level, and link all the processes relevant to the constitution of the target
phenomenon.
That said, there are philosophers (e.g., Hensel, 2013) who argue that
there is much more reduction in the mechanistic approach than is explic-
itly acknowledged by its proponents, because explanations of higher
order phenomena are formulated in lower level terms. Godfrey-Smith
(2014, p. 16) characterizes the mechanistic approach as reductionist in
a modest sense; that is, properties of the whole system are explained in
terms of its parts and their organization. This reductionism is weak since
the properties of the whole cannot be found on a lower level. And Bechtel
(2008, p. 129) uses the term mechanistic reduction, although he claims
that it is Janus-faced and that a mechanist can be a reductionist and
emergentist at the same time. This is so because, according to the mecha-
nistic approach, the behavior of the whole system surpasses the behavior
of its parts, and furthermore, it is context-dependent. In this regard, a
mechanistic explanation can be read as ontologically reductive (mental
phenomena are produced by physical mechanisms, but there are emer-
gent properties) and epistemologically weakly reductive or nonreductive
(explanations of mental phenomena are multilevel and include multiple
languages of description and methods of research). Furthermore, mecha-
nisms not only have emergent properties at a higher level; they are also
situated in an environment, and understanding that is equally important.
As Bechtel (2008) puts it,
the decomposition required by mechanistic explanation is reduction-

ist, but the recognition that parts and operations must be organized
into an appropriate whole provides for a robust sense of a higher
level of organization. A scientist seeking an account at this higher
level will find it essential to undertake independent study of the orga-
nization of the mechanism and how it engages its environment.
(p. 130)
To sum up, although mechanism seeks to explain phenomena in terms

of a mechanism’s parts and activities, it acknowledges that a complete
explanation is multilevel and requires theoretical integration of different
fields of research, not a reduction of one theory or discipline to another.
The mechanistic approach to integration develops the conception of
interfield theories (Darden & Maull, 1977). The key notion to under-
standing this approach is constraint. According to Craver (2007), “dif-
ferent fields are integrated when their findings provide constraints on
the space of possible mechanisms,” which as “conceived most inclusively,
contains all the mechanisms that could possibly explain a phenomenon”
(p. 247). The dimensionality of this space is defined by the properties and
internal organization of a hypothetical mechanism. Constraints provided
by different fields limit these dimensions. To put it differently, the results
of research in different fields narrow down the number of how-possibly
models of mechanisms by providing a hypothetical specification of the
searched for mechanism’s properties.
Constraints can be of different types. Craver (2007, p. 249), for exam-
ple, identifies several types of constraints in two categories: intralevel and
interlevel. The former are provided by research fields studying a phenom-
enon on the same level; for instance, one field studies anatomical proper-
ties such as the shape, size, and localization of neurons, and another field
investigates the electrophysiological properties of neurons, including the
characteristics of signal transmission. Constraints from these fields con-
tribute to an interfield theory of neuronal behavior. The second category
of constraints, interlevel constraints, concerns fields studying phenomena
on different levels of organization of a how-possibly mechanism. First,
such interlevel constraints may be epistemic and related to how a target
phenomenon is conceptualized—one field can make another field use a
different taxonomy. For example, thinking about navigation in space in
terms of spatial maps constrained research on a lower level to search
for a mechanism that could produce internal representations of space,
that is, the mechanism of long-term potentiation. Another example of
top-down interlevel constraints is functional constraints where a higher
level field delivers functional specifications for a mechanism on a lower
level. However, the direction of the constraining relation is not always
top-down. Research on a lower level can also influence how we think
about a studied phenomenon. For instance, thanks to research and lesion
studies from the field of neurology (Scoville & Milner, 1957), we know
that memory is not a homogeneous phenomenon but consists of several
related sub-capacities. Among interlevel constraints, temporal ones are
especially interesting in the context of multilevel explanations of men-
tal phenomena. First, notice that mental phenomena are intrinsically
temporal; that is, they occur in time, change over time, and are often
related to past experiences or refer to objectives in the future. Memory
and action planning are obvious examples, but they are not the only ones.
Studying such phenomena from an external observational perspective as
well as from a first-person perspective can be an important source of
constraints on the dynamics of the hypothetical mechanisms underlying
these phenomena.
3.4.3 Autonomy
The autonomy of integrated fields of research is a matter of ongoing
debate. A strong notion of autonomy assumes that scientific disciplines
are autonomous from each other if there are no direct constraints
between them. As I have shown above, this kind of autonomy is some-
times ascribed to first-person and psychological-functional explanations.
For instance, Fodor (1965) argues that psychological-functional explana-
tions of mental states tell us nothing about their neural realizations and,
vice versa, that neurological research will not shed light on functional
relations between mental states. Thus, psychological explanations can be
developed independently of neuroscience. Such a notion of autonomy
is, however, too strong if we want to integrate such different domains in
virtue of mutual constraints. As Piccinini and Craver (2011) argue, “func-
tional analysis and mechanistic explanation are not autonomous because
they constrain each other; in addition, we argue that they can’t possibly
be autonomous in this sense because functional analysis is just a kind of
mechanistic explanation” (p. 290). To put it differently, functional analy-
sis describes the functional properties of a possible mechanism responsi-
ble for the target phenomenon. Accordingly, functional analysis delivers a
functional sketch of a hypothetical mechanism responsible for the target
capacity. This sketch describes the system’s functional organization, but,
at the same time, the sketch ignores the structural details of the mecha-
nism, such as its physical components and their organization. When the
omitted details are added, the sketch becomes the scheme of the hypothet-
ical mechanism. Ultimately, the scheme becomes a complete explanation,
that is, a multilevel model of the actual mechanism responsible for the
explanandum phenomenon. In reply to this, Roth and Cummins (2017)
argue that proponents of mechanistic integration confuse explanatory
autonomy with confirmation autonomy. Functional-psychological expla-
nations are fully autonomous and thus cannot be integrated with mecha-
nism, but mechanistic neuroscience can verify which functional design is
correct. A mechanistic model can confirm a functional design, but it does
not follow that the model actually explains the target capacity. It is the
functional design that is the key to explaining the target phenomenon.
However, as Roth and Cummins (2017, p. 40) conclude, the autonomy
of functional explanation does not exclude that both functional analysis
and mechanism can constrain each other’s hypotheses.
That being said, we can consider integrated fields autonomous when
adopting a weaker notion of autonomy. According to Piccinini and
Craver (2011),
one scientific enterprise may be called autonomous from another if

the former can choose (i) which phenomena to explain, (ii) which
observational and experimental techniques to use, (iii) which vocab-
ulary to adopt, and (iv) the precise way in which evidence from the
other field constrains its explanations.
(p. 288)
The last point is crucial for theoretical integration. Furthermore, as

Kaplan (2017, p. 3) argues, explanatory autonomy can come in degrees.
Accordingly, a scientific discipline may be completely or partially autono-
mous. For instance, disciplines that investigate the same phenomenon
from different points of view are not completely independent in (i) choos-
ing the phenomenon, therefore they are not fully autonomous. The results
of one discipline investigating a phenomenon are relevant for another
discipline investigating the same phenomenon, and that relevancy can
come in the form of mutual constraints. However, their autonomy is
maintained in the methods of investigation they choose and the terminol-
ogy in which their explanations are formulated. For example, chronobi-
ology and computational cognitive science study circadian rhythms using
different methods and analytical tools. The former model is grounded
in empirical observation and formulated in terms of dynamical systems
theory. The latter uses computational modeling in order to propose a
hypothetical mechanism responsible for a particular behavior. However,
by observing mutual constraints, an integrated dynamical-mechanistic
model can be achieved (Bechtel & Abrahamsen, 2010). Another example
comes from explaining mental maladies, such as schizophrenia, which
are often informed by research from various fields of research includ-
ing psychiatry (also phenomenological psychiatry), psychology, soci-
ology, anthropology, and, last but not least, neurobiology (e.g., Engel,
1977; Kendler et al., 2011). In this respect, these fields of research remain
autonomous, yet they constrain each other in other respects.
To conclude, if the mechanistic approach does not aim to eliminate
higher level descriptions of the target phenomenon and it accepts the
importance of research from various fields of study, then it can be con-
ceived of as a strategy for integrating rather than unifying cognitive sci-
ence. As Craver (2007) puts it, “the reductive goal of globally relating
two fields through derivation of laws should be replaced by a mosaic
image of multiple fields making punctate contributions to an abstract
sketch of a mechanism” (pp. 246–247). Mechanistic integration is not
elegant; it does not produce homogenous and smooth explanations. On
the contrary, it gathers various research findings and theoretical contribu-
tions to constrain the space of possible mechanisms. As a result, a typical
mechanistic explanation consists of multiple levels. Such a model of inte-
gration requires that integrated research fields cannot be fully autono-
mous; they choose a phenomenon to be explained (which is often the
same phenomenon for another field) and terminology of description as
well as methods of investigation, but they have to be in a relation of
reciprocal constraints; that is, evidence or concepts from one field must
constrain explanations of the other.
3.5 Is Phenomenology Explanatory?

Can phenomenology derived from Husserlian tradition be considered a
type of one of the explanatory models introduced earlier, or does it afford
a unique model of explanation? Can phenomenology be integrated with
other research fields and feature as part of a multilevel explanation? Is
it plausible that phenomenology could constrain a mechanistic model
in some respect? Before I try to answer these questions, I briefly discuss
Husserl’s understanding of explanation.
Husserlian phenomenology was intended to be not only a methodical
study of consciousness (see Chapter 1) but also a strict (strenge) science
(Husserl, 2002), that is, a rigorous science with its own methodology and
ideal of exactness and precision. To some extent, Husserl’s phenomeno-
logical project is similar to positivist ideas in science, such as abandoning
folk language and introducing new scientific terminology, stressing the
precision of scientific statements, controlling commonsense beliefs and
their demarcation from scientific knowledge, and emphasizing the need
for systematic and transparent explanations (see Nagel, 1961). We can
find all these norms more or less applied in Husserlian phenomenology,
of course in their specific phenomenological formulation. Also, much as
science, phenomenology aims to formulate generalizations about the phe-
nomena it investigates. But Husserl also argues for phenomenology being
grounded in the certainty of inner intuition (Anschauung). As the famous
principle of all principles states, “every originary presentive intuition is a
legitimizing source of cognition” (Husserl, 1982, p. 44). This condition is
more difficult to accept by the standards of the scientific approach that
stress the necessity of testability. But the self-certainty of phenomenol-
ogy may be replaced with some sort of intersubjective corroboration by,
for example, comparing eidetic descriptions of experience (Gallagher &
Zahavi, 2012) or empirically as in Gallagher’s front-loaded phenomenol-
ogy (e.g., Gallagher & Brøsted Sørensen, 2006). It is also important to
notice that although phenomenological method includes a careful descrip-
tion of lived experience, it should not be considered a purely descriptive
endeavor. Husserl distances himself from the descriptive psychology of
Franz Brentano and conceives phenomenology as explanatory. A source
of confusion is that Husserl repeatedly introduces various descriptions
of acts of consciousness and their contents. Furthermore, as I showed
in Chapter 1, the phenomenological genetic approach, that is, the one
that investigates the origin of lived experiences and motivational rela-
tions between different sorts of mental states, is considered by Husserl to
be explanatory, in contrast to the static approach, which describes and
categorizes different types of mental contents and acts. Both the static
and genetic approaches give us an understanding of how experience is
produced, and both reveal motivational regularities involved in the expe-
rience of intentional objects. But only identification of the underlying
passive processes of conscious life, for example, related to affections and
associations, gives us an explanation of the constitution of experience.
In short, the genetic approach shows that phenomenology should not
be understood as a mere description of mental phenomena, because it
reveals structures and functions of consciousness that are involved in the
production of the studied experience.
In the following sections, I focus on the notion of phenomenological
explanations of mental states and personal behaviors. After considering
various interpretations of the phenomenological type of explanation, I
argue that phenomenology offers a sort of constitutive understanding of
studied mental phenomena rather than complete explanations.
3.5.1 Types of Phenomenological Explanation

When Husserl writes about the objective of phenomenology, he uses the
notion of explanation (Erklärung) of the structures of experience, that is,
disclosing or revealing structures that are implicit and hidden in everyday
lived experiences, which is contrasted with the clarification (Aufklärung)
by description applied in descriptive sciences, such as Brentano’s descrip-
tive psychology (Husserl, 1980). A phenomenological explanation is not
a mere description of one’s own inner states; it also differs from the causal
explanation endorsed by the natural sciences. As I showed in Chapter
1, phenomenology is a method of “seeing” structures of experience and
providing concepts to empirical psychology. For example, by varying
experience in imagination, a phenomenologist can apprehend in intuition
its invariant properties. According to Husserl, revealing the structures of
experience is genuinely explanatory—describing the generic structures of
consciousness explains how a particular experience is constituted or, to
put it differently, the types of activities of consciousness involved in the
production of experiences of a certain kind.
Despite the explanatory ambitions of phenomenology, it is not that
easy to say what type of explanations phenomenology is meant to pro-
duce. At first glance, it might seem that phenomenology fits well to the
distinction between the personal and sub-personal level of explana-
tion and is confined to the former. For example, Heath Williams (2020)
argues that Husserlian phenomenology offers genuine explanations on
the personal level (e.g., Husserl, 1989) and that such explanations are
neither causal nor deductive-nomological. Such personal-level phenom-
enological explanations are also irreducible to sub-personal levels, that
is, functional and physical, which constitute sub-personal explanations
according to Dennett (1969). In his argumentation, Williams refers to
Husserl’s “personalistic attitude” in which behavior is explained as a
“motivational nexus.” According to Husserl (1989, pp. 241–242), an
explanation of someone’s behavior can take the form of a clarification
of their motivations. As I have shown in Chapter 1, Husserl (e.g., 1977,
p. 108) introduced the notion of motivation to explicate quasi-causal
relations between mental states. We can use various types of motivation,
such as motivations of reason or associations, to explain why a person
behaved in a certain way. But as Husserl (1989) writes, “the ‘because-so’
of motivation has a totally different sense than causality in the sense of
nature. No causal research, no matter how far-reaching, can improve the
understanding which is ours when we have understood the motivation of
a person” (p. 241). In short, mental states are in motivational relations,
which means they can affect one another. These relations, however, are
not causal and lawlike generalizations about them have nothing in com-
mon with the laws of nature formulated in the “naturalistic attitude,”
which is different from the personalistic one. Moreover, according to
Williams, such motivational explanations cannot be expressed in lawlike
statements with a ceteris paribus clause, and thus cannot be seen as D-N
type explanations. This is so because “motivational accounts potentially
engage with the concrete and hence the singular, and might refer to my
particular, situated, empirical, personal character, history, or tempera-
ment in order to explain a singular instance” (Williams, 2020, p. 19). In
other words, an explanation of someone’s behavior in terms of motiva-
tions requires an understanding on the basis of empathy (Einfühlung)
that takes into account the target’s individual determinations rather than
knowing some sort of a general law or psychological platitude. The latter
is often referred to in folk-psychology explanations as a basis of infer-
ences about other’s behavior (e.g., Stich & Nichols, 2003). But Husserl’s
motivational explanation differs from personal level explanation, which
relies on folk theoretical concepts like belief, desire, and intention, for
two further reasons. First, mental states play a motivational rather than
causal role in Husserl’s work, unlike in the case of folk-psychological
explanations. Second, Husserl acknowledges that the motivational nexus
of our beliefs and actions can also include unconscious or pre-reflective
states, such as instincts, drives, and affections, which impairs the rational-
istic constraint crucial for the “intentional stance.”
I agree that there is a strong personal dimension in phenomenological
analyses of motivational relations and intentional formations, but I do not
perceive phenomenology as offering purely personal-level motivational
explanations of our mental lives. First, explaining the motivational nexus
underlying someone’s behavior, although useful in explaining someone’s
particular behavior, cannot be conceived of as an explanation of some
general mental disposition. That would require generalization, that is,
a justification to the effect that it can be applied to other cases and not
only to a given particular one. This is problematic since the motivational
nexus underlying someone’s actions is always determined by individual
history, habits, and associations. But I don’t think that inability to gen-
eralize a motivational nexus is some major problem for phenomenology,
because, as I argued in Chapter 1, Husserl was not primarily interested
in explaining the individual mental states and behaviors of a person,
but in explaining universal structures of consciousness common to all
human beings. That is why he applies reductive and variation methods
to individual experiences: to abstract from particular determinations and
extract generic structures of experience.
Second, personal-level categories are, so to speak, the tip of the ice-
berg of our mental lives. Husserl was aware of that, and in his genetic
phenomenology, we see an attempt to account for passive synthesis of
intentional states. The level of passive syntheses is above the sub-personal
understood as neural mechanisms, about which Husserl says nothing,
but it is below the active constitution in a person’s conscious acts. It is
the phenomenological pre-egoic dimension of passive constitutions, for
example, the constitutive functions of temporal syntheses, or different
sorts of passive associations and affective motivations that determine our
beliefs and desires on the personal level. Phenomenology is not only able
to analyze these sub-personal processes but also proposes a hierarchical
model of the levels of the constitution of experience. A simple stratifica-
tion differentiates two levels, that is, “the lower levels, psychic passivity,
and the specifically mental, namely, the life which proceeds in I-centered
acts” (Husserl, 1977, p. 99). More advanced analysis reveals several lev-
els of the constitution of experience (Rodemeyer, 2020), including hyletic
syntheses (original sensations), passive syntheses (associations, affectivity,
habitual determinations), active syntheses (egoic acts of sense formation,
e.g., acts of perception, judgments), the interpersonal (e.g., acts of empa-
thy), and intersubjective (related to being part of a community, history,
culture). At each of these levels, a specific type of constitutive function is
involved, for example, on the lower levels, functions of time-conscious-
ness (retention and protention) or functions related to associative moti-
vations. On the upper levels, these are noetic functions involved in acts
of judging, imaging, remembering, and so on. These phenomenological
models of multilevel constitution can be applied to particular phenomena
in order to show how that phenomenon is constituted on these levels. For
example, Lanei Rodemeyer (2020) considers how phenomenological lev-
els of constitution can help us understand eating disorders. Her analyses
show, for instance, that beliefs and meanings related to the distorted self
become embodied in patterns of behavior and how explicit beliefs are
passively motivated by tacit presumptions.
To sum up my argument, phenomenological analyses are not exclu-
sively at a personal level, nor do they concern explanations of the behav-
ior of a particular person. Passive constitution that precedes egoic activity
is included in the phenomenological understanding of consciousness and
is crucial for accounting for personal level activity. The main objective of
phenomenological investigations is the generic structures of experience,
not motivations of particular behaviors.
If phenomenology goes beyond the personal level and investigates
generic structures of experience, what type of explanation does it offer?
Alphonso Lingis (1986) considers phenomenological explanation in the
context of intuition (Anschauung), one of the key categories in transcen-
dental philosophy. According to him, phenomenology is different both
from the deductive explanation of formal science and from the inductive
reasoning of empirical science because it seeks to ground these sciences
in intuitive apprehension. As I mentioned in Chapter 1, intuitive appre-
hension is a methodological process in phenomenology related to the
application of imaginative variation, that is, the procedure of modify-
ing and replacing parts of an experienced object in order to apprehend
its invariant structure or, to put it differently, its eidos (Husserl, 1973,
para. 87). Apprehension of eidos suggests that the results of phenom-
enological investigations are universal; that is, they describe the general
structure of an experience of a certain type, not an individual experience.
Apprehending the eidos of various types of objects of apprehension also
reveals the “productive activity” (Husserl, 1969) of consciousness, namely,
forms of intentional functions, which generate experiences of a certain
kind. In this sense, phenomenology is a sui generis intuitive explanation
that, on the one hand, relies on bringing apprehensions in intuition to
their original form (eidos) and, on the other, revealing consciousness’
activities involved in the process of the production of experience.
Because phenomenology aims to express motivational regularities of
experience in the form of lawlike generalizations, the interpretation of
phenomenology as a sort of “intuitive” explanation can be developed
further to understand phenomenological eidetics as a form of nomologi-
cal explanation. Indeed, Husserl often talks about laws in reference to
our mental lives and uses terms such as “eidetic law” (1980), “essential
law” (2005), “lawful regularities,” “structural lawfulness,” and “lawful
functionality” (2001). According to Rochus Sowa (2012), “as a priori
research, Husserlian phenomenology aims at descriptive eidetic laws.
Husserl uses two linguistic or logical forms for these laws: generic judg-
ments and universal judgments” (p. 258). For example, an eidetic law
is that “any conceivable phenomenal color has some sort of phenom-
enal extension” (Sowa, 2012, p. 258). Another example of such laws
can be found in Husserl’s investigations of time-consciousness. These
are the so-called laws of time-consciousness involved in the apprehen-
sion of temporal objects, such as the law of modification of the now-
phase into retention, and further into a retention of retention (Husserl,
1991). Genetic-eidetic laws, which Husserl sometimes calls “laws of pas-
sivity” (1973), are introduced by Husserl, for example, in Experience
and Judgment (1973), where he discusses laws of judgment formation,
or in his manuscripts Analyses Concerning Passive and Active Synthesis
(2001), where he considers regularities of associations, affections, and
rememberings. Passive constitution concerns, in particular, affections that
elicit attention directed toward objects enabling associations and remem-
bering, which, in turn, contributes to expectations about the object.
Husserl (2001, para. 33) tries to identify some regularities in this process,
for example, laws of the propagation of affection.
So according to Sowa’s interpretation, phenomenology offers “eidetic
explanation” or “explanation by law” (Husserl, 1977), however, a better
name might be nomological-intuitive explanation, because the subsump-
tion under law does not have a formal deductive form but is grounded in
intuition (intuitive apprehension). Sowa (2012) argues that “the peculiar
scientificity of lawful eidetic-descriptive statements consists in this dou-
ble testability or falsifiability” (p. 261, emphasis in the original). In the
first case, phenomenological eidetic laws can be falsified in the process
of imaginary variation (Sowa, 2012, p. 259). Accordingly, the previously
mentioned method of eidetic variation is not only a way of seeing or
discovering eidetic laws but also of testing them, that is, by construing in
phantasy possible variants of the object of experience. If the procedure
of variation does not falsify the target claim, then it is acknowledged as a
sort of eidetic law “until further notice.” In the second case, Sowa (2012)
also mentions that hypothetical laws may be tested by “facts of scientific
or pre-scientific experience” (p. 259, emphasis in the original). Scientific
knowledge, for instance, neuropsychological or psychiatric, can, at least
in principle, falsify phenomenological laws by indicating an empirical
fact, for example, a case of a disease affecting cognitive functioning,
effectively establishing that some law is not universal.
Reading phenomenological explanation as a nomological-intuitive
explanation, it is important to remember two things. First, that it relies
on transcendental phenomenology in contrast to the phenomenological
psychology endorsed in this work. Second, that there are essential dif-
ferences between phenomenological laws and natural or formal laws
(Husserl, 1982, para. 6–9). As we can read in the first book of Ideas, “the
unrestricted universality of natural laws must not be mistaken for eidetic
universality” (Husserl, 1982, p. 15). Eidetic laws are different from the
causal laws of natural science. Phenomenological transcendental reduc-
tion shifts from the natural attitude toward the phenomenological and
brackets causal relations between worldly objects and mental states. For
example, as Husserl (2001) puts it, “the rubric ‘association’ characterizes
for us a form and a lawful regularity of immanent genesis that constantly
belongs to consciousness in general; but it does not characterize, as it does
for psychologists, a form of objective, psychophysical causality” (p. 162).
3.5.2 Phenomenological Understanding
The interpretations of phenomenological explanation discussed earlier
seem flawed. Understanding phenomenology as a personal explana-
tion seems to narrow down its scope and objective, that is, investigating
generic structures of experience. The second interpretation focuses on
the transcendental character of phenomenology and grounds phenom-
enological explanation in intuitive apprehension. Although it is true that
Husserlian phenomenological “seeing” is grounded in intuition, such a
reading does not elucidate what the form of this intuitive explanation
is. The third proposal also considers phenomenology a transcendental
philosophy and argues that the form of intuitive explanation is nomo-
logical; that is, phenomenological intuitions apprehend eidetic laws that
are expressed in the form of generic and universal judgments. However,
the interpretation of phenomenology as a sui generis nomological-intui-
tive explanation, although supported by Husserl’s terminology, construes
transcendental phenomenology as explanatorily weak. Recall that the
force of D-N explanations relies on the universality of the laws of nature
and the certainty of deductive inferences. But phenomenological laws do
not appear to be universal, nor are phenomenological claims deductive;
they are rather the result of the reductive methodology enabling essential
seeing. Moreover, phenomenological lawlike generalizations have limited
predictive power unless they are specified and applied to empirical phe-
nomena, which is hard to do when one applies transcendental reduction,
suspending causal relations. So a condition to make testable phenomeno-
logical predictions is to downplay the transcendental dimension, just as it
is downplayed in phenomenological psychology.
According to the phenomenological agenda, a phenomenologist sus-
pends claims about the causal structure underlying mental phenomena.
However, in some cases, we clearly have to refer to the sub-personal level
in order to fully explain someone’s behavior. The addiction mechanisms
discussed earlier or mental diseases such as depression are clear examples
of such cases. Generally speaking, it seems, on the basis of the debate
surrounding the integration of cognitive science, that in most cases a
complete explanation of a mental phenomenon requires addressing sev-
eral levels including the phenomenal level of experience, psychological
level of cognitive capacities and functions, and the level of neural mecha-
nisms. Clearly phenomenology cannot deliver such a complete explana-
tion alone. It is a result of integrating different frameworks and various
research fields. That said, it is also clear that phenomenology gives us an
insightful account of our mental lives and a general understanding of
the structures of consciousness that can be informative to other research
fields. But rather than a genuine explanation, phenomenology offers us
a specific clarification or way of understanding conscious phenomena.
Now, what kind of understanding does phenomenology provide, and
how is it related to insights from other disciplines?
It is often argued that understanding is inseparable from explanation,
that understanding is the main goal of explanation, or that understand-
ing why a certain phenomenon occurred is part of the explanation of that
phenomenon. However, according to Peter Lipton (2009), understanding
and explanation are distinct: understanding concerns the cognitive ben-
efits that an explanation provides but is not identical with an explanation
as such. Lipton’s line of reasoning is as follows: a good scientific explana-
tion provides cognitive benefits related to four kinds of understanding
or knowledge—of causes, of necessity, of possibility, and of unification
(Lipton, 2009, p. 43). But these epistemic benefits can also be provided
by different types of cognitive activities.
First, according to Lipton, causal information often comes from tacit
knowledge different from explanatorily explicit representations. For
example, seeing a physical model or graphical representation of a system
may give us an understanding of causal information without providing a
genuine explanation. And manipulation, for example, physical interven-
tions on parts of mechanisms, can be very informative with respect to the
causal relations between parts and their organization, but we would not
say that the resulting understanding provides us with a full explanation.
Second, an understanding of necessities may emerge from thought experi-
ments, but we would not say that thought experiments are explanatory.
For example, Galileo’s thought experiment concerning the acceleration
of massive objects shows us that acceleration is necessarily independent
of mass. He tells us to imagine two falling objects connected with a rope,
one heavier than the other. If acceleration depended on mass, then the
lighter object would slow down the heavier one. But when we consider
these objects together as one mass, they should apparently fall faster than
they would separated. Thus, thinking of acceleration as dependent on
mass leads to contradiction, and so we have to accept that they are inde-
pendent. The thought experiment conveys some worldly necessity, but
it does not tell us why it is the case. We may know that events happen
necessarily, but we can still not know why. Third, as Lipton (2009) argues
further, “we can gain actual understanding from merely potential expla-
nation. As in the case of thought experiments, the kind of understand-
ing gained is modal, but here what is gained is knowledge of possibility
rather than of necessity” (p. 49). Potential explanations give us an under-
standing of how things might be related, but they are not yet complete
explanations. Consider, for example, how-possibly mechanistic models;
they help us in understanding how a target phenomenon may possibly be
realized. Despite that, they are incomplete and merely possible explana-
tions; they give us an actual understanding by, for instance, showing the
type of actual mechanism responsible for the target phenomenon. Fourth,
unification, that is, understanding how different phenomena fit together,
can be achieved through sharing exemplar mechanisms in the scientific
community, which organize research practice, for example, allow for for-
mulating new problems or evaluating proposed solutions but that are not
explanatory.
I think that phenomenology should be considered in an analogous
manner—as an enterprise providing us with a better understanding of
mental phenomena but not complete explanations. The cognitive benefit
that phenomenological understanding provides concerns modal knowl-
edge of mental phenomena. Eidetic analyses give us insight into the nec-
essary conditions for an occurrence of an experience of a certain type,
and these necessities are often expressed in lawlike statements. Genetic
analyses provide us with an understanding of how a studied experience
may possibly be produced by the underlying passive processes of syn-
thesis, as, for example, in the case of the perception of a temporal object
such as a melody elucidated by functions of retention, protention, and
ur-impression (Husserl, 1991).
The phenomenological method of imaginative variation, which enables
“essential seeing” can be conceived of as analogous to thought experiments
which help us understand the necessities governing the world (as I noted
in Chapter 1, Husserl himself considered imaginative variation a form of
phenomenological experiment; see, e.g., Husserl 1980, pp. 44–45). A phe-
nomenologist varies experience in imagination in order to see that certain
modifications lead to contradiction or a change in the experienced object’s
identity, thus demonstrating what the necessary conditions for a certain
type of experience are. Such a procedure is not an explanation or a formal
proof but rather a “see it for yourself” strategy of providing understanding.
Notice that phenomenological thought experiments allow one to answer
some counterfactual questions regarding our mental lives. For example,
seeing an object from a certain perspective and imagining seeing it from
a different perceptual perspective allows me to answer what part of the
object I would see, what kind of perceptual experience I would have, if I
moved and saw it from that different perspective. That kind of prediction is
only possible in virtue of understanding the structure of perception.
Drawing on Michael Wheeler (2013), I call phenomenological under-
standing a constitutive understanding. Constitutive understanding con-
cerns “the identification, articulation and clarification of the conditions
that determine what it is for a phenomenon to be the phenomenon that
it is” (Wheeler, 2013, p. 143). According to Wheeler, phenomenological
analysis is a form of constitutive understanding; it reveals conditions
relevant to the target phenomenon and its key properties. Importantly,
this type of understanding is distinct from enabling understanding,
which
reveals the causal elements, along with the organization of, and the
systematic causal interactions between, those elements, that together
make it intelligible to us how a phenomenon of a certain kind could
be realized or generated in a world like ours.
(p. 143)
The mechanistic approach is an example of such an enabling understand-

ing, as its objective is to describe the causal mechanism underlying the
studied phenomenon. Wheeler brings up this distinction in the context
of naturalizing phenomenology, and he argues that both types of under-
standing—constitutive/phenomenological and enabling—are distinct, yet
they can mutually influence and constrain each other.
3.6 Conclusion: Toward Integration of Phenomenology With

Multilevel Mechanistic Explanation
In this chapter, I discussed key explanatory models applied in cognitive
science. I argued that the D-N model and the related idea of nomological
reduction and unification of science are outdated. In contemporary cogni-
tive science there are various models of explanation including personal,
functional, dynamical, and mechanistic explanations. I argued that all
these approaches have limitations when applied to multifaceted cognitive
phenomena; thus, an integrative approach seems to be an effective option.
I consider the mechanistic model of explanation crucial because it provides
a framework for integrating the diverse scientific work in cognitive science
into one explanation. Accordingly, different fields of research can be inte-
grated if they provide constraints on the space of possible mechanisms.
Considering phenomenology in the context of models of explanation
and the integration of cognitive science, I argued that phenomenology is
not merely descriptive, nor does it offer a complete explanation of mental
phenomena. The nomological-intuitive approach, which relies on tran-
scendental phenomenology, is distinct and autonomous from the natural
sciences, and thus it is not amenable to naturalization. On the contrary,
the weaker conception of phenomenology based on phenomenological
psychology (see Chapter 1) seems amenable to naturalization because
it is not fully autonomous; that is, it is autonomous in choosing its own
methods of investigation and vocabulary, but it shares explananda with
other research fields and thus can be constrained by them. It provides
a special kind of understanding that gives us insight into the necessary
conditions of the studied experience and shows how the experience is
possibly generated by underlying processes and functions of conscious-
ness. Importantly, this constitutive understanding is complementary with
enabling understanding, which focuses on the causal conditions of the
realization of the target phenomenon. Both types of understanding may
constrain one another and contribute to complete explanations.
As I have shown in the previous chapter, it is common to think of the
naturalization of phenomenology in terms of reciprocal constraints in the
debate. The new approach I defend relies on considering whether phe-
nomenology could provide constraints on the space of possible mecha-
nisms and hence be integrated with cognitive science in accordance with
the mechanistic integrative framework. Rethinking constraints in accor-
dance with the mechanistic framework leads us to a new approach to
naturalizing phenomenology that does not aim to find a unified theory of
cognition. It also does not conceive phenomenology as a part of experi-
mental design. It considers phenomenology a part of building multilevel
explanatory models of mechanisms. Various research fields participate
in this process by confining the space of how-possibly mechanisms, pro-
viding information about the mechanism’s components, organization,
functions, dynamics, and so on. In other words, the question of whether
phenomenology can be naturalized now concerns whether it can effec-
tively constrain the space of possible mechanistic models.
In the next part of the book, I address this issue in detail. Drawing on
Craver’s (2007) distinction between intralevel and interlevel constraints,
I argue that phenomenology can provide the latter. In particular, I con-
sider two hypotheses. The first functional hypothesis, which I develop in
Chapter 4, is formulated in a top-down manner. I start from Husserlian
phenomenology as a theory of acts of consciousness and experience and
show that phenomenological analysis shares similarities with functional
analysis, and thus, they can, in principle, provide interlevel functional
constraints on mechanistic models. Phenomenology, much like functional
analysis, employs a strategy of decomposition that is used to analyze the
contents of the experience and functions involved in the production of
the studied experience. The second hypothesis, discussed in Chapter 5,
is bottom-up and concerns dynamical constraints. I start by discussing
research in cognitive science in which DST is applied, including attempts
to combine DST with phenomenological methods in neurophenomenol-
ogy. I argue that application of DST in neurophenomenological method,
in particular in analysis of the diachronic structure of experience, is
weak, but it can provide us with an understanding of the dynamics of the
studied experience, for example, the phases of experience and evolution
of experience in time. This understanding of the dynamical character of
first-person experience is sometimes the key to explaining a target phe-
nomenon, and it can provide constraints on dynamical-mechanistic mod-
els. If I am right, then functionalism and dynamism may be considered as
bridging phenomenology with mechanistic cognitive science.
Notes
1 In the rest of this book, I use the expression absent the neo– prefix to refer to
the contemporary version of mechanism—as the mechanistic model of expla-
nation or simply mechanistic explanation.
2 Mutual manipulability and its application to mechanism is a controversial
topic (e.g., see Baumgartner & Gebharter, 2016). For the sake of clarity, I
draw on Glennan (2017, p. 44) and understand MM as an epistemic criterion
that serves as a useful instrument for delineating boundaries between relevant
and irrelevant components.
3 Contrary to the epistemic view, Craver (2007) endorses an ontic version of
mechanistic explanation, which acknowledges that “explanations are objec-
tive features of the world” (p. 27).
4 In mechanistic philosophy, the term phenomenal typically refers to natural phe-
nomena that are observable and describable from a third-person perspective.
References
Abraham, R. H., & Shaw, C. D. (1992). Dynamics: The geometry of behavior
(2nd ed.). Addison-Wesley Publishing Company.
Abramova, E., Slors, M. V. P., & Rooij, I. J. E. I. (2017). Enactive mechanistic
explanation of social cognition. In G. Gunzelmann, A. Howes, & T. Tenbrink
(Eds.), Proceedings of the 39th annual meeting of the cognitive science society
(pp. 45–50). Cognitive Science Society.
Andreasen, N. (1999). A unitary model of schizophrenia: Bleuler’s “fragmented
phrene” as schizencephaly. Archives of General Psychiatry, 56, 781–787.
Baddeley, A. (2007). Working memory, thought, and action. Oxford University
Press.
Bangu, S. (2017). Scientific explanation and understanding: Unificationism recon-
sidered. European Journal for Philosophy of Science, 7(1), 103–126.
Barker, M. J. (2019). Eliminative pluralism and integrative alternatives: The case
of SPECIES. The British Journal for the Philosophy of Science, 70(3), 657–681.
Barrett, D. (2014). Functional analysis and mechanistic explanation. Synthese,
191(12), 2695–2714.
Baumgartner, M., & Gebharter, A. (2016). Constitutive relevance, mutual manip-
ulability, and fat-handedness. The British Journal for the Philosophy of Science,
67(3), 731–756.
Bechtel, W. (1998). Representations and cognitive explanations: Assessing the
dynamicist’s challenge in cognitive science. Cognitive Science, 22(3), 295–317.
neuroscience. Taylor & Francis Group.
Bechtel, W. (2009a). Explanation: Mechanism, modularity, and situated cogni-
tion. In P. Robbins & M. Aydede (Eds.), The Cambridge handbook of situated
cognition (pp. 155–170). Cambridge University Press.
Bechtel, W. (2009b). Looking down, around, and up: Mechanistic expla-
nation in psychology. Philosophical Psychology, 22(5), 543–564.
doi:10.1080/09515080903238948
Bechtel, W. (2019). Resituating cognitive mechanisms within heterarchical net-
works controlling physiology and behavior. Theory & Psychology, 29(5), 620–
639. doi:10.1177/0959354319873725
Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative.
Studies in History and Philosophy of Science Part C: Studies in History and
Philosophy of Biological and Biomedical Sciences, 36(2), 421–441.
Bechtel, W., & Richardson, R. C. (2010). Discovering complexity: Decomposition
and localization as strategies in scientific research. MIT Press.
Bechtel, W., & Wright, C. (2009). What is psychological explanation? In P. Calvo
& J. Symons (Eds.), Routledge companion to philosophy of psychology (pp.
113–130). Routledge.
Beer, R. D. (2003). The Dynamics of active categorical perception in an evolved
model agent. Adaptive Behavior, 11(4), 209–243.
Bickle, J. (2003). Philosophy of neuroscience: A ruthlessly reductive approach.
Kluwer Academic Publishers.
Block, N. (1980). What is functionalism? In N. Block (Ed.), Readings in philoso-
phy of psychology (Vol. 1, pp. 171–184). Harvard University Press.
Bermúdez, J. L. (2000). Personal and sub-personal; A difference without a distinc-
tion. Philosophical Explorations, 3(1), 63–82.
Campbell, J. (2002). Rationality, meaning and the analysis of delusion. Philosophy,
Psychiatry and Psychology, 8, 89–100.
Chemero, A. (2000). Anti-representationalism and the dynamical stance.
Philosophy of Science, 67(4), 625–647. doi.org/10.1086/392858
Chemero, A. (2011). Radical embodied cognitive science. MIT Press.
Churchland, P. S. (1986). Neurophilosophy: Toward a unified science of the
mind-brain. MIT Press.
Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19.
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future
of cognitive science. Behavioral and Brain Sciences, 36(3), 181–204.
Colombo, M. (2013). Constitutive relevance and the personal/subpersonal dis-
tinction. Philosophical Psychology, 26(4), 547–570.
Colombo, M., & Heinz, A. (2019). Explanatory integration, computational phe-
notypes, and dimensional psychiatry: The case of alcohol use disorder. Theory
& Psychology, 29(5), 697–718.
Cosmides, L., & Tooby, J. (1995). From function to structure: The role of evo-
lutionary biology and computational theories in cognitive neuroscience. In M.
S. Gazzaniga (Ed.), The cognitive neurosciences (pp. 1199–1210). MIT Press.
741–765. doi:10.2307/2024640
Craver, C. F. (2006). When mechanistic models explain. Synthese, 153, 355–376.
Craver, C. F., & Darden, L. (2013). In search of mechanisms: Discoveries across
the life sciences. University of Chicago Press.
Craver, C. F., & Kaplan, D. M. (2020). Are more details better? On the norms
of completeness for mechanistic explanations. The British Journal for the
Philosophy of Science, 71(1), 287–319.
Darden, L., & Maull, N. (1977). Interfield theories. Philosophy of Science, 44(1),
43–64.
Dennett, D. C. (1969). Content and consciousness. Routledge and Kegan Paul.
Dennett, D. C. (1981). True believers: The intentional strategy and why it works.
In A. Heath (Ed.), Scientific explanation (pp. 53–75). Oxford University Press.
Dennett, D. C. (1991a). Consciousness explained. Little, Brown.
Dennett, D. C. (1991b). Real patterns. Journal of Philosophy, 88(1), 27–51.
Dennett, D. C. (2003). Who’s on first? Heterophenomenology explained. Journal
of Consciousness Studies, 10(9–10), 19–30.
Dennett, D. C. (2005). Sweet dreams: Philosophical obstacles to a science of con-
sciousness. MIT Press.
Drayson, Z. (2014). The personal/subpersonal distinction. Philosophy Compass,
9(5), 338–346.
Dupré, J. (1995). The disorder of things: Metaphysical foundations of the dis-
unity of science. Harvard University Press.
Eliasmith, C. (2001). Attractive and in-discrete: A critique of two putative virtues
of the dynamicist theory of mind. Minds and Machines, 11, 417–426.
Engel, G. (1977). The need for a new medical model: A challenge for biomedicine.
Science, 196(4286), 129–136.
Felleman, D. J., & van Essen, D. C. (1991). Distributed hierarchical processing in the
primate cerebral cortex. Cerebral Cortex, 1(1), 1–47. doi:10.1093/cercor/1.1.1
Fodor, J. (1965). Explanations in psychology. In M. Black (Ed.), Philosophy in
America (pp. 161–179). Cornell University Press.
Fodor, J. (1974). Special sciences (Or: The disunity of science as a working
hypothesis). Synthese, 28(2), 97–115.
Fodor, J. (1983). The modularity of mind. MIT Press.
Friston, K. J. (2010). The free-energy principle: A unified brain theory? Nature
Reviews Neuroscience, 11(2), 127–138.
Gallagher, S., & Brøsted Sørensen, J. (2006). Experimenting with phenomenol-
ogy. Consciousness and Cognition, 15(1), 119–134.
Routledge.
Glennan, S. (2017). The new mechanical philosophy. Oxford University Press.
Godfrey-Smith, P. (2014). Philosophy of biology. Princeton University Press.
Golonka, S., & Wilson, A. D. (2019). Ecological mechanisms in cognitive science.
Theory & Psychology, 29(5), 676–696.
Grobler, A., & Wiśniewski, A. (2005). Explanation and theory evaluation. In W.
R. Festa, A. Aliseda, & J. Peijnenburg (Eds.), Cognitive structures in scientific
inquiry: Poznan studies in the philosophy of the sciences and the humanities
(pp. 299–310). Brill | Rodopi.
Haken, H., Kelso, J. A. S., & Bunz, H. (1985). A theoretical model of phase transi-
tions in human hand movements. Biological Cybernetics, 51, 347–442.
Halina, M. (2017). Mechanistic explanation and its limits. In S. Glennan & P.
Illari (Eds.), The Routledge handbook of mechanisms and mechanical philoso-
phy (pp. 213–224). Taylor & Francis Group.
Hempel, C. G. (1965). Aspects of scientific explanation and other essays in the
philosophy of science. The Free Press.
Hempel, C. G., & Oppenheim, P. (1948). Studies in the logic of explanation.
Philosophy of Science, 15(2), 135–175.
Hensel, W. M. (2013). On reduction and interfield integration in neuroscience. In
M. Miłkowski & K. Talmont-Kamiński (Eds.), Regarding the mind, naturally:
Naturalist approaches to the sciences of the mental (pp. 167–181). Cambridge
Scholars Publishing.
Hodgkin, A. L., & Huxley, A. F. (1939). Action potentials recorded from inside a
nerve fibre. Nature, 144, 710–711.
Hohwy, J. (2013). The predictive mind. Oxford University Press.
Hurlburt, R. T., & Schwitzgebel, E. (2007). Describing inner experience?
Proponent meets skeptic. MIT Press.
Husserl, E. (1969). Formal and transcendental logic (D. Cairns, Trans.). Martinus
Nijhoff.
Nijhoff.
Husserl, E. (2001). Analyses concerning passive and active synthesis: Lectures on
transcendental logic (A. J. Steinbock, Trans.). Springer.
Hutchins, E. (1995). How a cockpit remembers its speeds. Cognitive Science, 19,
265–288.
Jack, A. I., & Roepstorff, A. (2002). Introspection and cognitive brain mapping:
From stimulus–response to script–report. Trends in Cognitive Sciences, 6(8),
333–339.
Kaplan, D. M. (2017). Integrating mind and brain science: A field guide. In D.
M. Kaplan (Ed.), Explanation and integration in mind and brain science (pp.
1–28). Oxford University Press.
Kaplan, D. M., & Craver, C. F. (2011). The explanatory force of dynamical and
mathematical models in neuroscience: A mechanistic perspective. Philosophy
of Science, 78(4), 601–627.
Kelso, J. A. S. (1995). Dynamic patterns: The self-organization of brain and
behavior. MIT Press.
Kendler, K., Zachar, P., & Craver, C. (2011). What kinds of things are psychi-
atric disorders? Psychological Medicine, 41(6), 1143–1150. doi:10.1017/
S0033291710001844
Komorowska-Mach, J. (2019). Introspection – one or more? Pluralism about self-
knowledge. Filozofia Nauki, 27(1), 5–25. doi.org/10.14394/filnau.2019.0001
Lakatos, I. (1980). The methodology of scientific research programmes.
Cambridge University Press.
Lingis, A. (1986). Phenomenological explanations. Martinus Nijhoff Publishers.
Lipton, P. (2009). Understanding without explanation. In H. W. de Regt, S.
Leonelli, & K. Eigner (Eds.), Scientific understanding: Philosophical perspec-
tives (pp. 43–63). University of Pittsburgh Press.
McDowell, J. (1986). Singular thought and the extent of ‘inner space.’ In J.
McDowell & P. Pettit (Eds.), Subject, thought, and context (pp. 137–168).
McDowell, J. (1994). The content of perceptual experience. Philosophical
Quarterly, 44, 190–205.
Millikan, R. G. (2002). Biofunctions. In A. Ariew, R. Cummins, & M. Perlman
(Eds.), Functions: New essays in the philosophy of psychology and biology.
Miłkowski, M. (2016). Integrating cognitive (neuro)science using mechanisms.
AVANT. The Journal of the Philosophical-Interdisciplinary Vanguard, VII(2),
45–67.
Miłkowski, M. (2017). Mechanisms and the mental. In S. Glennan & P. Illari
(Eds.), The Routledge handbook of mechanisms and mechanical philosophy
(pp. 74–88). Taylor & Francis Group.
Miłkowski, M., Clowes, R., Rucińska, Z., Przegalińska, A., Zawidzki, T., Krueger,
J., Gies, A., McGann, M., Afeltowicz, Ł., Wachowski, W., Stjernberg, F.,
Loughlin, V., & Hohol, M. (2018). From wide cognition to mechanisms: A silent
revolution. Frontiers in Psychology, 9. doi.org/10.3389/fpsyg.2018.02393
Miłkowski, M., & Hohol, M. (2020). Explanations in cognitive science:
Unification versus pluralism. Synthese. Advanced online publication. doi.
org/10.1007/s11229-020-02777-y
Mitchell, S. D. (2002). Integrative pluralism. Biology and Philosophy, 17(1),
55–70.
Nagel, E. (1961). The structure of science: Problems in the logic of scientific
explanation. Brace & World.
Newell, A., & Simon, H. A. (1972). Human problem solving. Prentice Hall.
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal
reports on mental processes. Psychological Review, 84, 231–359.
Oppenheim, P., & Putnam, H. (1958). Unity of science as a working hypothesis.
Minnesota Studies in the Philosophy of Science, 2, 3–36.
Oullier, O., De Guzman, G. C., Jantzen, K. J., Lagarde, J., & Kelso, J. A. S.
(2008). Social coordination dynamics: Measuring human bonding. Social
Neuroscience, 3(2), 178–192.
Piccinini, G. (2003). Data from introspective reports: Upgrading from common
sense to science. Journal of Consciousness Studies, 10(9–10), 141–156.
Piccinini, G. (2010). How to improve on heterophenomenology: The self-mea-
surement methodology of first-person data. Journal of Consciousness Studies,
17(3–4), 84–106.
Piccinini, G., & Craver, C. (2011). Integrating psychology and neuroscience:
Functional analyses as mechanism sketches. Synthese, 183(3), 283–311.
Port, R. (2003). Meter and speech. Journal of Phonetics, 31, 599–611.
Prinz, J. (2012). The conscious brain. Oxford University Press.
Putnam, H. (1960). Minds and machines. In S. Hook (Ed.), Dimensions of mind
(pp. 138–164). New York University Press.
Raja, V., Biener, Z., & Chemero, A. (2017). From Kepler to Gibson. Ecological
Psychology, 29(2), 146–160.
Rodemeyer, L. (2020). Levels of embodiment: A Husserlian analysis of gender
and the development of eating disorders. In C. Tewes & G. Stanghellini (Eds.),
Time and body: Phenomenological and psychopathological approaches (pp.
234–255). Cambridge University Press. doi:10.1017/9781108776660.019
Rosenthal, D. (2005). Consciousness and mind. Clarendon Press.
Roth, M., & Cummins, R. (2017). Neuroscience, psychology, reduction, and
functional analysis. In D. M. Kaplan (Ed.), Explanation and integration in
mind and brain science (pp. 29–43). Oxford University Press.
Roux, S. (2017). From the mechanical philosophy to early modern mechanisms.
In S. Glennan & P. Illari (Eds.), The Routledge handbook of mechanisms and
mechanical philosophy (pp. 26–45). Taylor & Francis Group.
Ruiz-Mirazo, K., & Moreno, A. (2004). Basic autonomy as a fundamental step in
the synthesis of life. Artificial Life, 10(3), 235–259.
Salmon, W. C. (1984). Scientific explanation and the causal structure of the world.
Princeton University Press.
Salmon, W. C. (1989). Four decades of scientific explanation. University of
Pittsburgh Press.
Sass, L. A. (2014). Explanation and description in phenomenological psychopa-
thology. Journal of Psychopathology, 20, 366–376.
Schacter, D. L. (2012). Constructive memory: Past and future. Dialogues in
Clinical Neuroscience, 14(1), 7–18.
Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral hip-
pocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 20,
11–21.
Scriven, M. (1975). Causation as explanation. Nous, 9, 3–16.
Schwitzgebel, E. (2012). Introspection, what? In D. Smithies & D. Stojlar (Eds.),
Introspection and consciousness (pp. 29–48). Oxford University Press.
Simon, H. A. (1962). The architecture of complexity. Proceedings of the American
Philosophical Society, 106(6), 467–482.
Smith, J. (2010). Seeing other people. Philosophy and Phenomenological
Research, 81(3), 731–748.
Smith, L. B., & Thelen, E. (2003). Development as a dynamic system. Trends in
Sober, E. (1983). Equilibrium explanation. Philosophical Studies, 43(2), 201–210.
Sowa, R. (2012). Eidetics and its methodology (E. A. Behnke, Trans.). In S. Luft
& S. Overgaard (Eds.), The Routledge companion to phenomenology (pp.
Stich, S., & Nichols, S. (2003). Folk psychology. In S. Stich & T. A. Warfield (Eds.),
The Blackwell guide to philosophy of mind (pp. 235–255). Blackwell Publishing.
Walmsley, J. (2008). Explanation in dynamical cognitive science. Minds and
Machines, 18(3), 331–348. doi:10.1007/s11023-008-9103-9
Watkins, J. W. (1957). Historical explanation in the social sciences. The British
Journal for the Philosophy of Science, 8(30), 104–117.
Weiskopf, D. A. (2011). Models and mechanisms in psychological explanation.
Synthese, 183, 313–338. doi:10.1007/s11229-011-9958-9
Wheeler, M. (2013). Science friction: Phenomenology, naturalism and cognitive
science. Royal Institute of Philosophy Supplement, 72, 135–167.
Williams, H. (2020). Husserl on personal level explanation. Human Studies, 43,
1–22. doi:10.1007/s10746-020-09537-4
Wimsatt, W. C. (1986). Forms of aggregativity. In A. Donagan, A. N. Perovich, &
M. V. Wedin (Eds.), Human nature and natural knowledge: Essays presented
to Marjorie Grene on the occasion of her seventy-fifth birthday (pp. 259–291).
Springer.
Woodward, J. (2003). Making things happen: A theory of causal explanation.
Woodward, J. (2017). Explanation in neurobiology. In D. M. Kaplan (Ed.),
Explanation and integration in mind and brain science (pp. 70–100). Oxford
University Press.
Wright, L. (1976). Teleological explanations. University of California Press.
Explorations in the dynamics of cognition. MIT Press.
van Gelder, T. (1998). The dynamical hypothesis in cognitive science. Behavioral
and Brain Sciences, 21(5), 615–628.
Ylikoski, P. K. (2017). Social mechanisms. In S. Glennan & P. Illari (Eds.), The
Routledge handbook of mechanisms and mechanical philosophy (pp. 401–
412). Taylor & Francis Group.
Zahavi, D. (2007). Killing the straw man: Dennett and phenomenology.
Phenomenology and the Cognitive Sciences, 6(1–2), 21–43.
Zednik, C. (2011). The nature of dynamical explanation. Philosophy of Science,
78(2), 238–263. doi:10.1086/659221
Part II
Phenomenology and
Mechanism
In Search of Constraints
4 Phenomenology and Functionalism
4.1 Introduction1
Functionalism is recognized as one of the most influential approaches to
mind and cognition—one usually perceived as incongruent with phenom-
enological approaches descending from the Husserlian tradition. I think
this is misleading, and there are arguments for connections and similari-
ties between these traditions. In philosophy of mind, functionalism may
be approached from a number of different perspectives (for an overview,
see, e.g., Block, 1980). First, it may be approached as a metaphysical
theory of mental states, which argues that to be a mental state is to be
a functional state individuated by causal relations to other mental states
and inputs and outputs. There are different types of such functionalist
theories of mind, including the most popular one: computation-repre-
sentation functionalism. The key idea behind this version of function-
alism is that mental states can be understood as Turing machine table
states (e.g., Putnam, 1960) and the mind as a sort of computer program.
In this fashion, Dreyfus (Dreyfus & Hall, 1982) and McIntyre (1986)
proposed computational-functionalist readings of Husserl. Second, func-
tionalism can be approached as an explanatory strategy called functional
analysis, which is applied in psychology and cognitive science. It con-
sists in decomposing a cognitive system into component processes and
capacities and describing its functional organization (Cummins, 1975;
Fodor, 1968). Functional analysis may differ depending on what type of
functionalism one is committed to, and it does not have to necessarily
entail computationalism (Piccinini, 2010). A proponent of computation-
representation functionalism will decompose a target system into a set
of specialized computational modules processing input information and
returning outputs and will produce an explanation in terms of program
execution and data manipulation. But Cummins-style functional analy-
sis, which relies on a more general systemic notion of function as a causal
role, will decompose the system into a set of sub-capacities that contrib-
ute to the system’s functioning. These sub-capacities are then described in
dispositional-behavioral terms, not in computational terms.
DOI: 10.4324/9781003035367-7
140 Phenomenology and Mechanism
In this chapter, I propose a functionalist reading of Husserlian phe-
nomenology and argue that such a reading opens a way toward inte-
grating phenomenology with cognitive science, in particular with the
mechanistic explanatory framework. First, I critically discuss a func-
tional-computational interpretation of Husserl coined in the discussion
between Dreyfus and McIntyre. I outline the deficiencies of Dreyfus’s and
McIntyre’s proposals, one of which is not addressing the notion of func-
tion in Husserl. Next, I propose shifting perspective and approaching the
relation between phenomenology and functionalism from a methodolog-
ical perspective. Methodological similarities were already indicated by
McIntyre (1986), and more recently, Paul Livingston (2005) has argued
that there is continuity between the phenomenological and functional
methods of conceptual and logical analysis of our experience. My argu-
ment is different. The key claim is that in Husserlian “functional phe-
nomenology,” as introduced in the first book of Ideas (Husserl, 1982),
we find an original notion of intentional function as well as a method
of decomposition that can be understood as analogous to the explana-
tory strategy of functional analysis. Finally, I argue that the proposed
functionalist reading of phenomenology opens a new approach to the
integration of phenomenology with cognitive sciences. Accordingly, phe-
nomenology provides constraints concerning the functional structure of
the phenomenon in question that can be applied to explanatory models.
I demonstrate this on two examples: vision studies and the evolutionary-
developmental account of first-person perspective.
4.2 Husserlian Phenomenology and Computational

Functionalism
One of the most important insights of phenomenological analysis is that
conscious experience is intentional; that is, it refers to something; it is
about something. In contemporary philosophy of mind, the intentional
property of mental states is often understood in terms of mental repre-
sentation. In short, a mental state is intentional if and only if it has rep-
resentational content. The representational interpretation of Husserlian
phenomenology arose in the debate between Dreyfus (1988; Dreyfus
& Hall, 1982) and McIntyre (1986). Three points are especially impor-
tant in this representational interpretation of Husserl. First, according
to Dreyfus, Husserl’s theory of noemata is a predecessor of contempo-
rary computationalism, such as Fodor’s representational theory of mind
(RTM; e.g., Fodor, 1975). McIntyre disagrees with such a strong compu-
tationalist reading, but he acknowledges that there are striking points of
agreement between Husserlian phenomenology and contemporary rep-
resentationalism. Second, the key similarity is that cognition has a medi-
ated character—in Fodor by mental representations and in Husserl by
noemata. Third, in both approaches, the theory of mental representations
Phenomenology and Functionalism 141
and their role in cognitive processes do not require reference to a mind’s
physical realization. They are, so to speak, “ontologically neutral.”
Arguing that Husserl was a precursor of a sort of representational
theory of mind requires the introduction and clarification of some phe-
nomenological terminology including such notions as noema, noematic
sense, and noesis. First, it needs to be noted that Husserl used the term
representation in a different sense than contemporary philosophers of
mind and cognitive psychologists do. For Husserl, the notion of repre-
sentation referred to something that is re-presented. Thus, he used the
term in analyses of acts of memory or imagination, which literally re-
present an object (the German terms Husserl used were Darstellung or
Vergegenwärtigung). In acts of perception, on the contrary, an object is
presented to consciousness, not represented. The German term for this
mode of apprehension is Gegenwärtigung. This, however, does not mean
that Husserl argues for a version of naïve epistemic realism. On the con-
trary, the aim of phenomenology is to break down this naivety and show
how the apprehension of an object is possible in virtue of cognitive struc-
tures. A representational reading of Husserlian phenomenology empha-
sizes that the theory acknowledges a cognitive structure that plays an
analogous role to contemporary mental representation, that is, the role of
mediation between consciousness and objects, namely, the noema.
As I showed in Chapter 1, Husserl introduced the notion of noema in
the first book of Ideas (1982, especially para. 87–127). In short, noema
is the intentional correlate of noetic functions of consciousness, in virtue
of which consciousness refers to objects. Both noesis and noema can be
distinguished and analyzed only by applying phenomenological reduc-
tion and attaining the phenomenological attitude. In his interpretation
of noema, Dreyfus draws on the seminal paper by Dagfinn Føllesdal
(1969) titled “Husserl’s Notion of Noema,” in which Føllesdal argues
that noema should be understood as an abstract unit of sense, that is, as
“an intensional entity, a generalization of the notion of meaning (Sinn,
Bedeutung)” (p. 681). According to Føllesdal (1969), “the noematic Sinn
is that in virtue of which consciousness relates to the object” (p. 682).
Therefore, noema is ontologically different from the object to which it
refers; it is the conceptual structure through which consciousness can
apprehend objects. Another important claim is that noemata are abstract;
that is, they are pure meanings that cannot be perceived as such. Noema
is also not a kind of “internal image” of an intended thing (this was
already ruled out by Husserl in his critique of internal image theory; see
Husserl, 1982, para. 90). In short, according to this reading, noema is
an abstract meaning that mediates between consciousness and external
objects and in virtue of which cognitive acts have reference.2
Understanding noemata as abstract structures mediating cognition is
crucial for a representational reading of phenomenology. McIntyre (1986)
writes: “Husserl’s noematic Sinne can be seen—up to a point—as a version
of what Fodor calls ‘mental representations,’ having both formal (or ‘syn-
tactic’) and representational (or ‘semantic’) properties and so forming a
kind of ‘language of thought’” (p. 101). But Fodor also believes, at least
in his early writings, that the semantic properties of representations can
be reduced to their syntax. Here lies the bone of the contention between
Dreyfus and McIntyre. Dreyfus claims that for Husserl, noemata play a
mediating role, but he adds that they are “formal structures,” that is, “strict
rules for possible syntheses” of simple components (Dreyfus & Hall, 1982,
pp. 10–11). If that is the case, then noemata can be considered a result of
syntactical operations. In short, noemata are computed similarly to mental
representations, and Husserl can be thought to be a formalist who believed
in “a mathesis of experience” (Dreyfus, 1988, p. 101) and anticipated the
computational approach to the mind. McIntyre (1986) disagrees on this
point and argues that although Husserl acknowledges that noemata have
both semantic and syntactic properties, it does not follow that the former
are reducible to the latter (p. 111). McIntyre is right, and as we will see
later, there are more arguments to the effect that Husserl’s thinking about
mental operations as mathematical was only a loose analogy.
The last point in a representational-functionalist reading of Husserl
is “ontological neutrality.” Generally speaking, functionalism does not
answer the ontological question of what there is. Functionalism answers
only the metaphysical question of how mental states are individuated,
namely, by their function, and functional states are defined by their causal
relations to one another and inputs and outputs (Block, 2007, p. 19).
However, causality is understood in abstraction here; that is, they are
not any specific causal relations, such as those between neurons. Thus,
functionalists argue that there can be multiple physical realizations of the
same functional state.
McIntyre sees opponents of naturalistic psychology in both Husserl
and Fodor. Up until the 1980s, Fodor (1980) defended a position called
“methodological solipsism,” which states that in pursuing a theory of
mind, we should not presuppose anything about the natural setting of
the mind or any specific causal relations between mental and physical
states. According to McIntyre (1986), “like the functionalists and the
computationalists, then, Husserl seeks abstract accounts that would cap-
ture what is common to various mental capacities, no matter how dif-
ferent in their natural make-up the entities having these capacities may
be” (p. 104). In other words, functionalists such as Fodor and phenom-
enologists such as Husserl believe that a theory of mental capacities is
independent of their physical realization. Mental states understood as
functional states can be physically realized in multiple ways; therefore,
to explain the mental functioning of a system, we do not have to refer
to its physical realization. Thus, the theory of mind, which, according to
Fodor, is a special science, is autonomous from natural sciences such as
biology or physics.
As I argued in Chapter 1, Husserl (e.g., 1964, 2002) was also critical of
reductive naturalistic explanations of consciousness because he believed they
were based on an erroneous application of the methods of natural science
to explain conscious phenomena. Phenomenology, as designed by Husserl,
is thought to fill this gap by delivering a method of investigating conscious-
ness and, ultimately, an autonomous and strict science of consciousness,
that is, a science of the general structures of consciousness. According to
McIntyre, phenomenological reduction (understood by him as transcenden-
tal reduction) plays a similar role to the principle of methodological solip-
sism. Phenomenological reduction, as a suspension of common beliefs about
physical nature and mental phenomena, leads us to a change in attitude from
natural to phenomenological. After reduction, our descriptions of mental
activities are, in principle, restricted to what is actually experienced. Such
purified descriptions of experience are then analyzed in order to discover the
structure of a specific experience. Noemata, here understood as a sort of men-
tal representation, can be disclosed only after reduction, and creating a formal
theory of noemata does not require reference to brain activity and so on.
Now, we need to consider the credibility of such a representational
reading. Next to the previously mentioned similarities, there are also dif-
ferences, some of which might be decisive for the plausibility of such a
functionalist reading of Husserl. McIntyre himself indicates several impor-
tant differences. For example, he acknowledges Husserl’s transcendental
phenomenology to be in opposition to the psychological level of Fodor’s
special science of the mind. However, whether this tension is genuine or
merely apparent is highly dependent on our reading of Husserl and his
so-called transcendental turn (see, e.g., Zahavi, 2003). As I argued in
previous chapters, transcendental phenomenology is not the only face
of Husserl’s phenomenological project, and, furthermore, transcenden-
tal phenomenology is not the only way to study consciousness (even if
for some phenomenologists it is the most important one). Husserl was
aware of that and, although critical of the naturalization of consciousness,
acknowledged the importance of experimental psychology. In the lectures
titled Phenomenological Psychology, Husserl (1977) shows how to elevate
psychology to a phenomenological level and introduces the idea of eidetic
psychology, which was to be a discipline mediating between transcen-
dental phenomenology and experimental psychology (Reynolds, 2017;
Thinès, 1977). As I argued in Chapter 1, this eidetic or phenomenological
psychology is thought to provide experimental psychology with “scientific
concepts of internality” (Husserl, 1977, p. 166) that is, concepts such as
intentionality, noesis, and noema, among others. Let me also remind the
reader that Husserl (1977, pp. 33–34) characterizes phenomenological
psychology as a priori, eidetic, grounded in intuition or pure description,
focusing on intentionality. Although phenomenological psychology applies
the reductive method, it is different from the transcendental reduction of
transcendental phenomenology (see Kockelmans, 1987). Husserl (1971)
calls it “eidetic reduction” or “phenomenological reduction” (1977), and
this type of reduction opens the possibility to study structures of experi-
ence without transcendental commitments. When we acknowledge this
weak concept of phenomenology, phenomenology appears to be comple-
mentary to psychology rather than opposed to it.
Neither Dreyfus nor McIntyre acknowledges the distinction between
transcendental phenomenology and phenomenological psychology and
the two related types of reduction; thus, they stress the opposition between
the psychological level and the transcendental level. But McIntyre is also
skeptical about the consistency of the phenomenological eidetic approach
in general with contemporary cognitive science. He claims that Husserl
takes this phenomenological reflection to be indubitably reliable,

and the pronouncements issuing from it are not supposed to be mere
speculative or inductive generalizations but necessary or ‘eidetic’
truths about consciousness. Claims such as these mark radical dif-
ferences between the methods Husserl characterizes as uniquely phe-
nomenological and those employed by contemporary cognitivists.
(McIntyre, 1986, p. 103)
This seems like a superficial reading of Husserl. The aim of Husserlian

phenomenology was to search for certain and apodictic truths about con-
sciousness, but they are not evident and clear from the beginning. Our
access to the structural features of experiences comes in degrees of clarity,
and the task of phenomenology is to develop the method and elaborate
experiential insights (Husserl, 1982, para. 67). For Husserl, such apodic-
tic knowledge is rather an ideal of self-knowledge beyond our capacities,
and building a phenomenological science of subjectivity is an infinite task
(see, e.g., Husserl, 1989b).
Phenomenology is also not infallible. Already Føllesdal (1988) noticed
that Husserl offers an account of the corrigibility of phenomenologi-
cal observations and statements. Indeed, in Formal and Transcendental
Logic, Husserl (1969) admits that “the possibility of deception is inherent
in the evidence of experience” (p. 156). As I argued in previous chapters,
phenomenological eidetic claims can be falsified in at least three ways.
The first is through intersubjective corroboration (Gallagher & Zahavi,
2012, pp. 30–31), that is, a phenomenological description of an experi-
ence of a specific type can be compared with others’ descriptions of the
same type of experience. Second, as Sowa (2012) argues, phenomeno-
logical eidetic claims can be falsified in the process of imaginary varia-
tion, that is, by construing in phantasy possible variants of the object of
experience. Third, phenomenological descriptions are, whether we like it
or not, confronted with the results of scientific research. For some phe-
nomenologists, scientific experiments may be sufficient evidence to falsify
phenomenological claims (e.g., Gallagher & Brøsted Sørensen, 2006).
To sum up, both Dreyfus and McIntyre agree that there are striking
similarities between Husserlian phenomenology and representational
theory of mind as proposed by Fodor. They disagree, however, over the
extent of the similarities. Whereas Dreyfus writes that Husserl’s theory
of intentionality “corresponds exactly” (Dreyfus & Hall, 1982, p. 3) to
Fodor’s RTM, McIntyre is much more reserved and stresses essential dif-
ferences. I agree with McIntyre that Husserl was not a formalist regard-
ing mental representations and that a strong computationalist reading
of his phenomenology is implausible. Both McIntyre and Dreyfus also
fail to recognize that there is a further level below the phenomenologi-
cal transcendental level, that of the eidetic analysis of phenomenological
psychology. This distinction is crucial for the functionalist reading of phe-
nomenology and the possibility of integrating it with cognitive science.
But there is another issue that McIntyre and Dreyfus do not address, and
it is perhaps the most important one, which every functionalist reading of
Husserl should face, concerning the difference in how the notion of func-
tion is understood between Husserl and contemporary functionalists.
4.3 The Notion of Function and the Idea of Functional

Phenomenology in Husserl
In this and the following section, I develop an alternative functionalist
reading of Husserlian phenomenology. I take functionalism primarily
as an explanatory strategy. The notion of function is at the center of
this approach. My argument is twofold: first, that Husserl used func-
tional terminology and that he introduced an original notion of inten-
tional function responsible for the production of intentional experience.
Phenomenology aimed at investigating intentional functions and, more
generally, any processes that play a role in constituting experience is
referred to as functional phenomenology by Husserl. Second, I argue
that functional phenomenology shares similarities with the functional-
ist explanatory strategy known as functional analysis, which relies on
decomposing the system’s capacity (explanandum) into a set of functions
(explanans) responsible for the capacity.
In functionalism, the function of a system’s component is usually under-
stood as the causal role it plays in the system. For example, a mental state
of a belief type, for example, “This cake looks delicious,” is in causal rela-
tion with other mental states, say, a desire (“I want to eat that cake”), and
contributes to behavior (eating the cake), assuming that there are no other
conflicting states. Causal-role functionalists believe that identifying men-
tal states functionally will allow for formulating lawlike generalizations
about interactions between them. But as McIntyre (1986) rightly states,
Husserl cannot himself be a functionalist of the standard ‘causal-role’

sort, i.e., he cannot explicate mental states in terms of their causal
relations to one another and to the world, for causality (in any natu-
ralistic sense) is ‘bracketed’ by phenomenological epoché.
(p. 104)
Physical causality is not something phenomenology investigates. As I

showed in Chapter 1, Husserl introduces the notion of motivation to
explicate quasi-causal relations between mental states. In short, motiva-
tion is a specific type of mental causation (e.g., Husserl, 1977, p. 108)
that relates mental states of different sorts. For example, a perceptual
state can motivate a belief, which, in turn, can motivate a subject to
have a specific emotional state or to perform a particular cognitive act.
Importantly, motivational relations are weaker than physical causality.
There is no certainty that a mental state will always motivate the same
set of other mental states; thus, it is implausible to explicate motivational
relations in the form of strict laws.
We can now examine how Husserl used the concept of function. It
is important to notice that the notion of function appears in Husserl’s
works very early. Before his phenomenological period, in Philosophy of
Arithmetic (Husserl, 2012), he was especially interested in methods of
constructing numerical systems and discussed arithmetical operations
concerning both the numerical construction of complex expressions as
well as the reduction of complex expressions to normal form. According
to Stefania Centrone (2010), in this period, Husserl used the procedural
notion of function,3 that is, function as a numeric procedure, as a rule,
prescription, or an algorithm in a general sense. Such a numeric proce-
dure takes one or more arguments and returns a numeric product, that
is, a complex term.
In the first book of Ideas, Husserl proposes his own original notion of
intentional function understood as a mental operation of constitution,
that is, the production of an object experience, and introduces the idea of
functional phenomenology. In paragraph 86 of Ideas I, he writes:
Nonetheless, the greatest problems of all are the functional prob-

lems, or those of the “constitution of consciousness-objectivities.”
These problems concern the way in which noeses, e.g., with respect
to Nature, by animating stuff and combining it into manifold-unitary
continua and syntheses bring about consciousness of something such
that the Objective unity of the objectivity allows of being harmoni-
ously “made known,” “legitimated” and “rationally” determined.
…
In this sense “function” (in an entirely different sense in contrast to
the mathematical one) is something wholly unique, grounded in the
pure essence of noesis. Consciousness is precisely consciousness “of”
something; it is of its essence to bear in itself “sense,” so to speak,
the quintessence of “soul,” “spirit,” “reason.” Consciousness is not a
name for “psychological complexes,” for “contents” fused together,
for “bundles” or streams of “sensations” which, without sense in
themselves also cannot lend any “sense” to whatever mixture,
and he adds at the end of the paragraph:
Not only with regard to the difficulties which it arrives at, but also
with regard to the ranking of problems from the standpoint of the
idea of an absolute cognition, [pure hyletic phenomenology] obvi-
ously stands far below the noetic and functional phenomenology
(both of which, moreover, are properly not to be separated).
(Husserl, 1982, para. 86)
In this rich passage, we see that Husserl identifies functional problems

with problems of object constitution or, to use the language of early
Husserl, object construction. An object experience is a product of func-
tions Husserl called noeses and their noematic correlates. The “stuff”
here is understood as hyle or sense-data, which is formed by noetic func-
tions. Noetic functions take the data, on the one hand, and noematic rep-
resentations, on the other, and produce an object experience. The concept
of noetic function might seem similar, to some extent, to the procedural
understanding of function Husserl considered in the context of numeric
systems. If that were the case, then the computational reading of phenom-
enology would gain a new argument. However, Husserl explicitly states
that his concept of function is completely different than the mathematical
one. In the cited paragraph, we do not see any justification for why he
rejects the mathematical meaning of function in the phenomenological
domain, but there are at least two good reasons for doing so.
First, already in the Philosophy of Arithmetic, Husserl notices that cal-
culations operate on signs not on meanings, and therefore we can dis-
tinguish two separate but connected structures: a conceptual one and
a signitive one (Centrone, 2010, pp. 330–331). Arithmetical operations,
that is, computations, concern the latter—they are operations on signs,
which can be considered without reference to their meaning. They are
syntactical operations of putting together signs and constructing com-
plex terms or, conversely, deconstructing complex terms into simple ones.
But, as Husserl states in the cited paragraph, consciousness bears itself a
“sense,” meaning, or, in other words, semantics, and thus reducing men-
tal functions to arithmetic-syntactic functions is implausible.
Second, as I argued in Chapter 1, Husserl sees the limitations of using
a mathematical analogy in a phenomenological context. Phenomenology
is not a mathematics of consciousness because it is a different type of
exact science. According to Husserl (2001b), phenomenology is, simi-
larly to mathematics, an eidetic science, but contrary to mathematics,
which investigates “formal essences,” it investigates “material essences”
or “inexact essences” (Investigation III, para. 9). Thus, phenomenologi-
cal laws are of a different nature than the strict laws of mathematics. It
seems, therefore, that although the mathematical understanding of func-
tion was Husserl’s initial inspiration, it was ultimately abandoned in pur-
suit of phenomenology.
Interestingly, Husserl also seems to reject the psychological notion of
function understood here as the procedure of synthesizing sensations
into “psychological complexes.” Although in a footnote Husserl (1982)
remarks that such a notion of psychological function is used in Carl
Stumpf’s “hyletic phenomenology” (p. 210), it can also be attributed to
experimental psychology, which Husserl criticized. Experimental psy-
chology cannot explain intentionality and has to be grounded in phe-
nomenological psychology. As Husserl (1977) writes, “all uniting takes
place here by means of the functions of apprehension which make of
the sense-data appearances of” (p. 126). These are intentional functions
and their noematic correlates, which make an adumbration an adumbra-
tion of something. Functional phenomenology, which employs analysis
of intentional structures, can be conceived as part of the previously men-
tioned phenomenological eidetic psychology, which precedes empirical
psychology and provides it with a genuine analysis of intentionality.
To sum up, Husserl elaborated an original understanding of function
as the operation of constituting unitary intentional experience. He states
that “the point of view of function is the central one for phenomenol-
ogy” (Husserl, 1982, p. 208), and he calls functional phenomenology the
investigation of such intentional functions in their various forms. Husserl
is neither a causal-role functionalist nor a computational functionalist
in the strict sense. Functional phenomenology does not investigate how
particular sensations, or syntactic components, are “fused together” but,
rather, how representational contents emerge in experience and how
these contents are motivationally interrelated. In the following, I argue
that his functional phenomenology shares similarities with the explana-
tory strategy called functional analysis.
4.4 Functional Analysis and Phenomenological Decomposition

As I have shown in Chapter 3, functional analysis, in the broadest sense,
is a method of explaining a system’s capacities in functional terms. It
relies on the decomposition of the target system’s capacities into a set of
sub-capacities (functions) and is usually followed by a sketch of the sys-
tem’s functional architecture. There are different types of functional anal-
ysis depending on the type of functionalism (Piccinini & Craver, 2011):
functional analysis by internal states, boxology, and task analysis. A brief
recapitulation of these types of analysis is in order. In computational
functionalism, the preferred form of explanation is functional analysis by
internal states. Importantly, such internal states are typically characterized
in representational and computational terms. Boxology comes from the
works of Fodor (1983) and his idea of the mind’s modularity. The modu-
larity thesis encapsulates cognitive functions and subfunctions in special-
ized modules, which are a system’s components distinguished through
functional description. Finally, there is task analysis (Cummins, 1975,
2000). This type of functional analysis can be divided into three steps.
First, the explanandum phenomenon is identified. In the case of cognitive
science, it is usually some cognitive capacity of the target system. Next,
we define the capacity as a dispositional property; that is, the system’s
cognitive capacity φ is a complex dispositional property D. Accordingly,
the system disposition exhibits a lawlike regularity in its behavior. Next,
we decompose this dispositional property D into a set of sub-dispositions
d1, d2, …, dn. In other words, the occurrence of sub-dispositions amounts
to disposition D. The final step is representing the system’s target capac-
ity decomposed in functional analysis in the form of a functional design,
which reveals its functional organization and captures all the relevant
relations between functions and their results.
Now considering Husserl’s functional phenomenology, one could ask
whether it employs a similar approach to any of these types of func-
tional analysis. Boxology and analysis by internal states seem not to have
much in common with the phenomenological approach. These types of
functional analysis are closely related to the computational-functionalist
approach. But as I argued earlier, a strong computationalist reading of
Husserl seems implausible. If that is the case, then we should proceed to
considering the similarities and differences between phenomenological
analysis and Cummins-style functional analysis.
At first glance, the difference between phenomenology and functional
task analysis concerns the conceptualization of the target phenomenon,
which, in functional analysis, is a capacity understood as a dispositional
property of a system whereas, in phenomenology, it is a lived experi-
ence of a certain type. However, the situation is not so clear. In fact, he
occasionally also uses dispositional language when he discusses personal
or real psychic subjects. For example, Husserl (e.g., 1977) considers the
habitual and intellectual dispositions of a person or writes about “latent
capacities (dispositions)” (1989a, p. 264) of the ego that can be exhibited
under certain conditions. Importantly, such dispositions are on different
levels of the mental. As he states in the second book of Ideas,
in the psychic sphere, we find groups of dispositions as properties

of an inferior level and, built on them, though not in the manner of
mere ‘summation’ but instead in that of a ‘constitution,’ we find uni-
ties of property of a higher level as ones which consequently manifest
themselves, as unitary, in the properties of the inferior level and in
their experiential changes.
(Husserl, 1989a, pp. 130–131)
Functional phenomenology is in the business of describing the constitu-
tive relations and related functions of consciousness.
A more significant difference between phenomenology and functional
task analysis concerns the very understanding of function. In Husserl,
the central notion is intentional function, that is, functions responsible
for different sorts of intentional formations of experience. For Cummins
(1975), something has a function in the system if it contributes to the
system’s capacity. As we have seen, on a certain level of abstraction, these
approaches are not so far from each other. For example, a perceptual
apprehension of an object, which is a paradigmatic case of perceptual
intentionality for Husserl, can also be considered a capacity. Furthermore,
both approaches identify functions with relevancy for the target capacity,
so, for instance, following Husserl (1997), one can argue that functions
of bodily motility contribute to the constitution of the experience of spa-
tial orientation. The difference is found, however, in a deeper understand-
ing of this relevancy. For Cummins, it would be causal relevancy; for
Husserl, as I argued earlier, causal relations are replaced by motivational
relations, which are lawlike; however, they differ from strict causal laws.
Rather than focusing on one particular type of relevance, I propose to
instead rely on the more general notion of constitutive relevancy. A func-
tion is constitutively relevant to the target phenomenon (a capacity) if it
contributes to its occurrence. We abstract, however, from establishing the
specific nature of this relevancy.
The key similarity between functional analysis and functional phenom-
enology is that they both employ a decompositional explanatory strategy.
Functional phenomenology employs its own type of decompositional
strategy, that is, decomposition of unitary experience into its contents and
constitutive functions. On the one hand, it fragments and categorizes the
contents of experience; on the other, it identifies representational functions.
4.4.1 Phenomenological Decomposition
One of the key ideas in phenomenology is that in the naïve attitude we
encounter objects in the world which are, so to speak, ready-made. The
task of phenomenology is to break down this naivety and show how
objects of experience are constituted in the activity of consciousness.
Attaining the phenomenological attitude through the reduction is the
first step. What is next is the process of analyzing an experience in order
to extract its structures. Taking into account the idea of functional phe-
nomenology, I argue that it employs a method that relies on decomposing
an experience into its noematic contents and related functions involved
in the production of experience. I call this type of decomposition static,
since it is closely related to the idea of static phenomenology and later
developed into the genetic approach (see Chapter 1 for an introduction
to static and genetic phenomenology).
Static decomposition is, relatively speaking, the simplest and most
intuitive way to analyze experience, and it is already present in Husserl’s
early works, in which he employs the procedure of fragmenting objects
of experience. In Logical Investigations, he writes that “the division of a
whole into a plurality of mutually exclusive pieces we call a piecing or
fragmentation (Zerstückung) of the same” (Husserl, 2001b, p. 29). In the
third investigation, he develops a general mereological theory of wholes
and parts, and specifically, he discusses piecing experiences, that is, phe-
nomenological contents.
There is no space here for a detailed discussion of this complex mereo-
logical theory, so I mention only its main ideas. The key distinction intro-
duced at the beginning of the investigation is the distinction between
independent and “non-independent” parts. A part is anything that can
be distinguished in an object (Husserl, 2001b, pp. 5, 29). An independent
part can also be called a piece (or portion), since it can be a separate
object of analysis. A dependent part is tied to other contents and cannot
be separated in reality, only in abstraction. Sometimes Husserl also calls a
dependent part a moment or an abstract part. Moreover, each part can be
composed of other parts; thus, it can also be decomposed; the whole pro-
cedure of decomposition is iterative and assumes the existence of differ-
ent levels of decomposition. Husserl (2001b) also introduces the relation
of foundation: “a content of the species A is founded upon a content of
the species B, if an A can by its essence (i.e., legally, in virtue of its specific
nature) not exist, unless a B also exists” (p. 34). The relation of founda-
tion is important because it expresses the necessary (a priori necessity
based on “material essences”) relations between parts and wholes.
Let us consider an example: perceiving an apple tree. First, we see the
tree’s shape, the colors of its trunk, leaves, and fruits. When we touch
its bark, we experience shapes, texture, and temperature. We can also
sense the smell of the fruits and hear the leaves rustle in the wind. All
these sensory contents are parts of the unitary experience of an apple
tree. Whereas some of them are independent parts and can be analyzed
as separate pieces, others are nonindependent (or abstract parts) because
they need to be analyzed in reference to one another. For instance, we can
consider the shape of a leaf separately, but a color cannot exist without a
shape, and the tactile experience of texture, without extension. When we
move around the object, the shape, the size of the parts, and the colors
vary—these are partial moments of the perceptual experience. The whole
experience of the tree can be thus divided into different aspects contain-
ing different sensory, as well as representational, contents.4
Importantly, this decomposition can be accompanied by another phe-
nomenological method, namely, imaginary variation, in which the con-
tents of experience are varied in order to apprehend the experience’s
invariant structure. After the procedure of variation, the studied experi-
ence is no longer seen as a particular experience of an individual subject
but as a type of experience, for example, a type of perceptual experience.
What is especially interesting for us here is that the decomposition of
contents is supplemented in Husserl with the decomposition of noetic
functions. Units of meaning, or noemata in Husserlian terminology, iden-
tified in the decomposition of contents are related to noetic functions.
Thanks to noetic functions, sensory contents are correlated with con-
ceptual (noematic) contents, which can then be expressed in the form of
judgments. To put it differently, noetic functions play the role of repre-
sentational functions, which give meaning to sense-data by correlating it
with mental representations.
Husserl, however, was not satisfied with static decomposition and saw
that the simple dichotomy of sensory matter and noetic form did not
cover the whole process of the production of experience. He recognized
that the level of intentional functions related to the ego’s activity is just
one among several levels of constitution, including the level of passive
synthesis. As I argued in previous chapters, in order to explain the active,
as well as passive, synthesis of experience, Husserl develops the idea of
genetic phenomenology, which aims to explain the genesis of the very pro-
cess of constitution. The active level of constitution concerns all processes
related to all cognitive activities accessible to the subject’s awareness, for
example, producing judgments. Noemata understood as linguistic mean-
ings, or as mental representations, are part of this active synthesis. By the
passive level of constitution, Husserl understands processes that influence
cognitive acts and their contents but are not explicitly aware. An exam-
ple of such passive syntheses are different sorts of associations related
to affectivity and kinesthesis as well as previous experiences (see, e.g.,
Welton, 1982).
According to Husserl (1973), the static approach, which describes the
constitution of an intentional object as “the product of an objectivating
operation of the ego” (p. 72), provides “an understanding of intentional
accomplishment” and “reveals to us the graduated levels of intentional
objects that emerge in founded apperceptions of a higher level as objec-
tive senses and in functions of sense giving, and it reveals to us how they
function in them, etc.” (2001a, p. 629). But the static approach should
be supplemented by a genetic investigation that explains the passive pro-
cesses accompanying and influencing the acts of the ego. Genetic phe-
nomenology discloses that under noetic functions, there are motivational
and associative functions that also contribute to our experience.
To see how the genetic approach contributes to phenomenological
decomposition, recall what was said in Chapter 1 concerning Husserl’s
model of time-consciousness. The cognitive capacity to produce the expe-
rience of an object that extends in time can be divided into three subfunc-
tions: retention (retaining in consciousness parts of the object that are
no longer present), protention (anticipating the parts of the object not
yet present), and the primal- or “ur-impression” (sensory receptivity in
the present moment). According to Husserl, we do not experience the
momentary “now” of a temporal object as sensorially stimulating (the
ur-impression is, so to speak, a product of idealization). On the con-
trary, the lived experience of “now” is a product of all these functions,
which means that the sensory core of “now” is always accompanied
by retentional-protentional formations. Husserl also divides the reten-
tional function into two subfunctions: so-called longitudinal retention
and transverse retention. The former is directed toward the “how” of the
stream of consciousness, namely, into succeeding phases synthesized into
a non-objectified continuity. The latter is directed toward what appears
in the flow, that is, a temporal object. As I mentioned in Chapter 1, the
static model was revised by Husserl in the development of genetic phe-
nomenology (see Kortooms, 2002, pp. 175–284). For example, different
functions of affectivity and embodiment, such as kinesthesis, are included
in the genetic model of time-consciousness (Pokropski, 2015). In this
way, Husserl demonstrated that “the functions of corporeality belong to
the functions of the passive pre-constitution” (Landgrebe, 1981, p. 56).
Associative functions were another addition to the model, that is, asso-
ciations with past experiences influencing the occurrent experience. Thus,
the retentional function as described earlier was called “near retention,”
because it is involved with the production of our experience of the living
present; it retains the fading phases of perception, for example, the phases
of visual adumbrations of perceptual objects—and for the associative
process, Husserl introduced a new term: “far retention” (see Rodemeyer,
2006, pp. 88–90). Far retention is not limited to a current experience and
reaches back to past experiences. It is a form of passive association con-
cerning a given perceived object. Far retention plays an important role in
object recognition; that is, when we recognize an object, its meaning is
enriched by associations with previous experiences of that object.
To sum up, Husserlian phenomenology applies an explanatory strategy
of decomposition that shares similarities with functional decomposition
as applied in functionalism. Phenomenological decomposition has two
main levels: the decomposition of experiential contents and related func-
tions. But as Husserl argues, this is merely a descriptive method, and
it should be supplemented with the explanatory genetic approach. The
genetic approach decomposes the passive processes (such as affectivity,
embodiment, associations) accompanying the activities of the ego.
4.5 Toward Functional-Mechanistic Naturalization

If I am right and functional phenomenology can be read as similar to
the functionalist explanatory strategy, it is worth considering the con-
sequences of such a claim for the naturalization of phenomenology.
Functionalist naturalization is addressed concisely in the editors’ intro-
duction to the volume Naturalizing Phenomenology as a plausible way
of non-reductively integrating phenomenology with cognitive science
(Roy et al., 1999, pp. 71–72). Two ways of applying functionalist nat-
uralization to Husserlian phenomenology are mentioned. The first one
consists in developing Husserlian terminology in the direction of func-
tional computationalism. Such interpretations of Husserl were proposed,
as we have seen, by Dreyfus and McIntyre. Is the representational reading
of phenomenology a plausible approach to naturalization, as Roy and
his colleagues (1999) hypothesize? McIntyre (1986) claims that “neither
Fodor nor Husserl—neither cognitive science nor transcendental phe-
nomenology—claims to offer a naturalistic theory about how mental
processing actually takes place in human minds or brains” (p. 103). The
reason for that is that both positions acknowledge “ontological neutral-
ity.” Furthermore, as McIntyre rightly argues, the strong computational-
ist reading of Husserl is incorrect, because the semantic properties of
intentional contents are irreducible to syntactic ones. It seems therefore
that another weaker notion of representation is needed to pursue this
approach to naturalization.
The second path toward functionalist naturalization, which I endorse,
is “based on the idea that Husserlian phenomenological descriptions have
an intrinsically functional dimension, although not of a computational and
causal nature” (Roy et al., 1999, p. 72). The editors of the volume do not
develop this idea further and quickly conclude that “the problem is, therefore,
to estimate to what extent such a functional dimension of phenomenology
can really open the doors to an integration of phenomenological descriptions
into a naturalist framework” (Roy et al., 1999, p. 72). In this section, I argue
that the functional reading of phenomenology proposed earlier reveals its
“functional dimension” and opens a new approach to naturalization.
The idea of functional phenomenology and the method of phenom-
enological decomposition reveals that there is a functional dimension in
phenomenological descriptions. Putting it concisely, functional phenom-
enology describes and analyzes processes that contribute to experiences
of a certain type. Functional phenomenology can be located at the same
level as phenomenological psychology, which is different from transcen-
dental phenomenology. In this way, we avoid problems with naturalizing
the transcendental and acknowledge the mutual relation between phe-
nomenological theory and empirical research. Functional phenomenology
provides general concepts of intentional functions involved in the consti-
tution of the studied experience which can be applied in empirical studies.
An objection to the functional reading of phenomenology formulated
by Witold Płotka (2021) stresses that the distinction in Husserl’s works
between phenomenological psychology and transcendental phenomenol-
ogy is not clear-cut. This is so because the eidetic and the transcendental
aspects of phenomenological method cannot be clearly separated. I agree
that there is no clear-cut distinction here. For example, phenomenologi-
cal psychology applies concepts that are elaborated in the transcendental
attitude, such as noesis and noema. Although the distinction is not pre-
cise, it is sufficient for introducing a functional reading of phenomenol-
ogy as a project separate from transcendental phenomenology and yet, at
the same time, related to it. In other words, functional phenomenology,
much like phenomenological psychology, is distinct from transcendental
phenomenology, but it is not fully autonomous from it; that is, it assumes
concepts from transcendental phenomenology. The distinctiveness of
functional phenomenology consists in its methodology and explanatory
aim, which is different from the objective of transcendental phenomenol-
ogy. Functional phenomenology does not aim to explain transcendental
subjectivity and its role in the very foundation of reality as transcendental
phenomenology does (e.g., Moran, 2013) but investigates types of inten-
tional experience, providing decomposition and descriptions of functions
involved in the production of these types of experience. As such, analy-
ses conducted in functional phenomenology may provide constraints on
empirical research investigating mental phenomena without being car-
ried out in the transcendental attitude.
But it is still not clear how describing that functional dimension is
conducive to naturalizing phenomenology and how that naturalization
should proceed. Recalling what was covered in Chapter 3 about the theo-
retical integration of cognitive science, constraints are the key to inte-
grating different fields of study—something these fields can provide for
multilevel mechanistic models. Functional analysis provides characteriza-
tion and decomposition of cognitive functions, the realization of which
can be investigated in another field. Thus, functional analyses in general,
including Cummins-style analyses can be read as constraining mechanis-
tic models (Piccinini & Craver, 2011).
Now, let us consider how functional phenomenology could fit into the
theoretical integration of cognitive science. First, notice that functional
phenomenology as a study of the structures of consciousness is distinct
but not fully autonomous from other fields of research addressing con-
sciousness; the results from these fields may constrain phenomenology
and vice versa. It is autonomous, however, in choosing its own methods
of investigating and analyzing mental phenomena, such as decomposition
and its descriptive terminology. As I argued in Chapter 3, I do not believe
that phenomenology can deliver a complete explanation of mental phe-
nomena on its own. It is unable to address causal relations or the struc-
tural organization of underlying mechanisms. The key point is, however,
the nature of cognitive phenomena, which are complex and multifaceted.
In order to explain this class of phenomena, an integration of different
fields of research, including phenomenology, is required. Phenomenology
can significantly contribute to an integrative explanation.
As I argued previously, phenomenology in its historical, as well as con-
temporary, form can deliver structural descriptions of the studied expe-
rience or cognitive capacity; that is, phenomenological analysis is able
to identify invariant structures of experience. This already constrains
further research to a greater degree than does a simple description of
the explanandum phenomenon in folk-psychological terms. But my argu-
ment goes further. I propose to think of functional phenomenology as a
decompositional strategy similar to the functional decomposition applied
in psychology. Phenomenological decomposition, in an analogous man-
ner to functional decomposition, breaks down the target capacity into a
set of functions responsible for producing the experience under study. If
that is the case, then functional phenomenology can, in principle, provide
functional constraints.
In the following, I consider two examples of how phenomenological
analyses of constitutive functions contribute to a multilevel explanation
of a target capacity. The first example concerns a naturalistic account of
subjectivity understood as first-person orientation; the second one shows
how the results of phenomenological analyses can be connected to the
neuroscience of vision.
4.5.1 Example 1: First-Person Perspective

The first example is Neisser’s (2015) proposal of an evolutionary-devel-
opmental account of subjectivity understood as the first-person perspec-
tive, which is supported by phenomenological analyses of temporality.
The methodological side of Neisser’s approach consists of three levels:
(1) providing structural and dynamic analyses in terms of the first-

person perspective, (2) identifying neuropsychological functions for
the structural and dynamic forms, (3) connecting these functions
with underlying mechanisms via the fundamentally historical and
comparative methods of evolutionary developmental biology.
(p. 4)
The objective is hence to integrate phenomenological analyses of struc-

tures of subjectivity with related neuropsychological functions and bio-
logical mechanisms which realize these functions.
Neisser proposes to think of subjectivity as identification-free self-
reference (Evans, 1982). In the first-person perspective, there is no
question as to whom the experience is given, or to put it differently, sub-
jective experience is always self-referential. As Neisser (2015) argues,
“considering subjectivity as the identification-free form of the first
person perspective makes a functional analysis possible. Ecologically
considered, subjectivity serves an orienting function for an animal in
the environment of its concern” (p. 3). In other words, the first-per-
son perspective consists in how things relate to the subject here and
now. The self-orientating function is essentially spatial and temporal.
Here phenomenology comes onto the scene. Neisser refers to Husserl’s
theory of time-consciousness, in particular to his analyses of the reten-
tional-protentional structure of time-consciousness. The function of
retention provides the subject information about his previous posi-
tion and location. Protention is directed toward the near future, so
it furnishes the experience of “now” with expectations about what
will soon happen in the environment. Importantly, first-person spatio-
temporal orientation is affectively silent. A subject navigating in the
environment may recognize places as, for example, potentially danger-
ous, safe, nourishing, and so on. From among these functions of time-
consciousness, protention, in particular, has an affective aspect related
to subjective expectations about what will happen if I stay here, what
will change if I go there, and the like. To sum up, in Neisser’s (2015)
words, “the core of for-me thought is an affectively valanced sense of
directionality in an environment” (p. 108).
The next level of Neisser’s explanation of subjectivity is to map the
functions that amount to the general capacity of first-person orienta-
tion in an environment onto neurobiological mechanisms that were
developed in the course of evolution. Neisser (2015) calls this level of
explanation “homology thinking,” which “names a set of historical and
comparative explanatory strategies, used throughout biology, capable
of distinguishing the different paths by which biological structures
develop” (p. 6). Mechanisms identified in homology and comparative
biology may then be investigated further through experimental inter-
ventions. In order to integrate the phenomenological and philosophi-
cal analyses of the first-person perspective with neurobiological and
evolutionary models, Neisser refers to the olfactory spatial hypothesis
(Jacobs, 2012) and the functional model of parallel maps (Jacobs &
Schenk, 2003). According to the former, the spatial orientation sys-
tem emerged from the evolutionarily prior olfactory system, which
“evolved as a specialization for mapping spatiotemporal stimuli into
functional associative memory structures” (Neisser, 2015, p. 95). These
functional structures are cognitive maps localized in hippocampal sub-
regions (the dentate gyrus, area CA1, CA3). To be precise, the maps
consist of two parallel and specialized maps integrated into one. The
first is the bearing map that provides directional information about
the gradation of stimuli (increasing or decreasing intensity). The sec-
ond map is the topological map—encoding the topological properties
of an environment. The integrated map puts topological characteris-
tics onto the bearing space and thus allows an animal to effectively
navigate in an environment, to have expectations about the places it
is heading, and to retain information about places visited in the past.
The hippocampal representation of the surrounding environment is
created in the process of learning. Here, affective neuroscience pro-
vides a plausible explanation introducing the concept of primary affec-
tive modes (Panksepp & Biven, 2012), in particular the seeking mode
that motivates and rewards an animal through the dopamine system
for exploring an environment and remembering (marking on the map)
ecologically important places.
To conclude, Neisser’s account of subjectivity is an interesting exam-
ple of how to integrate phenomenological analysis of the first-person
perspective with naturalistic explanations, in particular evolutionary-
developmental ones. The explanation spans levels of first-person expe-
rience, psychological functions, and underlying neural mechanisms.
The methodological steps are congruent with the idea of functional
phenomenology introduced earlier. First, phenomenological analysis
introduces functions (the identification-free self-reference functions of
time-consciousness: retention and protention) that are constitutive for
the capacity of experiencing the world from the first-person perspective.
Second, the constitutive functions are matched with specific functional
models (navigation in an environment, cognitive maps), which, in turn,
are mapped onto neural mechanisms (hippocampal cognitive maps, pri-
mary affective modes generated in the mesolimbic system). As a result,
Neisser’s proposal can be read as a “phenomenologically valid” natural-
istic account of subjectivity.
4.5.2 Example 2: Vision Studies

Let me begin by briefly recalling how Husserl analyzes visual perception.
First of all, Husserl notices that although we see material objects, what
is actually perceived is an object’s adumbrations. The process of visually
perceiving an object can thus be fragmented into a succession of adum-
brations of the object. In Ideas I, Husserl (1982) writes that
of necessity a physical thing can be given only ‘one-sidedly’; and that

signifies, not just incompletely or imperfectly in some sense or other,
but precisely what presentation by adumbrations prescribes. A physi-
cal thing is necessarily given in mere ‘modes of appearance’ in which
necessarily a core of ‘what is actually presented’ is apprehended as
being surrounded by a horizon of ‘co-givenness’ which is not given-
ness proper, and of more or less vague indeterminateness.
(p. 94)
On the one hand, we perceive and recognize the object of perception, and
the object’s meaning (the noematic core) remains constant. But, on the
other, the flow of adumbrations changes in accordance with bodily move-
ments. Importantly, Husserl remarks that the co-perceived adumbrations
of an object are, to some extent, undetermined, but the indeterminacy can
also be reduced by bodily movements; that is, bodily movements generate
new adumbrations, for instance, revealing a side of the object that was
previously not in view.
For Husserl (1989a), the body is “the medium of all perception; it is
the organ of perception and is necessarily involved in all perception” (p.
61, emphasis in the original). Visual perception is therefore an embodied
process that is characterized by bodily spatial orientation and constitutes
the visual perspective. As Husserl (1989a) writes,
the Body acquires as the bearer of the zero point of orientation, the
bearer of the here and the now, out of which the pure Ego intu-
its space and the whole world of the senses. Thus each thing that
appears has eo ipso an orienting relation to the Body, and this refers
not only to what actually appears but to each thing that is supposed
to be able to appear.
(p. 61)
Another aspect of embodied perception is motility. The body is moving,

and thus, it determines (in a motivational sense) the successive adumbra-
tions of a perceived object. At the same time, we feel our body in move-
ment in a kinesthetic experience. As Husserl (1989a) concludes,
we constantly find here this two-fold articulation: kinesthetic sen-

sations on the one side, the motivating; and the sensations of fea-
tures on the other, the motivated. The like holds, obviously, for touch
and, similarly, everywhere. Perception is without exception a unitary
accomplishment which arises essentially out of the playing together
of two correlatively related functions.
(p. 63, emphasis in the orignal)
These two functions constitute an “if–then” system of perception and

action; that is, if my body moves over there, then I will see an object so
and so. Finally, visual perception is, for Husserl (1991), an essentially
temporal process, that is, to see a material thing requires, on the one
hand, the retention of adumbrations already past and, on the other, the
protention of adumbrations that are expected to appear with respect
to movement. If the protentional intentionality is fulfilled, that is, the
objects continue to appear as expected, then the perception of objects
goes on. When the intention is not fulfilled, then the expectation is not
satisfied and the process of perception is interrupted, which can lead, for
example, to modification of the meaning of a perceptual object or change
in bodily action.
To sum up, Husserl’s analysis of visual experience identifies several
key features of visual perception (adumbrations, spatial perspective, tem-
porality) as well as functions involved in the production of experience
(sensory receptivity coupled with bodily activity; retention and proten-
tion, which constitute a perceptual object in time). On a general level,
phenomenology conceives visual perception as an embodied and active
process; thus, perception should be conceived as coupled with bodily
action and dynamic spatial perspective. A more specific constraint con-
cerns the role of retention and protention, which suggests that visual per-
ception consists of cognitive functions of retaining the past content of
visual experience and anticipating it in the near future.
It was recently argued that these phenomenological constraints are
best satisfied by the anticipation-fulfillment (AF) model of visual percep-
tion (Madary, 2017). Drawing on Husserlian phenomenology, Michael
Madary 2017 formulates three specific constraints, according to which
visual perception is perspectival (we see material objects from one per-
spective at a time, but their properties can be seen from multiple per-
spectives), temporal (visual experience is continuous in time), and
indeterminate (the visual field has a central region of clear perception
and surrounding regions of indeterminacy), and argues that these con-
straints are satisfied by the AF model. In short, the AF model assumes
that vision relies on constant anticipation and fulfillment (or disappoint-
ment) of sensory content as a consequence of self-generated movement
(or its neural simulation in the case of an inability to move). In particular,
Madary (2017, pp. 41–58) defends two claims. First, the specific anticipa-
tion claim, which states that anticipation of perceptual content is always
specific and factual rather than general. Second, the perceptual connect-
edness claim, according to which “if S substantially changes her perspec-
tive on o, her visual phenomenology will present different views of o’s
factual properties” (Madary, 2017, p. 42). In other words, self-generated
change of spatial perspective on the perceived object leads to change of
representations of the object’s properties, such as shape, size, and so on.
Importantly, the subject is able to anticipate this self-generated change,
although not precisely, as factual properties may appear from different
perspectives. Thus, the nature of visual perception may be characterized
as both specific and, to some extent, undetermined.
Madary (2017, p. 27) calls phenomenological constraints descriptive
because they can be discovered through analysis of the first-person visual
experience and without experimental investigation. However, it seems
that phenomenological constraints are not merely descriptive in the sense
that they are inferred from simple descriptions of how one is experienc-
ing the world visually, but they also concern structural invariants of expe-
rience that can be discovered with phenomenological method, including
the method of imaginary variation. Moreover, as I argued in Chapter
3, phenomenology delivers not only descriptions but constitutive under-
standing, which indicates the constitutive conditions for the experience in
question. It seems, therefore, that phenomenological constraints are more
than merely descriptive.
At first glance, it might seem that phenomenology of vision provides
interlevel conceptual constraints. As established in Chapter 3, interlevel
constraints are provided by research fields that study a given target
phenomenon on different levels (Craver, 2007). Conceptual constraints
concern the way we think about the explanandum and underlying mech-
anism. According to the phenomenology of visual perception, we should
think about it in terms of anticipation and fulfillment. This constrains the
space of possible mechanisms to those that can produce internal repre-
sentations of the immediate future on the basis of input sensory data and
other relevant information (e.g., environmental context).
But taking into account the idea of functional phenomenology, the
Husserlian account of visual perception may provide stronger con-
straints than conceptual constraints, namely, functional ones. Functional
constraints are another type of interlevel constraint, that is, one field
of research investigates a target function and its component sub-func-
tions, whereas another field studies these function’s implementation.
Phenomenological decomposition unfolds anticipation and fulfillment to
specific constitutive functions of retention, protention, sensory receptiv-
ity, and self-generated movement. These constraints concern the general
functionality that a system has to implement in order to exhibit the stud-
ied visual capacity. In other words, phenomenology provides a general
functional sketch that can then be specified in a functional “how-possi-
bly” model. Recall that how-possibly models are incomplete models of
mechanisms that hypothetically could be responsible for the target capac-
ity. How-possibly models contain black boxes and lack many details with
respect to their realization. Once supplemented with these details and
confronted with empirical data, a how-possibly model may become a
“how-actually” model (see Chapter 3 for a longer discussion).
Can this rough phenomenological model of vision be specified into
the form of how-possibly model? According to Madary (2017), the AF
model of vision can be expressed in terms of Bayesian predictive pro-
cessing (Clark, 2013, 2015; Hohwy, 2013). Accordingly, anticipation is
understood as a prediction that is derived from sub-personal inferences
based on a hierarchical probabilistic model of the world. In short, the
system computes the perceptual state that is most likely to be next on
the basis of prior probability, the likelihood of the hypothesis, and the
input data. Such a model is still an incomplete functional model, but it
tells us more about specific functions that the responsible mechanisms
we are looking for should realize. Furthermore, the predictive processing
approach has empirical consequences concerning the nature of underly-
ing neural mechanisms, for example, that they should be computational
mechanisms.5 The next step in building a multilevel explanation is to
search for actual neural mechanisms underlying the predictive coding.
The model provides empirical implications concerning the organization
of the neural mechanism and testable hypotheses, yet empirical research
on predictive coding is still at a very early stage. For example, it is hypoth-
esized that cortical predictive processing requires feedback connections
(e.g., Madary, 2017, pp. 123–128). Another implication states that there
should be two functionally distinct subpopulations of neurons: One is
responsible for representing the current sensory state and computing
expectation on this basis, and the other population encodes prediction
errors, that is, activated in the case of mismatch between expectations
and sensory input (Clark, 2013, p. 12).
The example of vision shows that the results of phenomenological
analysis of experience are coherent with some results from other research
fields and can contribute to a multilevel explanation. Phenomenology,
generally speaking, investigates the target phenomenon on the level of
first-person experience and may provide conceptual constraints on other
research fields. The application of phenomenological decomposition
allows providing stronger constraints concerning the functions of con-
sciousness involved in the constitution of visual experience. These con-
straints can be specified in the functional model of vision formulated
on the basis of predictive coding. It is important to emphasize that I am
not claiming that the predictive processing model corresponds exactly to
the phenomenological model of vision. Some authors even contrast these
approaches (Zahavi, 2017; cf. Piekarski, 2017). Madary (2017) rejects
the possibility that “there is a one-to-one mapping between visual antici-
pations and predictive neural signals” (p. 158). According to him, the
relation between the phenomenological level and the sub-personal level
realizing the predictions is of structural correspondence. In my view, this
is a constraining relation between different research fields allowing for
integration; that is, phenomenological analysis provides general constitu-
tive constraints, which are satisfied by the functional predictive process-
ing model, which, in turn, constrain the space of underlying mechanisms.
4.6 Conclusion
In this chapter, I considered two different functionalist interpretations of
Husserlian phenomenology. I argued that a strong computational reading
of phenomenology is implausible for several reasons. Another function-
alist reading of phenomenology, one that I argued in favor of, adopts
a methodological perspective and points to the similarities between the
explanatory strategy of functional analysis and the phenomenological
method of decomposition. I argued that Husserl developed an original
notion of intentional function and applied it to decompose experiences
and identify the functions responsible for the production or constitution
of those experiences. According to this interpretation, phenomenologi-
cal descriptions of the intentional structure of experience can be read as
analogous to functional descriptions. Finally, I considered the implica-
tions of such a functionalist reading for the project of naturalization. I
argued that nonreductive naturalization should be understood in terms
of an integration of research fields that relies on providing constraints.
Taking under consideration two case studies, the first-person perspective
and vision studies, I have shown that functional phenomenological analy-
ses can provide general constraints concerning functions involved in the
constitution of the experience in question and thus become a part of a
chain of interlevel constraints constituting a multilevel integrative expla-
nation. Phenomenological constraints based on the analysis of first-per-
son experience can be specified on the sub-personal level in the functional
model of cognitive maps—and visual perception based on predictive cod-
ing theory. These functional models, in turn, provide constraints on the
models of underlying neural mechanisms.
Notes
1 An earlier version of this chapter appeared as “Phenomenology and
Functional Analysis: A Functionalist Reading of Husserlian Phenomenology”
(Pokropski, 2020).
2 This is not the only possible interpretation of Husserl’s noema (for an over-
view, see Drummond, 1997). For example, a different reading of noema was
proposed by Gurwitsch (2009), who argues for perceptual noema. Perceptual
noema, in short, is an adumbration of a thing seen from a certain perspective.
Gurwitsch’s noema is not abstract and imperceivable but, on the contrary, is
concrete and perceivable. Thus, perception could be conceived as more direct.
However, some argue (e.g., Dreyfus, 1982) that it is an extension of Husserl’s
notion of noema rather than Husserl’s original idea.
3 Another concept of function is the set-theoretical concept, according to which
function is an abstract relation between two sets. The set-theoretical understand-
ing of function is not present in Husserl’s works, but, interestingly, it was used by
Bolzano, whose works influenced Husserl (Benoist, 2002; Centrone, 2010).
4 A more recent example of the phenomenological decomposition of the con-
tents of experience can be found in the micro-phenomenological studies of
first-person experience conducted by Petitmengin et al. (2013, 2019). The
micro-phenomenological method is a second-person approach and begins
from a phenomenological interview. The descriptions of experience from
micro-phenomenological interviews are analyzed and decomposed in order to
extract the structure of the studied experience (see Chapter 5 for an extended
discussion of micro-phenomenology).
5 Note, however, that some (e.g., van Es, 2020) defend an instrumentalist read-
ing of predictive processing models.
References
Benoist, J. (2002). Husserl and Bolzano. In A. T. Tymieniecka (Ed.), Phenomenology
world-wide: Analecta Husserliana (pp. 98–99). Springer.
Block, N. (1980). What is functionalism? In N. Block (Ed.), Readings in philoso-
phy of psychology (Vol. 1, pp. 171–184). Harvard University Press.
Block, N. (2007). Consciousness, function, and representation: Collected papers.
MIT Press.
Centrone, S. (2010). Functions in Frege, Bolzano and Husserl. History and
Philosophy of Logic, 31(4), 315–336. doi:10.1080/01445340.2010.511795
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future
of cognitive science. Behavioral and Brain Sciences, 36(3), 1–24.
Clark, A. (2015). Surfing uncertainty: Prediction, action, and the embodied mind.
741–765. doi:10.2307/2024640
Dreyfus, H. (1982). Husserl’s perceptual noema. In H. L. Dreyfus & H. Hall
(Eds.), Husserl, intentionality and cognitive science (pp. 97–123). MIT Press.
Dreyfus, H. (1988). Husserl’s epiphenomenology. In H. Otto & J. Tuedio (Eds.),
Perspectives on mind (pp. 85–101). Reidel.
Dreyfus, H., & Hall, H. (1982). Introduction. In H. L. Dreyfus & H. Hall (Eds.),
Husserl, intentionality and cognitive science (pp. 1–27). MIT Press.
Drummond, J. J. (1997). Noema. In L. Embree (Ed.), Encyclopedia of phenom-
enology (pp. 494–499). Springer.
Evans, G. (1982). The varieties of reference. Clarendon Press.
Fodor, J. (1968). Psychological explanation. Random House.
Fodor, J. (1975). Language of thought. Harvard University Press.
Fodor, J. (1980). Methodological solipsism considered as a research strategy in
cognitive psychology. The Behavioral and Brain Sciences, 3(1), 63–73.
Fodor, J. (1983). The modularity of mind. MIT Press.
Føllesdal, D. (1969). Husserl’s notion of noema. The Journal of Philosophy,
66(20), 680–687. doi:10.2307/2024451
Føllesdal, D. (1988). Husserl on evidence and justification. In R. Sokolowski
(Ed.), Edmund Husserl and the phenomenological tradition (pp. 107–129). The
Catholic University of America Press.
Gallagher, S., & Brøsted Sørensen, J. (2006). Experimenting with phenomenology.
Consciousness and Cognition, 15(1), 119–134. doi:10.1016/j.concog.2005.03.002
Routledge.
Gurwitsch, A. (2009). The structure of perceptual noema (J. Garcia-Gómez,
Trans.). In J. Garcia-Gómez (Ed.), The collected works of Aron Gurwitsch
(1901–1973) (Vol. 1, pp. 185–307). Springer.
Hohwy, J. (2013). The predictive mind. Oxford University Press.
Husserl, E. (1964). The idea of phenomenology (W. P. Alston, Trans.). Martinus
Nijhoff.
Nijhoff.
Husserl, E. (1971). “Phenomenology” Edmund Husserl’s article for the
Encyclopaedia Britannica (1927): New complete translation by Richard E.
Palmer. Journal of the British Society for Phenomenology, 2(2), 77–90.
Husserl, E. (1977). Phenomenological psychology (J. Scanlon, Trans.). Martinus Nijhoff.
Husserl, E. (1989a). Ideas pertaining to a pure phenomenology and to a phenom-
Husserl, E. (1989b). The crisis of European sciences and transcendental phenom-
Husserl, E. (1997). Thing and space: Lectures of 1907 (R. Rojcewicz, Trans.).
Kluwer.
Routledge.
Husserl, E. (2012). Philosophy of arithmetic: Psychological and logical investiga-
tions with supplementary texts from 1887–1901 (Vol. 10). Springer Science &
Business Media.
Jacobs, L. (2012). From chemotaxis to the cognitive map. Proceedings of the
National Academy of Sciences, 109(Suppl. 1), 10693–10700.
Jacobs, L., & Schenk, F. (2003). Unpacking the cognitive map: The parallel map
theory of hippocampal function. Psychological Review, 110(2), 285–315.
Kockelmans, J. J. (1987). HusserI’s original view on phenomenological psychol-
Kortooms, T. (2002). Phenomenology of time. Springer.
Landgrebe, L. (1981). The phenomenology of Edmund Husserl. Cornell
University Press.
Livingston, P. (2005). Functionalism and logical analysis. In D. W. Smith & A.
L. Thomasson (Eds.), Phenomenology and philosophy of mind (pp. 19–40).
Madary, M. (2017). Visual phenomenology. MIT Press.
McIntyre, R. (1986). Husserl and the representational theory of mind. Topoi,
5(2), 101–113. doi:10.1007/BF00139224
Panksepp, J., & Biven, L. (2012). The archaeology of mind: Neuroevolutionary
origins of human emotion. W. W. Norton & Company.
Petitmengin, C., Remillieux, A., Cahour, B., & Carter-Thomas, S. (2013). A gap in
Nisbett and Wilson’s findings? A first-person access to our cognitive processes.
Consciousness and Cognition, 22(2), 654–669.
Petitmengin, C., Remillieux, A., & Valenzuela-Moguillansky, C. (2019). Discovering
the structures of lived experience: Towards a micro-phenomenological analysis
method. Phenomenology and the Cognitive Sciences, 18(4), 691–730.
Piekarski, M. (2017). Commentary: Brain, mind, world: Predictive coding,
neo-Kantianism, and transcendental idealism. Frontiers in Psychology, 8.
doi:10.3389/fpsyg.2017.02077
Piccinini, G. (2010). The mind as neural software? Understanding functional-
ism, computationalism, and computational functionalism. Philosophy and
Phenomenological Research, 81(2), 269–311.
Płotka, W. (2021). A commentary on Pokropski’s functionalist reading of
Husserlian phenomenology. Frontiers in Psychology, 12. doi:10.3389/
fpsyg.2021.614842
Pokropski, M. (2015). Affectivity and time: Towards a phenomenology of
embodied time-consciousness. Studies in Logic, Grammar and Rhetoric, 41(1),
161–172.
Pokropski, M. (2020). Phenomenology and functional analysis: Afunctionalist
reading of Husserlian phenomenology. Phenomenology and the Cognitive
Sciences, 19, 869–889.
Putnam, H. (1960). Minds and machines. In S. Hook (Ed.), Dimensions of mind
(pp. 138–164). New York University Press.
Rodemeyer, L. (2006). Intersubjective temporality: It’s about time. Springer.
Roy, J.-M., Petitot J., Pachoud, B., & Varela F. J. (1999). Beyond the gap: An intro-
Sowa, R. (2012). Eidetics and its methodology (E. A. Behnke, Trans.). In S. Luft
& S. Overgaard (Eds.), The Routledge companion to phenomenology (pp.
Thinès, G. (1977). Phenomenology and the science of behaviour. George Allen
& Unwin.
van Es, T. (2020). Living models or life modelled? On the use of models in
the free energy principle. Adaptive Behavior. Advanced online publication.
doi:10.1177/1059712320918678
Welton, D. (1982). Husserl’s genetic phenomenology of perception. Research in
Phenomenology, 12(1), 59–83.
Zahavi, D. (2003). Husserl’s phenomenology. Stanford University Press.
Zahavi, D. (2017). Brain, mind, world: Predictive coding, neo-Kantianism, and
transcendental idealism. Husserl Studies, 34(2–4), 47–61.
5 Phenomenology and Dynamical
Modeling
5.1 Introduction
In this chapter, I consider whether phenomenology could provide dynam-
ical constraints on the space of possible mechanisms and thus contrib-
ute to integrative dynamical-mechanistic explanations. Dynamical
constraints concern the dynamics of the phenomenon in question, for
instance, that it is oscillatory or consists of three subsequent phases. The
chapter is structured as follows: First, I discuss the relation between the
dynamical approach and the mechanistic model of explanation in the
context of the integration of research fields in cognitive science. I argue
that these frameworks are not opposed to one another but rather are
complementary and that dynamical-mechanistic explanation is the new
gold standard of explanation in cognitive science. Second, I consider
applications of phenomenology in cognitive science in which the dynami-
cal nature of the studied phenomena is emphasized, namely neuro- and
micro-phenomenology. In particular, I discuss neurophenomenological
studies of epilepsy and confront their results with dynamical modeling
of epileptic seizures in neuroscience. I argue that neurophenomenology
provides dynamical constraints that are mainly conceptual and thus too
weak to contribute to dynamical-mechanistic models. I conclude with
a discussion of how to revise neurophenomenology in order to provide
stronger constraints. In particular, I argue for formalizing pheno-dynam-
ical models.
5.2 Integrating the Dynamical and Mechanistic Frameworks

In Chapter 3, I introduced the idea of mechanistic integration of cogni-
tive science, according to which different fields of research are integrated
when they provide constraints on the space of possible mechanisms
(Craver, 2007). At first glance, it may seem that the dynamical approach
stands in opposition to the mechanistic model of explanation. The point
of contention is the decomposability of the system exhibiting the tar-
get phenomenon and the localization of activities in components of
DOI: 10.4324/9781003035367-8
mechanisms—these are the two key explanatory strategies of the mecha-
nistic framework. On the one hand, some dynamicists (e.g., Lamb &
Chemero, 2014) argue that cognitive systems are non-decomposable;
thus, they cannot supplement the mechanistic framework, which depends
on assuming the target system’s decomposability or near-decomposabil-
ity. On the other hand, mechanists (e.g., Bechtel & Richardson, 2010)
argue that decomposability, as well as localization, are heuristic strate-
gies the results of which can be revised or falsified in the research pro-
cess. It does not follow, however, that the mechanistic approach is wrong.
For example, an initial assumption about the direct localization of visual
processing in the occipital lobe was changed in the course of research
to a composition of cognitive subfunctions localized in distributed areas
of the brain. The simple heuristic of direct localization failed and was
replaced by complex localization. But this model of visual processing is
still mechanistic (e.g., Bechtel, 2008). We can push this idea further and
argue that decomposition and localization strategies may fail in general
and leave us with cognitive models the functions of which are nonlocaliz-
able in the parts of the brain but that are nevertheless still explanatorily
useful models (Stinson, 2016).
Generally speaking, decomposability should not be understood as an
ontological category but as an epistemological one. We can decompose
a target system in many ways depending on our explanatory interests
and knowledge. Furthermore, there are different degrees of decompos-
ability, and, accordingly, a system may be decomposable, nearly decom-
posable, minimally decomposable, or non-decomposable (Simon, 1962;
Wimsatt, 1986). The decomposability and localization heuristics fail
in the case of systems that do not have parts or operations that can
be distinguished. Connectionist networks are good examples, as their
behavior emerges from the organization of relatively simple but highly
integrated elements (Bechtel & Richardson, 2010, p. 228). Network
models are not classical mechanistic models because they are non-
decomposable, yet they are in some sense mechanistic because the sys-
tem’s behavior is produced by its components and activities within the
system. What is crucial in network systems is not the contribution of
individual parts but their organization.
The heuristics of decomposition and localization may also fail when
parts and operations interact with each other in a nonlinear fashion and
when nonlinear interactions appear between parts and the behavior of
the whole system; then the behavior of the parts is system dependent and
may be called emergent. But as Bechtel and Richardson (2010) argue,
“when these two conditions are met, the systemic behavior is reason-
ably counted as emergent, even though it is fully explicable mechanisti-
cally” (p. xlvi). Such mechanistic explanation is different from the classic
picture, in which we identify and empirically study the real parts and
operations of a mechanism; it mainly relies on computational modeling.
Phenomenology and Dynamical Modeling 169
Furthermore, to capture the emergent character of the studied behavior,
such a model often involves formal and conceptual tools from dynamical
systems theory; thus, it is called a dynamical-mechanistic explanation.
According to Bechtel and Abrahamsen (2010), such explanations may be
built on research on circadian rhythms, whereby our understanding of
the underlying mechanisms was extended by computational dynamical
modeling. The variables of the computational models captured key prop-
erties of parts, their nonlinear operations, and the complex organization
of the mechanism; simulations have shown that these models are able to
produce the target dynamical phenomenon.
Another example of this kind of dynamical-mechanistic explanation
comes from network neuroscience (M. Silberstein & Chemero, 2013;
Zednik, 2019). Network neuroscience is a research field investigating how
the brain coordinates and integrates distributed processes, which often
vary in level of organization, localization, and timescale (for an overview,
see Sporns, 2011). According to Michael Silberstein and Chemero (2013),
“most of this coordination represents patterns of spontaneous, self-orga-
nizing, macroscopic spatiotemporal patterns, which resemble the on-the-
fly functional networks recruited during activity” (p. 962). As they argue,
most models which account for these self-organizing networks share
several key generalizations, including that in such network processes,
the temporal dimension is essential, and thus, they should be viewed as
dynamically evolving graphs. In this approach, which draws on graph
theory and dynamical systems theory, the basic units of explanation are
brain multiscale networks and their distributed connections, which are
represented in graphical form, having specific topological properties. For
example, it is argued that from among the various network topologies,
the so-called small-world networks are especially interesting, as such net-
works can be found in the brain (Sporns, 2011). How do we explain
these network processes? As Silberstein and Chemero claim, explana-
tions in network neuroscience are pluralistic and draw on various frame-
works including mechanistic, dynamical, and statistical-causal ones, each
of which contributes to network simulations. But the main explanatory
work is done by the graph models:
The explanatory point is that such graphical simulations allow us to

derive, predict, and discover a number of important things such as
mappings between structural and functional features of the brain;
cognitive capacities; organizational features such as degeneracy,
robustness, and plasticity.
(M. Silberstein & Chemero, 2013, p. 964)
Importantly, predictions are not the only thing to be gained from network
simulations; thus, one cannot make the predictivism objection often formu-
lated against covering law and dynamical explanations (see Chapter 3 for
a discussion). On the contrary, they allow us to understand why a studied
network exhibits certain properties in virtue of its dynamical and graphical
properties.
So, the mechanistic explanatory strategies of localization and decom-
position may fail, but that does not rule out the plausibility of the mecha-
nistic account. Nor is it true that all dynamical cognitive systems are
always non-decomposable. They can be decomposed in numerous ways,
but the key to doing it right is capturing how a system’s components are
dynamically coupled, that is, how they interact with each other and how
these interactions contribute to the behavior of the system. An example
of a decomposable dynamical system is the already mentioned percep-
tual agent model proposed by Beer (2000, 2003; see Chapter 3). The
two components of the system are described in the model by a sepa-
rate list of equations (16 equations in total). The fact that these compo-
nents are coupled is expressed in the equations by variables describing
one component being parameters for the other components, and vice
versa. Beer’s model shares features of both dynamical and mechanistic
explanations. A further example is related to the decomposability of eco-
logical systems. As Sabrina Golonka and Andrew Wilson (2019) argue,
the ecological approach does not have to be limited to purely dynami-
cal explanations—it can support mechanistic ones too. Accordingly, “the
ecological approach grounds psychological explanations in the types of
parts and operations that are amenable to dynamic causal mechanistic
models” (Golonka & Wilson, 2019, p. 679). The key to decomposing the
target cognitive system in an ecologically meaningful way is to choose
the relevant scale of decomposition. For Golonka and Wilson, this scale
is defined by ecological information, that is, kinematic patterns in ambi-
ent arrays which specify dynamical properties. In short, an organism uses
these kinematic patterns to coordinate action and perceive the properties
of an environment, such as affordances, in such a way that the “eco-
logical information specifying task dynamics constrains action coordi-
nation and control to the space of task-relevant solutions” (Golonka &
Wilson, 2019, p. 683). To put it differently, a cognitive system’s degrees of
freedom are limited by ecological information in order to shape actions
and achieve the goal in an optimal fashion. Furthermore, Golonka and
Wilson suggest that ecological information may also constrain neural
coordination processes. Taking into account ecological information,
they reconsider the model of rhythmic movement proposed by Geoffrey
Bingham (2004) as an example of a dynamical-mechanistic model. The
model differs from the famous Haken–Kelso–Bunz (HKB) model (Haken
et al., 1985; see Chapter 3), which was purely dynamical, in that it allows
one to make predictions about physical components of underlying mech-
anisms. The refined model is based on the assumption, supported by
empirical research, that rhythmic coordination depends on the percep-
tion of relative phase:
The information variable must be kinematic, it must specify relative
phase, it must be made of state variables (to preserve autonomy), it
must break the symmetry between relative phases in a manner that
matches the phenomena catalogued above, and it must be detectable
by both vision and proprioception.
(Golonka & Wilson, 2019, p. 688)
These examples show, on the one hand, that mechanism is not blindly
committed to decomposability and localization and, on the other, that
the dynamical approach is not limited to non-decomposable and emer-
gent phenomena. If that is the case, then the mechanistic framework
and the dynamical framework should be perceived as complementary
rather than exclusive (e.g., Kaplan & Bechtel, 2011; M. Silberstein &
Chemero, 2013; Zednik, 2011). For some (Golonka & Wilson 2019),
this “hybrid” dynamical-mechanistic model is the new “gold standard” of
explanation for cognitive science. Such a model combines the advantages
of the two approaches; it demonstrates the organization and composition
of the mechanism responsible for the target phenomenon and explores
the mechanism’s functioning in simulations. In particular, according
to Bechtel and Abrahamsen (2010), dynamical modeling may contrib-
ute to our understanding of a given mechanism in several ways: (1) by
demonstrating that the modeled mechanism is able to exhibit the target
phenomenon; (2) providing a larger space of parameter values and thus
allowing one to study the behavior of the mechanism, which would be
difficult to do empirically; (3) helping identify which parts and operations
are crucial for producing the target phenomenon; (4) helping explain the
mechanism’s malfunctions; (5) revealing the conditions under which the
mechanism may be coupled with another mechanism, thereby exhibit-
ing collective coordinated behavior; (6) allowing one to explore possible
alterations in relations between multiple mechanisms. From among these,
the last two are especially closely tied to dynamical modeling (Golonka
& Wilson, 2019, p. 677) and may contribute to studies of extended and
distributed systems (Zednik, 2011).
Now, taking into account that the dynamical and mechanistic
approaches may be understood as complementary, we can consider
whether phenomenology, especially where it is read through the lens of
dynamical systems theory, might provide dynamical constraints and thus
contribute to dynamical-mechanistic models.
5.3 Neurophenomenology and Dynamical Systems Theory

Neurophenomenology is certainly one of the first and key applications
of phenomenological method in cognitive science that explicitly refers to
dynamical systems theory. A brief review of the main points of neurophe-
nomenological method is in order, as introduced in Chapter 2. According
to Varela (1996), the methodological components of neurophenomenol-
ogy are thought to provide a toolkit for the scientific study of first-person
experience and will ultimately lead to bridging the “explanatory gap”
related to the hard problem of consciousness (Chalmers, 1996). These
steps are (1) a phenomenological description and analysis of first-per-
son experience, (2) an application of the formal apparatus of dynamical
systems theory (DST), and (3) EEG recordings of subjects’ brain activ-
ity. It is argued that such a method can facilitate studies of experience
and validation of disclosed experiential structures, which could be cor-
related with subjects’ neural activity. The working hypothesis of neuro-
phenomenology states that “phenomenological accounts of the structure
of experience and their counterparts in cognitive science relate to each
other through reciprocal constraints” (Varela, 1996, p. 343). These con-
straining relations between the phenomenal and the neural are under-
stood in neurophenomenology in terms of “generative passages” (Lutz
et al., 2002). Whereas the first methodological component of neurophe-
nomenology related to interviews and collecting first-person data was a
subject of discussion (e.g., Bockelman et al., 2013; Ciechanowski, 2017;
Jedličková & Müller, 2019), the application of DST did not attract as
much attention.
Recall that DST provides methods for both quantitative and qualita-
tive analysis of the behavior of the target system (see Chapter 3). The
quantitative methods require formal models and rely on analyses of the
target system’s behavior in relation to values and parameters included
in the equations. Qualitative analysis reveals features of long-term sys-
tem dynamics, such as stable and unstable states or attractors (regions in
the state space toward which the system evolves). Besides these methods
of analysis, there is also dynamic description (e.g., van Gelder & Port,
1995), which relies on the application of DST terminology, such as stable
state, attractor, or bifurcation, to describe a given system’s behavior. Such
a description is useful when the phenomena in question are too complex
to be expressed in a formal mathematical model. Van Gelder and Port
(1995) argue that using DST’s conceptual apparatus can shed new light
on those phenomena, but it is important to remember that the descrip-
tions will remain merely metaphorical if not supported by qualitative and
quantitative analyses. As I argue in the following, neurophenomenology
does not go beyond mere dynamic description.
Before the first neurophenomenological experiments were conducted
by Lutz et al. (2002), the terminology of DST was already present in
the theoretical study of time-consciousness proposed by Varela (1999;
see also van Gelder, 1999). Varela uses DST to connect findings about
the synchronization of neuron assemblies and the structure of time-con-
sciousness proposed by Husserl. Accordingly, on the neural level, neurons
behave as nonlinear oscillators, which when coupled give rise to the emer-
gence of neural assemblies understood as “distributed local networks of
neurons transiently linked by reciprocal dynamic connections” (Varela et
al., 2001, p. 229). These assemblies participate in a phenomenon called
large-scale integration, that is, the integration of nonlocal groups of neu-
rons (typically localized more than 1 cm from each other), the mechanism
of which, according to Varela et al. (2001), relies on phase synchroniza-
tion. As it is hypothesized, processes of large-scale integration through
synchronization are thought to be responsible for the emergence of con-
scious experience. For example, visual perception and the problem of the
binding of visual information is an area of research where phase syn-
chrony as a mechanism of integration has been applied (e.g., Roskies,
1999). As Varela (1999, p. 283) argues, these dynamical neural processes
shed light on the mechanism responsible for the experience of the “now,”
that is, the experience of the “now” phase in the flow of visual informa-
tion. Varela refers to Husserl, who argued that experience of the “now”
has a three-part structure: retention, protention, and “ur-impression” (see
Chapter 1). Accordingly, each “now” phase consists of retained experi-
ence that is already past, current reception of sensory impressions, and
protention (anticipation) of incoming experience in the near future (see
Chapter 1 for a longer discussion). Although Varela does not propose
any formal dynamical model of time-consciousness, he does apply DST
terminology, including attractor and bifurcation,1 to describe both neural
processes of synchronization and temporal change of perceptual experi-
ence, which allows him to think about the constitution of the “now” in a
novel dynamical fashion. As he writes,
as the model (and the data) shows, the synchronization is dynami-

cally unstable and will constantly and successively give rise to new
assemblies. We may refer to these continuous jumps as the trajecto-
ries of the system. Each emergence bifurcates from the previous ones
given its initial and boundary conditions. Thus each emergence is still
present in its successor.
(Varela, 1999, p. 283)
On the level of experience, these bifurcations may correspond, for exam-

ple, to changes in visual perception of image sequences, when one percept
is being switched for another. Importantly, these perceptual switches are
accompanied by retention of past phases and anticipation of incoming
phases. Although Varela’s proposal is purely speculative, it shows that
it is possible to give a dynamical interpretation of key categories of the
Husserlian theory of time-consciousness, which allows one to relate them
to underlying neural processes.
In the famous first neurophenomenological experiment conducted by
Lutz et al. (Lutz, 2002; Lutz & Thompson, 2003), the stakes were raised
and the application of DST was postulated to have the form of a formal
model (Lutz, 2002, p. 149; Lutz & Thompson, 2003, p. 34). Recall that
the experiment concerned degrees of perceptual readiness (see Chapter 2
for a detailed description of the experiment). The objective of the experi-
ment was to allow subjects to describe and categorize their experience of
perceptual readiness and to correlate these phenomenological categories
with EEG data. In the second-person process, researchers trained subjects
and together identified several phenomenological categories of the readi-
ness state (called phenomenological clusters, PhCs), which were then cor-
related with their dynamical neural signatures (DNSs). The application
of a formal dynamical model was thought to allow a more disciplined
dynamical analysis of the neural system’s behavior underlying the stud-
ied perceptual change and correlation with phenomenological clusters.
To the best of my knowledge, a formal dynamical model was never cre-
ated, and Lutz et al. provided only a general dynamic description of both
relata. More specifically, Lutz introduced dynamical expressions of DNSs
and corresponding PhCs. However, he admitted the dynamical analysis
was done at a “speculative level” (Lutz et al., 2002, p. 156); thus, it seems
that the coupling between behavioral, neurophysiological, and first-per-
son data that Lutz et al. (2002) postulated should be understood as a
relation of mere correlation—not fulfilling the criteria for “coupling” in
the DST sense: in DST, one can model the coupling of a system’s compo-
nents (or processes) by applying variables describing one coupled compo-
nent to equations describing the other component as parameters, and vice
versa (e.g., as done in the perceptual agent model discussed in Chapter 3).
To sum up, neurophenomenology introduced a dynamic description
of experience and the corresponding neural processes. It has, so far,
failed to deliver a formal dynamical model that would allow for genuine
dynamical analysis revealing the target system’s behavioral and experi-
ential trajectory. Only such a dynamical model can provide a dynamical
explanation in terms of a description of the target system’s phase space
that allows for prediction of its behavior and formulation of hypotheses
concerning the underlying mechanism. A formal model would also make
it possible to couple the first-person and the neural levels in a stronger
sense; that is, such a model would show the relations between experien-
tial categories and underlying neural dynamics.
5.4 Discovering the Dynamics of Experience—Epilepsy Study

The study of epilepsy is an interesting example in which various research
fields and explanatory frameworks converge and contribute to a model
of the malady, including such frameworks as causal-etiological, mecha-
nistic, and dynamical. Importantly, epilepsy has a diverse phenomenol-
ogy, including behavioral signs, such as bodily stiffness or convulsions
during seizures, and first-person experiences, such as sensory auras or
various interoceptive sensations in the time preceding a seizure. Generally
speaking, epilepsy is a condition characterized by seizures caused by
sudden synchronous neuronal activity in the brain leading to a distur-
bance in brain functioning (for an overview, see, e.g., Davis et al., 2005,
pp. 155–164). Although in most cases the cause of epilepsy may be estab-
lished, for example, brain tumor or metabolic disorders, the mechanism
responsible for seizures is still largely unknown. Epileptic seizures come
in different types and forms, and they vary from simple and partial (focal)
seizures characterized by experiences of strange sensations or auras (e.g.,
Foldvary-Schaefer & Unnwongse, 2011; Liu et al., 2017) to absence sei-
zures in which a person suddenly loses awareness for several seconds to
generalized tonic-clonic seizures, which include the loss of consciousness
and bodily stiffness (tonic stage, lasting around 20–30 s) and convulsions
(clonic stage, 20–60 s). A seizure is often called the ictal phase, which is
distinguished from the interictal phase, that is, a period between seizures
when an epileptic individual has no epileptic onset or any symptoms.
The period after the seizure is called the postictal phase; it may last from
minutes to days and, depending on the type of epilepsy, it manifests with
various symptoms, including general confusion, migraine, cognitive defi-
cits such as memory impairments, and, in some cases, also psychiatric
symptoms, such as postictal psychosis or delirium (e.g., Pottkämper et
al., 2020). Some types of seizures may also be preceded by a preictal or
prodromal phase in which abnormal EEG signals can be detected (e.g.,
Mormann et al., 2003). Some patients experience prodromic sensations
during this phase (Petitmengin et al., 2006) and often remember the char-
acter of the prodrome experience, for example, its being visual, interocep-
tive, or cognitive.
In the study conducted by Petitmengin et al. (2006, 2007; Petitmengin,
2010), a neurophenomenological method was applied to explore the pre-
ictal experience preceding epileptic seizures. The objective was to explore
the prodromic sensations of epileptic patients and to find correlations
between these experiences and EEG data. Allow me to briefly note that
Petitmengin et al. (2007, p. 747) also refer to the three methodological
components of neurophenomenology introduced by Lutz: (1) description
of the dynamics of subjective experience, (2) analysis of neural activ-
ity (EEG), and (3) establishing correlations between these two dynami-
cal structures. The core assumption is that experience is not a passive
process; it unfolds in time and includes pre-reflective processes, that is,
processes that are not evident to the subject, but the subject can become
aware of them after shifting their attention. And the brain is also con-
sidered a complex dynamical system the activity of which develops in
time. However, in Petitmengin’s (2006) version of neurophenomenol-
ogy, to capture the dynamical and pre-reflective character of experience,
a second-person method called elicitation interview is applied. During
the interviews, the interviewer asks the interviewee to evoke the stud-
ied experience and redirects the interviewee’s attention to experienced
sensations, feelings, thoughts, and so on. When the specific experience is
established, the interviewee’s attention is shifted toward the pre-reflective
structure and dynamics of this experience—for example, the succession
of sensations or modifications in the modality of the experience. Then,
in neurophenomenological analysis, a hypothetical correspondence
between this “pheno-dynamic structure” and “neuro-dynamic structure”
is established. The correspondence was thought to take the form of a
homeomorphism (Petitmengin et al., 2007, p. 755; see Chapter 2 for a
critique of homeomorphism as a constraining relation). To establish this
correspondence, DST is again utilized.
The elicitation interview was used to examine the subjective experi-
ence of the preictal phase preceding seizures. The interviewees were nine
patients with partial epilepsy. The interviews confirmed the existence of
the preictal phase of epileptic episodes on a phenomenological level; that
is, they revealed specific abnormal experiences in the period preceding a
seizure. Most of the patients reported experience of negative symptoms
such as tiredness, weakness, fragility—some reported strange bodily feel-
ings (“body was abandoning me”; “heat inside my body which rises from
my stomach to my head”; Petitmengin et al., 2006, p. 301). Importantly,
these premonitory sensations were present even a few hours before onset
and were progressive up to the seizure. This finding seems to cohere with
previous EEG studies that revealed phase-scattering (desynchronization)
of neuronal assemblies related to the epileptogenic focus from a few
minutes up to hours before a seizure (e.g., le van Quyen et al., 2001;
Mormann et al., 2003). As Petitmengin et al. (2006, p. 304) hypothesize,
the prodromic symptoms may correspond to the loss in synchronization.
Interestingly, some patients were able to recognize facilitating factors and
develop countermeasures to avoid seizures. These findings were inter-
preted as a phenomenology-guided discovery of the preictal state (Lutz,
2007) and an example of the fruitfulness of the neurophenomenological
approach, showing “mutual constraints” at work (Petitmengin, 2010).
Certainly, neurophenomenological findings based on this interview
method allow one to formulate testable hypotheses. For example, descrip-
tions of progressive sensations in the preictal phase suggest that seizures
do not begin suddenly but that there is a transition from the interic-
tal phase to ictal phase (Petitmengin et al., 2007, p. 750), that they can
be anticipated even hours before onset. Furthermore, nonlinear analysis
of EEG signals suggests that preictal changes in synchronization do not
necessarily relate to the epileptogenic focus but may also be related to
other areas of the brain. The hypothesis is supported by phenomenologi-
cal interviews in which patients reported general distress and problems
with concentration, which might be related to global dysfunctions of
brain activity (Petitmengin et al., 2007, p. 758). This example shows that
phenomenology, in this case understood as a second-person method of
analysis of experience and correlation with third-person data, delivered
hypotheses concerning the mechanism responsible for preictal symptoms
and is probably related to onset. The first hypothesis concerns temporal
and dynamical constraints, namely, that the hypothetical mechanism is
already active in the preictal phase and its functioning is progressive as
the symptoms develop until the seizure. The second hypothesis is related
to the localization of the mechanism, which is a distributed localization
rather than a simple one.
It seems, however, that calling these neurophenomenologically guided
findings a discovery (Lutz, 2007) is dubious. First, it is questionable
whether the first-person data actually constrained third-person models of
epileptic seizures in a significant fashion. Preictal decrease in synchrony—
with the earliest signs detectable up to a few hours before onset—was
discovered before elicitation interviews were carried out (e.g., Mormann
et al., 2003). Based on those findings, Michel le van Quyen et al. (2003)
developed a neuro-dynamical model in which neuronal population related
to the epileptogenic focus gradually falls out of normal large-scale syn-
chronization patterns and is therefore more susceptible to recruitment in
epileptic processes. This renders the “mutual constraints” narration ques-
tionable because it seems that there was no phenomenology-guided dis-
covery of the preictal state to begin with—it had already been discovered
when the study was conducted. At best, the phenomenological interviews
confirmed that prodromic sensations may appear earlier than previously
thought. And the idea of seizure abortion is not new either. Already in
the 19th century, the physician John Hughlings Jackson (1868) described
a case of a patient who was able to prevent seizures when experienc-
ing prodromic aura. That being said, phenomenological interviews could
contribute to epileptic patients having a better self-understanding of their
condition and thus finding their own individual abortion strategies.
Second, in subsequent papers, Petitmengin et al. (2007, p. 758;
Petitmengin 2010, p. 492) argued that structural correspondence
between pheno-dynamical and neuro-dynamical structures had roughly
been established. An intensification of negative prodromal symptoms was
thought to correspond to a progressive desynchronization in neuronal
populations. On the contrary, it is known that acute positive symptoms
(motor hyperactivity typical of seizures) result from a hyper-synchroniza-
tion of neurons that begins in the epileptogenic area. However, whether
such pheno- and neuro-dynamical correspondences may be considered a
“formal” homeomorphic relation is disputable. As I argued in Chapter 2,
the relation of homeomorphism may only be established between spaces
that share some key properties such as dimensionality. It seems that this
condition is not fulfilled in the case of the pheno-dynamical structure
identified in the prodrome and multidimensional representations of
neuro-dynamics. Petitmengin et al. (2007) argue that the
neuronal dynamic is not organized in a sequential order, as the

‘computer’ metaphor would have it. … The sequential character
is replaced by a parallel process of network synchronization. One
could take the analogy of an orchestra: suddenly, groups of distant
instruments start playing on the same rhythm.
(pp. 748–749)
But first-person descriptions of the experience, developed through the

neurophenomenological approach, actually lack genuine multidimen-
sional dynamics; their character is rather sequential or one-dimen-
sional—they take experiences to be invariant progressions of separate
phases and sensations. In the case of epilepsy, these are prodromic sen-
sations constituting the preictal phase, which turns into a seizure (ictal
phase). Of course, one can describe the phases of the experience of an
epileptic episode as attractors through which the trajectory of the sys-
tem moves in phase space. But such a description remains purely con-
ceptual if it is not formalized with a mathematical specification of the
system’s values and conditions for transitions between phases or attrac-
tors (or repellents) toward which the system tends (or which the system
avoids). It seems, therefore, that this “rough homeomorphism” merely
takes the form of a positive correlation between two hypothetical quan-
titative variables: the positivity of symptoms and the degree of syn-
chrony between neural populations (see Petitmengin et al., 2006, Fig. 1).
Naturally, positing such a hypothetical covariance is far from the neuro-
phenomenological goal of establishing exact correlations between rigor-
ous descriptions of phenomenal and neural dynamics (see Petitmengin
et al., 2007, p. 747). To the best of my knowledge, no follow-up study
that would make this correlation more credible was ever conducted,
although one was planned (p. 758). As I argued in Chapter 2 and earlier,
establishing a homeomorphic relation between the phenomenological
and neural levels seems unlikely. Formally coupling them would require
serious modeling work, in which defined experiential values and param-
eters related to first-person experiences from different phases of epilep-
tic episodes (mainly preictal and postictal) would need to be included
in equations concerning underlying neural dynamics. Such an approach
would require a dynamical-mechanistic model of epilepsy.
In fact, there are several proposals of dynamical models of the epileptic
brain from which neurophenomenology could draw in order to propose
its own model. For example, Fernando Lopes da Silva et al. (2003) pro-
posed a dynamical model of the epileptic brain, the phase space of which
consists of two stable states: interictal (normal attractor) and ictal (sei-
zure attractor). The general assumptions underlying this model state that
epilepsies are “dynamical diseases of brain systems since they are mani-
festations of the property of neuronal networks to display multistable
dynamics” (Lopes da Silva et al., 2003, p. 540) and that the epileptic
brain differs from the normal brain in that some of its control parameters
are abnormal, which makes transitions between ictal and interictal states
easier (see Figure 5.1). Three bifurcation routes are considered; first, the
transition occurs rapidly, which means that the two attractor’s basins are
close to each other and the system may switch between the two states if
the initial conditions and parameters are changed. Because the attractor
basins are so close, even random fluctuations may trigger phase tran-
sitions. This route illustrates a type of epilepsy called a nonconvulsive
absence seizure (see also Suffczyński et al., 2004), which manifests with
a sudden loss of awareness. A hypothetical mechanism responsible for
the small distance between these two states of a neural network is “low-
threshold Ca2+ channels and/or GABAA and GABAB receptors, due to
genetic and/or developmental defects” (Lopes da Silva et al., 2003, p.
541). Second, a mixed route, in which the attractor basins are distanced
but an external stimulus may lead to a transition from normal interictal
neuron behavior to an abnormal ictal state. An example is reflex epilepsy,
in which a seizure is often triggered by a visual stimulus. In the normal
brain, distributed areas are integrated through a mechanism of phase syn-
chronization (e.g., Varela et al., 2001). The hypothetical mechanism of
this route relies on the disturbance of large-scale synchronization through
an increase of phase clustering or decrease of phase dispersion at some
frequency. According to the third route of the model, the attractor basins
are distanced, but the interictal attractor gradually deforms to ictal. In
this case, the initial distance between attractors is large enough that ran-
dom fluctuations do not trigger a phase transition. The state change is
progressive and may be activated in a neural network by a change in
control parameters related to the molecular mechanisms maintaining the
Figure 5.1 P
hase space of the (A) nonepileptic and (B) epileptic brain model.
Note that in the case of the nonepileptic brain the distance between
the seizure attractor and normal attractor (concentrated basin of
attraction) is large; thus, the transition does not occur. In the case of
the epileptic brain, the distance is smaller and fluctuations of parame-
ters may lead to phase transition. Adapted from “Dynamical Diseases
of Brain Systems: Different Routes to Epileptic Seizures,” by Lopes
da Silva et al., 2003, IEEE Transactions on Biomedical Engineering,
50(5), p. 542. Copyright © 2003 by IEEE.
stability of the network caused by endogenous factors such as hormones
(Lopes da Silva et al., 2003, p. 543–544).
The models proposed by Lopes da Silva et al. (2003) and Piotr
Suffczyński et al. (2004) are further examples of dynamical-mechanistic
models that extend our understanding of the mechanisms responsible for
the transition of neural dynamics leading to seizures. These simulations
provide a means to study a greater range of parameters than possible
through empirical studies and show nonlinear interaction and coordina-
tion between neural mechanisms involved in epilepsy. These models are
also a good example of how dynamical models can tell us something
about mechanisms’ parts and their organization; that is, they include
parameters and values related to actual components of neural mecha-
nisms, such as membrane (e.g., Ca2+ currents), synaptic (e.g., thresh-
old of GABAA), and network properties (e.g., thalamocortical system;
Suffczyński et al., 2004, p. 483).
It is plausible that some of the model’s values could stand for experi-
ential states, such as prodromal sensations in the preictal phase, that is,
the progressive negative symptoms reported by patients in Petitmengin’s
(2007) study, or cognitive deficits in the postictal phase (Pottkämper et
al., 2020). Such modification of the model would, however, require quan-
tification of experiential categories.
5.5 Micro-Phenomenology—Toward Diachronic Models of

Experience
The elicitation interview method was recently reformulated as a micro-
phenomenological interview (Petitmengin et al., 2019). Whereas the prin-
ciples of conducting the interview remained the same, the understanding
of what constitutes synchronic and diachronic dimensions of experi-
ence—and the methods of their abstraction and analysis—changed.
Micro-phenomenological interview focuses on subjective singular experi-
ence and explores its pre-reflective micro-cognitive structure that, as it is
hypothesized, is shared by other subjects:
Once detected by a research team, such a structure constitutes a

hypothesis, likely to be supported or not supported by analyzing
experiences of the same type. In other words, the singular and non-
reproducible character of the content of a given experience does not
prevent the intersubjective validation of experiential structures.
(Petitmengin et al., 2019, p. 701)
In order to detect such a hypothetical structure of experience, after col-

lecting descriptions, the verbatim is prepared, and the analysis starts with
the identification of regularities of both synchronic and diachronic char-
acter. Whereas analysis of the former, that is, analysis of the content of
experience, is common to other qualitative methods, for example, inter-
pretative phenomenological analysis (Smith et al., 2009), investigating
the latter is specific to micro-phenomenological analyses, which “rely on
the Husserlian principle, resumed by Vermersch (2012), that lived expe-
rience unfolds in time, without losing sight of the hypothesis that this
temporal structure might be created and sustained by a micro-activity”
(Petitmengin et al., 2019, p. 702). To put it differently, synchronic analy-
sis concerns the structure of the content of experience at a given moment
in time, whereas diachronic analysis concerns the evolution of the experi-
ence in time.
Researchers analyzing transcripts from interviews differentiate between
statements that refer to the singular content of experience and those that
refer to its generic structure, that is, to invariant properties characteris-
tic of an experience of a specific type. The next step is identification of
descriptemes in these detected structural statements, that is, description
components of experiential events, associated with minimal phenomeno-
logical units of meaning. These basic structural elements, both synchronic
and diachronic, are distinguished through operations of abstraction and
form more general experiential categories. Three kinds of abstraction
operations are performed on the transcript material to extract experien-
tial categories and phases:
• Classification/instantiation—defines a category as a class and its

instances (tokens). For example, “localization” is an experiential cat-
egory, and “a tightness in the chest” is one of its possible individual
occurrences.
• Aggregation/fragmentation—defines a general category as composed
of subcategories. Fragmentation specifies the dimensions of which
a given higher order experiential category consists. For example,
“bodily feeling” may be perceived as having a particular temperature
occurring at a given location and so on. Vice versa, aspects of expe-
rience (subcategories) may be aggregated into a general category.
In diachronic analysis, this operation distinguishes phases and sub-
phases of experience separated by transitional events.
• Generalization/specialization—“characterizes a category as a
set of specialized categories characterized by specific properties”
(Petitmengin et al., 2019, p. 705). For example, global bodily feelings
experienced in the whole body may be quite different from local-
ized bodily feelings experienced in a specific body part, yet both
represent different examples of a general “bodily feeling” category.
Similarly, a general category of “feeling phase” may consist of differ-
ent subphases.
Abstracted categories are treated as parameters that can assume a qualita-

tive “value” (e.g., a bodily feeling may have a location or temperature),
and these specialized categories are pooled together to form a constel-
lation of experiential space at a given moment in time—the synchronic
dimension of experience. But an experience may consist of multiple
phases, and throughout the experience, the parameters of descriptive cat-
egories may covary; the categories may even emerge, fade, or transform
into one another. The evolution of experience in time is captured in dia-
chronic analysis. For example, “the operation of diachronic segmentation
makes it possible to divide the description of an experience into phases, a
phase into sub-phases, a sub-phase into gestures, and so on until the level
of very subtle micro-acts and microprocesses” (Petitmengin et al., 2019, p.
709). Phases of experience are taken to be relatively invariant successions
of experiential category configurations and their dynamically changing
values. To establish the end of one phase and the beginning of another, the
interviewer tries to identify so-called transitional events: “modification in
the subject’s experience which in turn induces significant transformations
in his/her experience” (Petitmengin et al., 2019, p. 728). Various categories
of events may be transitional: a perceptual event, a change in the intensity
of a sensation, or a train of thought ending with a consequential conclu-
sion. From the point of view of diachronic analysis, it is important that
a transitional event typically separates two phases of experience; thus, it
provides us with significant information about the diachronic structure of
experience. The next step in the analysis is the construction of a dynami-
cal line that “represents the evolution of the values of a descriptive cat-
egory throughout the different phases of an experience” (Petitmengin et
al., 2019, p. 710). For example, the category of bodily feeling may change
(e.g., modification of its intensity or localization) in successive phases of
experience identified in diachronic analysis. Thus, the dynamical line com-
bines the synchronic dimension of experience with the diachronic one.
Showing several dynamical lines, called a range, helps us understand the
overall dynamics of the studied experience—for example, some experien-
tial values may change simultaneously across all phases whereas others
may fade when a transitional event occurs. The iterative process of inter-
view and analysis also allows us to refine transitional events or identify
new ones, which, in turn, may lead to reorganization of the diachronic
structure.
What Petitmengin et al. proposed is an advanced methodology for
studying and representing the structure of experience in graphical form.
According to Petitmengin et al. (2019, pp. 724–725), an advantage of the
micro-phenomenological approach over other qualitative methods is the
testability of hypotheses concerning phenomenological structures. Several
modes of such validation are proposed, including cross-verification; that
is, the same set of descriptions is analyzed by different researchers and
their results are compared or the verification of the hypothetical struc-
ture of a studied experience by comparing it with the results of another
research team studying the same type of experience. In the context of
integrating phenomenology with neuroscience, another mode of verifi-
cation seems interesting, namely neurophenomenological validation; for
example,
when the distribution (or partition) of neuro-electrical recordings

into ‘phenomenological clusters’ according to the values of an expe-
riential category, enables the detection of distinct neural configura-
tions or ‘signatures,’ where until now only noise was perceived, this
provides a strong confirmation of [the] validity of these structures
(and therefore of the reliability of the reports from which they were
abstracted).
(Petitmengin et al., 2019, p. 725)
The problem with neurophenomenological studies, including those in a

refined micro-phenomenological approach, is that their application of
DST is insufficient to create a genuine formal dynamical model. Although
the proponents of neurophenomenology refer to DST as the key compo-
nent of their methodology, including the application of formal models
(see Lutz, 2002), its application is usually limited to dynamic descriptions
in terms such as phase space, phase transition, and so on. In the works
of Lutz, Petitmengin, and Varela, we also find schematic diagrams that,
however, represent only the general idea of a dynamical interpretation of
the target system (whether it is some type of experience or neural activity;
e.g., Petitmengin et al., 2007). These diagrams are far from phase dia-
grams or analyses of a target system’s dynamics on the basis of a formal
model. Similarly, the phases of experience identified in micro-phenome-
nological diachronic analyses are sequential entities and cannot be simply
understood as a system’s phases in a dynamical phase portrait. To draw
an actual phase diagram, one needs to establish the key variables and
parameters describing the evolution of the studied behavior in time and
find the relations between them described by equations. Furthermore,
a genuine dynamical analysis indicates a set of attractors determining
the system’s behavior in the long run, that is, the system’s tendency to
head toward a specific region in a phase space and exhibiting a specific
type of activity (e.g., steady state, limit cycle, or chaotic). Without such
advanced dynamical analyses, it seems unlikely that a relation between
pheno-dynamical and neuro-dynamical structures can be established. In
particular, as discussed in Chapter 2, it seems unlikely that it could estab-
lish a homeomorphic relation, as postulated by Petitmengin et al. (2007),
between the diachronic structure of experience (a one-dimensional suc-
cession of an experience’s phases) and the dynamics of brain processes,
which is n-dimensional (where n stands for the number of variables defin-
ing the state space representing neural network dynamics).
It seems that the goals set by neurophenomenology and micro-
phenomenology are yet to be achieved. However, even though
neurophenomenological studies have delivered less than promised so far,
neurophenomenology with its interview method can be considered a use-
ful scientific tool for gathering reliable and detailed first-person data. The
method has been used to examine the structures of particular experiential
categories, such as in the rubber hand illusion (Valenzuela-Moguillansky
et al., 2013) or sensory substitution (Kałwak et al., 2018), to obtain
richer first-person descriptions (e.g., of various forms of meditation; e.g.,
Przyrembel et al., 2019), to explore the micro-dynamics of psychedelic
experience (Timmermann et al., 2019) and the practical implications for
individual seizure prevention (Petitmengin et al., 2006), and even to coun-
teract subjects’ metacognitive fallibility (Petitmengin et al., 2013). Proper
application of DST in neuro- and micro-phenomenological method, as
originally asserted, would allow one to verify claims concerning the cor-
respondence of dynamics on the experiential and neural level and would
integrate neurophenomenology with the dynamical-mechanistic model of
explanation.
5.6 Conclusion: Toward Phenomenologically Informed

Dynamical Constraints
The methodology of studying first-person experience and the underly-
ing neural processes remains undeveloped. Neurophenomenology and
micro-phenomenology are important methodological contributions. As I
argued, neurophenomenological method relies on three key components:
phenomenological description and analysis of first-person experience,
dynamical systems theory, and neuroimaging (EEG). The first compo-
nent has been continuously developed, especially the method of phenom-
enological interview, which is today a source of important insights. The
neuroimaging techniques applied in neurophenomenology have also been
improved, for instance, intracranial EEG has been used (Petitmengin &
Lachaux, 2013). As I argued, the second component—the application of
dynamical systems theory—is still very limited, which undermines the
whole project. Development of the application of DST in neurophenom-
enology is critical for the main initial objective of the project, namely
apprehending the dynamics of first-person lived experience in relation to
the underlying distributed neural processes (Varela, 1996). The explana-
tory power of neurophenomenology lies in the development of dynamical
models and their integration with a mechanistic framework. Dynamical
models can be a powerful tool for analyzing the evolution of a target
system’s behavior, but they can also tell us something about the compo-
nents, operations, and organization of the underlying mechanism, as is
the case in dynamical-mechanistic models. But what kind of constraints,
in particular dynamical constraints, can neurophenomenology (and
other phenomenological applications in cognitive science) deliver in the
dynamical-mechanistic model of explanation?
Neurophenomenological studies of epilepsy lead to hypotheses con-
cerning the distributed localization of neural mechanisms responsible for
prodromic symptoms (Petitmengin et al., 2007). Generally speaking, it is
known that first-person data obtained in clinical interviews with epileptic
patients can inform us of the character of seizures and the localization
of the epileptic focus. Information concerning the character of auras is
particularly important. Auras are often early symptoms of the ictal phase,
and their character varies depending on the localization of the sympto-
matogenic zone in the brain. The most common type of aura is a sen-
sory-visual aura, the focus of which is localized in the occipital lobe. For
instance, a simple visual aura such as flashing or moving lights is typical
of activation of the primary visual cortex and contiguous visual associa-
tion areas (Foldvary-Schaefer & Unnwongse, 2011, p. 161). Identifying
the type of aura helps establish where the epileptogenic focus is. Nancy
Foldvary-Schaefer and Kanjana Unnwongse (2011) argue that “the local-
izing value of auras in identifying the most likely lobe of origin was found
to be as good as that of EEG and imaging” (p. 161). This can be read as
a clear example of first-person data providing constraints on the localiza-
tion of the underlying mechanism.
Various methods of phenomenological interview (e.g.,; Høffding &
Martiny, 2016; Petitmengin et al., 2007) can certainly deliver nontriv-
ial structural descriptions of a studied experience, including its dynam-
ics or diachronic structure. In particular, the micro-phenomenological
method of diachronic analysis can deliver a description of experience
phases and transitional events. The interview process also helps to access
and uncover the pre-reflective content of our mental activities. However,
the “dynamical descriptions” of experience proposed in the neuro- and
micro-phenomenological approaches are the weakest way of using DST
in cognitive science, as it only applies the conceptual apparatus of DST in
the most general sense (van Gelder & Port, 1995).
An example illustrating that stronger dynamical constraints can be
formulated on the basis of first-person investigations is research on the
visual aura of migraines (for an overview, see, e.g., Schott, 2007). The
aura is a common subjective experience related to the onset of migraines
which manifests in a visual hallucinatory experience. Illustrations made
by migraineurs play a significant role in understanding the type of experi-
ence these auras embody along with understanding the responsible mech-
anism. These graphical representations revealed that migraine auras often
have a uniform structure, which typically consists of a characteristic scin-
tillating zigzag pattern (sometimes called a fortification figure due to it
resemblance of castle fortifications) followed by a scotoma. Interestingly,
the migraine aura is a dynamical phenomenon; that is, the visual pattern
propagates and moves through the visual field. For example, the scintil-
lating pattern may appear in the center of the visual field and then grows
and moves toward the periphery where it disappears. The aura spread
typically lasts for a dozen minutes. In 1941, Karl Spencer Lashley stud-
ied the dynamics of aura and provided illustrations of his own migraine
experience showing how the aura pattern changed over time (see Figures
5.2 and 5.3). The illustrations allowed Lashley to deduce that the spatio-
temporal spread of an aura is constant and can be estimated at approxi-
mately 3 mm/min, which is consistent with other observations.
These illustrations were key to discovering the neural mechanism which
underlies the aura experience. In 1958, Peter Milner (1958) noticed that
there is a similarity between the speed of propagation of these scintillat-
ing visual patterns and the velocity of neural phenomenon called cortical
spreading depression (CSD). More recent studies (e.g., S. Silberstein, 2004)
show that there is a growing body of evidence that CSD constitutes the
underlying mechanism of the migraine visual aura.2 In short, CSD is a wave
of neuronal depolarization that moves across the cortex with the approxi-
mate speed of 3 to 5 mm/min and is marked by a brief phase of increased
cerebral blood flow followed by a longer phase of hypoperfusion. The for-
mer phase of CSD causes the positive symptom of scintillating fortification
figures, and the latter phase is associated with the negative symptom of
scotoma (Dahlem & Müller, 2003). The example of migraine aura studies
shows that a careful examination of first-person experience can be of key
importance for empirical studies and finding an underlying mechanism. In
particular, it shows that analyses of first-person data can provide dynamical
constraints on mechanistic explanation, that is, the velocity of aura spread
that corresponds to the temporal characteristics of CSD mechanisms.
It seems that neurophenomenology can also provide stronger and more
specific dynamical constraints on the space of possible mechanisms. For
Figure 5.2 S patiotemporal evolution of migraine visual aura 1. X indicates the

fixation point, and numbers represent the temporal span in minutes.
From “Patterns of Cerebral Integration Indicated by the Scotomas
of Migraine,” by K. S. Lashley, 1941, Archives of Neurology and
Psychiatry, 46(2), pp. 333–334. Copyright 1941 by the American
Medical Association. Reprinted with permission.
Figure 5.3 S patiotemporal evolution of migraine visual aura 2. X indicates the

fixation point, and numbers represent the temporal span in minutes.
From “Patterns of Cerebral Integration Indicated by the Scotomas
of Migraine,” by K. S. Lashley, 1941, Archives of Neurology and
Psychiatry, 46(2), pp. 333–334. Copyright 1941 by the American
Medical Association. Reprinted with permission.
example, phenomenological interviews with epileptic patients confirmed the

existence of the preictal phase, characterized by a specific phenomenology,
and allowed for the formulation of hypotheses about the temporal span
of the phase, that is, when the neural desynchronization begins and how it
progressively develops until the seizure (Petitmengin et al., 2007). It seems
that the neurophenomenological approach could also contribute to bet-
ter understanding the postictal phase that is also characterized by a rich
phenomenology (Pottkämper et al., 2020). Such a study would be of key
importance since the mechanism of epileptic seizure termination and tran-
sition to postictal phase is still not fully understood. An especially interest-
ing question concerns why after some seizures we can observe a postictal
generalized EEG suppression (PGES), which in some cases lead to sudden,
unexpected death in epilepsy (SUDEP). Dynamical models, such as the one
proposed by Prisca Bauer et al. (2017), allow for formulating hypotheses
concerning the nature of this mechanism. For example, one hypothesis
states that “the duration of the postictal period is related to the connec-
tivity parameter at the end of the seizure reflected by the oscillatory fre-
quency of the model” (Bauer et al., 2017, p. 661). Generally speaking, it is
hypothesized that the mechanism responsible for seizure termination aims
to restore the brain’s normal functioning, but if activated too strongly it
results in PGES. A study of postictal first-person experience could shed
new light on these underlying processes.
Taking into account these examples of constraints formulated on the
grounds of the first-person experience, it seems that neurophenomenology
has the resources to directly contribute to creating a dynamical-
mechanistic model of the experience in question and the responsible
mechanism. Building such a model would, however, require modifications
in neurophenomenological methodology, that is, moving beyond mere
dynamic descriptions and a more extensive application of DST. Descriptive
phenomenological categories the qualities of which change over time
would have to be quantified and their values, as well as their dynamics
of change, projected onto the system’s phase portrait and included in the
model’s equations. Such a methodological move would be an important
step in integrating phenomenology with cognitive science.
Notes
1 The notion of bifurcation or phase transition has also been applied to other
types of cognitive processes (see, e.g., Spivey et al., 2009).
2 As we have seen, an aura is also a symptom of an epileptic seizure. According to
some models, epileptic seizures and CSD responsible for the migraine aura have
a common underlying mechanism related to the ion concentration in neuronal
membranes and cells’ external microenvironment (e.g., Ullah et al., 2015).
References
Bauer P. R., Thijs, R. D., Lamberts, R. J., Velis, D. N., Visser, G. H., Tolner, E. A.,
Sander, J. W., Lopes da Silva, F. H., & Kalitzin, S. N. (2017). Dynamics of con-
vulsive seizure termination and postictal generalized EEG suppression. Brain,
140(3), 655–668.
neuroscience. Taylor & Francis Group.
Bechtel, W., & Richardson, R. C. (2010). Discovering complexity: Decomposition
and localization as strategies in scientific research. MIT Press.
Beer, R. D. (2003). The dynamics of active categorical perception in an evolved
model agent. Adaptive Behavior, 11(4), 209–243.
Bingham, G. P. (2004). A perceptually driven dynamical model of bimanual rhyth-
mic movement (and phase perception). Ecological Psychology, 16(1), 45–53.
Bockelman, P., Reinerman-Jones, L., and Gallagher, S. (2013). Methodological
lessons in neurophenomenology: Review of a baseline study and recom-
mendations for research approaches. Frontiers in Human Neuroscience, 7.
doi:10.3389/fnhum.2013.00608
Ciechanowski, L. (2017). Has the philosopher’s stone of the interaction between
first- and third-person data finally been found? Constructivist Foundations,
12(2), 203–205.
Dahlem, M. A., & Müller, S. C. (2003). Migraine aura dynamics after reverse
retinotopic mapping of weak excitation waves in the primary visual cortex.
Biological Cybernetics, 88(6), 419–424.
Davis, L. E., King, M. K., & Schultz, J. L. (2005). Fundamentals of neurologic
disease. Demos Medical Publishing.
Foldvary-Schaefer, N., & Unnwongse, K. (2011). Localizing and lateralizing fea-
tures of auras and seizures. Epilepsy & Behavior, 20(2), 160–166.
Golonka, S., & Wilson, A. D. (2019). Ecological mechanisms in cognitive science.
Theory & Psychology, 29(5), 676–696. doi:10.1177/0959354319877686
Haken, H., Kelso, J. A. S., & Bunz, H. (1985). A theoretical model of phase transi-
tions in human hand movements. Biological Cybernetics, 51, 347–442.
Høffding, S., & Martiny, K. (2016). Framing a phenomenological interview:
What, why and how. Phenomenology and the Cognitive Sciences, 15, 539–564.
doi:10.1007/s11097-015-9433-z
Jackson, J. H. (1868). Case of convulsive attacks arrested by stopping the aura.
The Lancet, 1, 618–619.
Jedličková, L., & Müller, M. (2019). Problems and hopes of neurophenomenol-
ogy and first-person neuroscience. e-Rhizome, 1(2): 138–159. doi:10.5507/
rh.2019.008
Kałwak, W., Reuter, M., Łukowska, M., Majchrowicz, B., & Wierzchoń, M. (2018).
Guidelines for quantitative and qualitative studies of sensory substitution expe-
rience. Adaptive Behavior, 26(3), 111–127. doi:10.1177/1059712318771690
Kaplan, D. M., & Bechtel, W. (2011). Dynamical models: An alternative or
complement to mechanistic explanations? Topics in Cognitive Science, 3(2),
438–444.
Lamb, M., & Chemero, A. (2014). Structure and application of dynamical mod-
els in cognitive science. Proceedings of the Annual Meeting of the Cognitive
Science Society, 36, 809–814.
Lashley, K. S. (1941). Patterns of cerebral integration indicated by the scotomas of
migraine. Archives of Neurology and Psychiatry, 46(2), 331–339. doi:10.1001/
archneurpsyc.1941.02280200137007
le van Quyen, M., Martinerie, J., Navarro, V., Boon, P., D’Havé, M., Adam, C.,
Renault, B., Varela, F., & Baulac, M. (2001). Anticipation of epileptic seizures
from standard EEG recordings. The Lancet, 357(9251), 183–188.
le van Quyen, M. (2003). Disentangling the dynamic core: A research program
for a neurodynamics at the large-scale. Biological Research, 36(1), 67–88.
Liu, Y., Guo, X. M., Wu, X., Li, P., & Wang, W. W. (2017). Clinical analysis of
partial epilepsy with auras. Chinese Medical Journal (Eng.), 130(3), 318–322.
doi:10.4103/0366-6999.198918
Lopes da Silva, F. H., Blanes, W., Kalitzin, S. N., Parra, J., Suffczyński, P., & Velis,
D. N. (2003). Dynamical diseases of brain systems: Different routes to epileptic
seizures. IEEE Transactions on Biomedical Engineering, 50(5), 540–548.
Lutz, A. (2002). Toward a neurophenomenology as an account of generative passages:
A first empirical case study. Phenomenology and the Cognitive Sciences, 1, 133–167.
Lutz, A. (2007). Neurophenomenology and the study of self-consciousness.
Lutz, A., Lachaux, J. P., Martinerie, J., & Varela, F. J. (2002). Guiding the study
of brain dynamics by using first-person data: Synchrony patterns correlate
with ongoing conscious states during a simple visual task. Proceedings of the
National Academy of Sciences, 99(3), 1586–1591.
Lutz, A., & Thompson, E. (2003). Neurophenomenology integrating subjective
experience and brain dynamics in the neuroscience of consciousness. Journal of
Consciousness Studies, 10(9–10), 31–52.
Milner, P. M. (1958). Note on a possible correspondence between the scotomas
of migraine and spreading depression of Leão. Electroencephalography and
Clinical Neurophysiology, 10(4), 705.
Mormann, F., Kreuz, T., Andrzejak, R. G., David, P., Lehnertz, K., & Elger, C.
E. (2003). Epileptic seizures are preceded by a decrease in synchronization.
Epilepsy Research, 53(3), 173–185.
Petitmengin, C. (2006). Describing one’s subjective experience in the second per-
son: An interview method for the science of consciousness. Phenomenology and
the Cognitive Sciences, 5(3–4), 229–269. doi:10.1007/s11097–006–9022–2
Petitmengin, C. (2010). A neurophenomenological study of epileptic seizure
anticipation. In S. Gallagher & D. Schmicking (Eds.), Handbook of phenom-
enology and cognitive science (pp. 471–499). Springer.
Petitmengin, C., Baulac, M., & Navarro, V. (2006). Seizure anticipation: Are neu-
rophenomenological approaches able to detect preictal symptoms? Epilepsy &
Behavior, 9(2), 298–306.
Petitmengin, C., & Lachaux, J.-P. (2013). Microcognitive science: Bridging expe-
riential and neuronal microdynamics. Frontiers in Human Neuroscience, 7.
doi:10.3389/fnhum.2013.00617
Petitmengin, C., Remillieux, A., Cahour, B., & Carter-Thomas, S. (2013). A gap in
Nisbett and Wilson’s findings? A first-person access to our cognitive processes.
Petitmengin, C., Remillieux, A., & Valenzuela-Moguillansky, C. (2019). Discovering
the structures of lived experience: Towards a micro-phenomenological analysis
method. Phenomenology and the Cognitive Sciences, 18(4), 691–730.
Pottkämper, J. C. M., Hofmeijer, J., van Waarde, J. A., & van Putten, M. J. A. M.
(2020). The postictal state—What do we know? Epilepsia, 61(6), 1045–1061.
Przyrembel, M., Vrticka, P., Engert, V., & Singer, T. (2019). Loving-kindness med-
itation—A queen of hearts?: A physio-phenomenological investigation on the
variety of experience. Journal of Consciousness Studies, 26(7–8), 95–129.
Roskies, A. (1999). The binding problem: Special issue. Neuron, 24, 7–125.
Schott, G. D. (2007). Exploring the visual hallucinations of migraine aura: The
tacit contribution of illustration. Brain, 130(6), 1690–1703.
Silberstein, M., & Chemero, A. (2013). Constraints on localization and decom-
position as explanatory strategies in the biological sciences. Philosophy of
Science, 80(5), 958–970.
Silberstein, S. D. (2004). Migraine. The Lancet, 363(9406), 381–391.
Simon, H. A. (1962). The architecture of complexity. Proceedings of the American
Philosophical Society, 106(6), 467–482.
Smith, J. A., Flowers, P., & Larkin, M. (2009). Interpretative phenomenological
analysis: Theory, method and research. Sage.
Spivey, M. J., Anderson, S. E., & Dale, R. (2009). The phase transition in human
cognition. New Mathematics and Natural Computation, 5(1), 197–220.
Sporns, O. (2011). Networks of the brain. MIT Press.
Stinson, C. (2016). Mechanisms in psychology: Ripping nature at its seams.
Synthese, 193, 1585–1614.
Suffczyński, P., Kalitzin, S., & Lopes Da Silva, F. H. (2004). Dynamics of non-
convulsive epileptic phenomena modeled by a bistable neuronal network.
Neuroscience, 126(2), 467–484.
Timmermann, C., Roseman, L., Schartner, M., Milliere, R., Williams, L. T.,
Erritzoe, D., & Turton, S. (2019). Neural correlates of the DMT experience
assessed with multivariate EEG. Scientific Reports, 9(1), 1–13.
Ullah, G., Wei, Y., Dahlem, M. A., Wechselberger, M., & Schiff, S. J. (2015). The
role of cell volume in the dynamics of seizure, spreading depression, and anoxic
depolarization. PLOS Computational Biology, 11(8). doi:10.1371/journal.
pcbi.1004414
Wimsatt, W. C. (1986). Forms of aggregativity. In B. Donagan, A. N. Perovich, Jr.,
& M. V. Wedin (Eds.), Human nature and natural knowledge: Essays presented
to Marjorie Grene on the occasion of her seventy-fifth birthday (pp. 259–291).
Springer.
Valenzuela-Moguillansky, C., O’Regan, J. K., & Petitmengin, C. (2013). Exploring
the subjective experience of the “rubber hand” illusion. Frontiers in Human
Neuroscience, 23. doi.org/10.3389/fnhum.2013.00659
van Gelder, T. (1999). Wooden iron? Husserlian phenomenology meets cognitive
science. In J. Petitot, F. J. Varela, B. Pachoud, & J.-M. Roy (Eds.), Naturalizing
Explorations in the dynamics of cognition (pp. 1–43). MIT Press.
Varela, F. (1996). Neurophenomenology: A methodological remedy for the hard
problem. Journal of Consciousness Studies, 3(4), 330–349.
Varela, F. (1999). The specious present: A neurophenomenology of time conscious-
ness. In J. Petitot, F. J. Varela, B. Pachoud, & J.-M. Roy (Eds.), Naturalizing
Varela, F., Lachaux, J. P., Rodriguez, E., & Martinerie, J. (2001). The brain-
web: Phase synchronization and large-scale integration. Nature Reviews
Neuroscience, 2, 229–239. doi:10.1038/35067550
Vermersch, P. (2012). Explicitation et phénoménologie. PUF.
Zednik, C. (2011). The nature of dynamical explanation. Philosophy of Science,
78(2), 238–263.
Zednik, C. (2019). Models and mechanisms in network neuroscience.
Philosophical Psychology, 32(1), 23–51.
6 Conclusion
Toward Methodologically Guided
Mutual Constraints
This book considered the prospects of methodological and theoretical

integration of phenomenology with the mechanistic explanatory frame-
work in cognitive science. I presented phenomenology primarily as a the-
ory of consciousness and method of eidetic investigation. This moderate
concept of phenomenology, which complements the Husserlian idea of
phenomenological psychology, plays down the transcendental dimension
of phenomenological philosophy and focuses on potential phenomeno-
logical contributions to other research fields in cognitive science. As such,
phenomenology can make a valuable contribution to studying multifac-
eted mental phenomena, delivering structural descriptions of first-person
experience, and providing analyses of the experience in question in order
to explicate its invariant (eidetic) properties. According to this concep-
tion, phenomenology differs from introspection, is not anti-naturalistic,
and the problem of qualia is not its primary concern. On the contrary,
phenomenology is interested in studying invariant generic structures of
experience, rather than particular experiences, and introduces the con-
cepts of these structures into empirical research. The proposed notion of
phenomenology is, as I argued, amenable to naturalization understood as
integration with the explanatory framework of cognitive science.
Reviewing the debate about naturalizing phenomenology and discuss-
ing various proposals of naturalized phenomenology revealed that all
these proposals suffer from deficiencies. The constraints in the proposals
which were thought to relate phenomenology to other research fields in
cognitive science proved either too weak (e.g., conceptual constraints)
or implausible (e.g., isomorphism). Furthermore, the idea of naturalizing
phenomenology was coined in opposition to reductive and computational
approaches to the mind and thus did not include more recent discussions
of nonreductive integration of cognitive science or multilevel mechanistic
explanations (e.g., Craver, 2007). Thus, rethinking the project of natural-
ization required addressing various explanatory models applied in cogni-
tive science as well as the project of explanatory integration. By assessing
explanatory frameworks, I have shown that each of these models has
DOI: 10.4324/9781003035367-9
certain limitations, and each of them is incapable of providing a complete
explanation of complex mental phenomena. Mechanistic integration of
cognitive science provides a remedy (Craver, 2007), according to which
various research fields contribute to a complete multilevel explanation
by providing constraints on the space of mechanisms. In this context,
the type of explanation phenomenology offers, if any, was also consid-
ered. The conclusion was that phenomenology does not provide genuine
explanations but rather gives us a constitutive understanding of mental
phenomena. It was hypothesized that phenomenological understanding
can provide constraints and thus contribute to multilevel mechanistic
explanations. In particular, two types of constraints were considered,
namely, functional and dynamical constraints.
In searching for functional constraints, we went back to Husserlian
phenomenology and his idea of functional phenomenology (Husserl,
1982), which introduced the notion of constitutive functions of con-
sciousness, that is, functions involved in the production of intentional
experience. It was argued that functional phenomenology relies on a
method of phenomenological decomposition, which shares similarities
with the analytical method of functional decomposition applied in psy-
chological explanations (Cummins, 1975). Much like functional analy-
sis, phenomenology can provide general characterizations of cognitive
functions constitutive for the target experience. By considering examples
of vision studies and a naturalistic account of the first-person perspec-
tive, it was shown that such a characterization can be further specified in
functional models, which, in turn, can constrain how-possibly models of
underlying mechanisms.
Considering phenomenological-dynamical constraints required reject-
ing the purported opposition between dynamical and mechanistic
approaches and instead thinking in terms of their integration. Dynamical-
mechanistic explanations address both the dynamics of the behavior in
question and the components of the underlying mechanisms (e.g., Bechtel
& Abrahamsen, 2010). Assessing the application of dynamical systems
theory in neurophenomenological studies (e.g., Lutz, 2002), in particu-
lar in studies of epilepsy (e.g., Petitmengin et al., 2007), I argued that
it is unsatisfying and limited to mere dynamic descriptions of the tar-
get phenomenon, and thus, it cannot genuinely constrain the space of
mechanisms.
The case of dynamical analyses of epilepsy providing stronger con-
straints on mechanisms was shown on the example of a dynamical-
mechanistic model of the epileptic brain (Lopes da Silva et al., 2003).
Discussing this model, I suggested that it could be modified in order to
include experiential parameters related to subjective experience of the
preictal or postictal phase. Such a phenomenological revision of the model
would be an example of mutual constraints between first-person analyses
of experience, third-person neuroscience, and dynamical modeling. The
Conclusion 195
history of studies concerning migraines and visual auras is also an illus-
trative example, one that suggests that analysis of first-person experience
may provide key information about the dynamics of the phenomenon in
question and thus lead to the discovery of underlying mechanisms.
Integrating phenomenology with the mechanistic framework can be
conceived of as a new approach to naturalization, which, as I argued,
is nonreductive and leaves integrated fields a certain dose of autonomy.
Generally speaking, mechanistic integration of cognitive science does not
offer one elegant, unified theory, but rather, it creates a patchwork of var-
ious research fields involved in the study of cognitive phenomena. Thus,
this way of naturalizing phenomenology does not assume epistemologi-
cal continuity between phenomenology and the natural sciences and does
not attempt to create a unified theory of cognition. Nor does it aim to
reduce phenomenology to other research fields, in particular to neuro-
biology. Such reductionism would run contrary to the ideas of explana-
tory pluralism, interfield integration, and the conviction that explanation
of mental phenomena requires the cooperation of multiple approaches
addressing various aspects (and levels) of the same target phenomenon.
In the proposed approach to naturalization, phenomenology is also not
merely applied to improve experimental design, as it is in front-loaded
phenomenology (e.g., Gallagher, 2010). In a nutshell, the idea of natu-
ralization through mechanistic integration has it that phenomenology
can contribute to building multilevel explanatory models by providing
intralevel constraints on the space of mechanisms, that is, by providing
information limiting the space of possibilities of mechanistic solutions.
As I argued, the character of the constraints that phenomenology can
provide is twofold: functional constraints concern constitutive functions
of consciousness involved in the production of experience and dynamic
constraints concern the dynamical landscape of experience.
6.1 Constraining Phenomenology
The earlier considerations mainly concerned theoretical and methodolog-
ical grounds for the integration of phenomenology with the mechanistic
framework, in particular how phenomenology can constrain explanatory
mechanistic models. But the relation between the mechanistic framework
of cognitive science and phenomenology is not one-directional. Thus it is
now important to consider what phenomenology gains from this integra-
tion, how it is constrained, and whether we can think of this relation as
“mutual enlightenment” (e.g., Gallagher, 2010).
To allow for this mutual relation between phenomenology and cog-
nitive science, we have to acknowledge that phenomenological claims
are not infallible. The infallibility of phenomenological theories was
often a part of the critique of phenomenology, stressing the opposition
between the phenomenological approach and the natural sciences, which
is grounded in the ideas of falsifiability and corroboration. But it is a
critique of phenomenology that is unsupported. Husserl himself admit-
ted that phenomenological evidence is fallible, even if it is considered
apodictic. For example, as he writes in Formal and Transcendental Logic,
the possibility of deception is inherent in the evidence of experience

and does not annul either its fundamental character or its effect. …
Even an ostensibly apodictic evidence can become disclosed as decep-
tion and, in that event, presupposes a similar evidence by which it is
‘shattered.’
(Husserl, 1969, p. 156)
To put it differently, we can err in our eidetic intuitions about the nature
of experience, and these errors can be detected by confronting them with
the evidence of a new experience. Confirmation or falsification of eidetic
claims can also be achieved on other grounds. The phenomenological
description in question can be confirmed or falsified through intersubjec-
tive corroboration, that is, by comparing it with others’ descriptions of
the same type of experience (Gallagher & Zahavi, 2012) or by confront-
ing the description with empirical evidence (e.g., Gallagher & Brøsted
Sørensen, 2006). An interesting historical example of how a phenom-
enological theory was confirmed by empirical research is discussed by
Philipp Berghofer (2020). As he argues, phenomenological investigations
of illusory and hallucinatory experience lead Husserl to the key distinc-
tion between perceptual states and judgments. Accordingly, perceptual
experience representing an object O does not imply having a belief that
O. In other words, one can have a perceptual experience that is incon-
gruent with their judgments about the perceptual object. According to
Berghofer, this phenomenological distinction was later supported by
empirical research conducted by Vittorio Benussi, who was a member of
Meinong’s school of experimental psychology in Graz. Benussi studied
various optical illusions, such as the Müller-Lyer illusion, and how sub-
jects react to them. His conclusions were that optical illusions should be
considered inadequate representations rather than errors of judgment and
that perception, in general, is an experience of a perceptual object’s pres-
ence, which does not involve judgments, although it may lead to them.
Thus, Berghofer constates, experimental research led Benussi to conclu-
sions similar to those that can be found in Husserlian phenomenology.
Benussi provides a clear example of convergence between phenomeno-
logical claims and the results of experimental research. But as Gallagher
(2010) rightly remarks,
generally, we should not think of experiments as simply accepting

the phenomenological description. Rather they can test and verify
that description, and either extend its application, or send it back to
Conclusion 197
the phenomenological drawing board. In such cases we should have
a mutual enlightenment between phenomenology and the empirical
sciences.
(p. 32)
To put it simply, experimental research should test phenomenologi-

cal claims and, if they fail, motivate phenomenologists to revise their
theories. For instance, Gallagher and Brøsted Sørensen (2006) discuss
empirical research on the sense of bodily agency and ownership. These
two phenomenological concepts were applied to investigate how sub-
jects evaluate their agency with respect to task-related bodily movements.
According to Gallagher, experimenting with sense of agency expands
phenomenological analysis in several respects, for example, that the sense
of agency is sensitive to temporal discrepancy between perception and
proprioception or that the phenomenon of agency is closely related to
social cognition. These experimental results push phenomenologists to
rethink and revise the phenomenological concept of agency, which again
can be applied to experimental design.
How can we rethink the relationship between phenomenology and
cognitive science in the context of the proposed naturalization through
mechanistic explanatory integration? Generally speaking, mechanistic
explanations aim to create multilevel models of the mechanism respon-
sible for the explanandum phenomenon. The model is not a result of
tidy deductive reasoning but a progressive process of gathering con-
straints of various sorts from different research fields. In Chapter 4, I
argued that phenomenological analysis can be understood as analogous
to functional analysis; that is, just as in functional analyses, phenomenol-
ogy can decompose the target experiential capacity and propose a func-
tional sketch of it that may be useful in developing a mechanistic model
(Piccinini & Craver, 2011). It is clear that different functional organiza-
tions of the system performing the same capacity are possible. Thus, the
sketch lacks explanatory power unless it is supplemented with mecha-
nistic details and functions mapped onto components of the mechanism
which are plausibly responsible for the target capacity. In other words,
filling the sketch in with information about the mechanism’s components
verifies whether the functional sketch is plausible. I propose to think
about testing phenomenological claims as analogous to the verification of
a functional design. Accordingly, mechanistic models can tell us whether
functional decomposition, through Cummins-style functional analysis or
phenomenological analysis, is plausible and finds support in simulations
or empirical evidence.
The issue of simulation raises yet another possibility, namely how
dynamical-mechanistic models and simulations relate to and constrain
phenomenological theories. My general and rather speculative answer is
that they can play the role of extensions of the phenomenological method
of imaginary variation and lead to new eidetic intuitions. As I mentioned
in Chapter 3, phenomenological eidetic claims can be falsified in the pro-
cess of imaginary variation, that is, by construing in phantasy possible
variants of the object of experience (Sowa, 2012). But the procedure of
imaginary variation is limited and depends on the phenomenologist’s
individual predispositions and epistemic intuitions. It seems that Husserl
also recognized this issue. For example, in the third book of Ideas he
discusses the limitations of our imaginary intuitions and considers how
artificial arrangements in psychological experiments can evoke new intu-
itions and eidetic insights. As he writes,
the way the external arrangements function here is in principle com-

pletely of the same sort as those which the geometers use. The beauti-
ful collections of models that our mathematical institutes keep serve
fruitful investigation and teaching, just as well as the drawings on the
board and on paper. In principle they achieve no more than the geo-
metrical phantasy over into which they are obviously simultaneously
called to lead one; they serve indeed the grasping of essence, but they
excite exemplary intuitions, and that has the advantage spoken of.
(Husserl, 1980, p. 45)
In contemporary cognitive science, dynamical-mechanistic models

can play an analogous role to the mathematical models mentioned by
Husserl; that is, such models allow seeing experiential variants that are
inaccessible to individual imagination. To put it differently, dynamical-
mechanistic models that contain information about first-person expe-
rience can reveal new aspects of experience. Furthermore, simulations
can test phenomenological generalizations such as those concerning the
dynamics of experience. For example, a simulation of change in experi-
ential state represented in a model as a phase transition may reveal that it
happens gradually rather that instantly and thus may lead to revision of
the corresponding phenomenological description.
By acknowledging that there are mutual constraints between phenom-
enology and mechanistic explanations in cognitive science, we have to
accept that phenomenology, understood as a theory of consciousness and
method of investigating mental phenomena, is not fully autonomous (of
course, this does not apply to transcendental phenomenology which is
a purely philosophical endeavor unconstrained by the natural sciences).
As I argued in Chapter 3, explanatory autonomy comes in degrees.
Phenomenology is autonomous in choosing its own method of investi-
gation as well as in applying its own terminology. But the very fact that
other disciplines also investigate the same sort of mental phenomena,
although from different perspectives and with different methods, limits
phenomenological autonomy. To put it briefly, phenomenology has to
reckon with evidence from other research fields involved in cognitive
Conclusion 199
science and accept that this evidence can lead to a revision of phenom-
enological theories. That being said, I do not consider this limitation of
autonomy a sort of reductionism. In fact, thinking in terms of integrative
explanations implies that many disciplines in cognitive science, including
psychology, are not fully autonomous.
Finally, introducing phenomenology, or some other type of investiga-
tion of first-person experience, to consciousness studies seems necessary
for its progress. Without some kind of perspective from “the inside,”
without accounting for first-person experience, we will remain like the
blind men describing the elephant, at best investigating some surface
phenomena without any understanding of the deeper structures of con-
sciousness. I believe that the successful integration of accounts of first-
person experience, especially of phenomenological origin, may lead to a
genuine scientific account of consciousness in all its complexity.
References
Computational modeling of circadian rhythms as an exemplar for cogni-
tive science. Studies in History and Philosophy of Science Part A, 41(3),
321–333.
Berghofer, P. (2020). Phenomenology and experimental psychology: On the pros-
pects and limitations of experimental research for a phenomenological episte-
mology. Journal of Transcendental Philosophy, 1(1), 85–108.
741–765.
Gallagher, S., & Brøsted Sørensen, J. (2006). Experimenting with phenom-
enology. Consciousness and Cognition, 15(1), 119–134. doi.org/10.1016/j.
concog.2005.03.002
Gallagher, S. (2010). Phenomenology and non-reductionist cognitive science. In
D. Schmicking & S. Gallagher (Eds.), Handbook of phenomenology and cogni-
tive science (pp. 21–34). Springer.
Routledge.
Nijhoff.
Lopes da Silva, F. H., Blanes, W., Kalitzin, S. N., Parra, J., Suffczyński, P., & Velis,
D. N. (2003). Dynamical diseases of brain systems: Different routes to epileptic
seizures. IEEE Transactions on Biomedical Engineering, 50(5), 540–548.
1(2), 133–167.
Sowa, R. (2012). Eidetics and its methodology (E. A. Behnke, Trans.). In S.
Luft & S. Overgaard (Eds.), The Routledge companion to phenomenology
(pp. 254–265). Routledge.
Index
Page numbers followed by ‘n’ refer to note numbers.
Abraham, R. H. 88 bearing map 157

Abrahamsen, A. 48, 95, 108–109, Bechtel, W. 48, 77, 90, 95–108,
169, 171 114–115, 168–171
Abramova, E. 99 Beer, R. D. 88, 91–92, 100, 109, 170
absence seizure 175, 179–180 Benoist, J. 163n3
affectivity 20, 152–153 Berghofer, P. 196
affordances 47–48, 170 Bermúdez, J. 79
analysis: diachronic 34, 181–182, 185; Bickle, J. 42, 114
dynamical 88, 93, 172, 174, 183; bifurcation 88, 172–173, 179
functional 83–87, 116–117, 139– Bingham, G. 170
140, 145, 148–150, 156, 162; Biven, L. 157
interpretative phenomenological 181; Block, N. 83, 139, 142
logical 140; phenomenological 22, Bockelman, P. 172
34–36, 127, 149; synchronic 34, 181 boxology 85, 148–149
Andreasen, N. 74 brain-centrism 47–48
A-not-B error 91 Brentano, F. 119–120
anticipation-fulfillment model of Brøsted Sørensen, J. 53, 144, 197
vision 160–162
anti-naturalism 13, 28–31, 44–45 Campbell, J. 35, 74
anti-representationalism 90 causality 23, 100, 120, 124, 142, 146
Ashburner, M. 56 causal-role 145
attitude: natural 15–16, 18, 21, 46, Centrone, S. 146–147, 163n3
124; personalistic 120; Chalmers, D. J. 26–27, 31, 99
phenomenological 15, 18, 28, 45, Chaminade, T. 53
124, 141, 150 Chemero, A. 75, 88, 90, 168–169
attractor 65, 88, 172–173, 178; Chrudzimski, A. 67n3
seizure 178–179 Churchland, P. S. 42, 114
aura 174–177, 185–188 Ciechanowski, L. 172
circadian rhythms 48, 117, 169
Baddeley, A. 86 clarification (Aufklärung) 119–120
Balduzzi, D. 27, 31 Clark, A. 99, 112, 161–162
Bangu, S. 73, 111 cognitive map 157–158, 163
Barker, M. 112 Colombo, M. 34–35, 79, 110
Barrett, D. 86 comparative biology 157
Bauer P. 188 computationalism 85–86, 139–145,
Baumgartner, M. 129n2 149, 154
202 Index
consequence-etiology 85 error: of misidentification 33;
constraints 4–5, 43–44, 51–67; underspecification 32–33
conceptual 52–56; dynamical 167, essence 16–17; formal and material
177, 184–188; functional 156, 161, 29, 147–148; inexact 29, 148; of
194; homeomorphism 63–66; perception 22, 24–25
homomorphism 62; isomorphism essential seeing 16, 25, 29, 124, 126;
57–60; on space of mechanisms eidetic 21
114–116 Evans, G. 156
Cornsweet effect 57 evolutionary-developmental biology
Cornsweet, T. N. 57 7, 156
corroboration 119, 144, 196 explanation: causal 58, 73–74, 120;
Cosmides, L. 108 componential 94, 107;
coupled oscillators 104 constitutive 79, 94, 107;
Craver, C. F. 3–5, 32, 35, 67, 73, 75, deductive-nomological 75–78;
85–86, 94–118 dynamical 87–94; eidetic 21, 123;
Cummins, R. 7, 48, 77, 84–87, 117, epistemic 74; functional 83–87;
139, 149–150, 155, 197 mechanistic 94–110; modal 74;
naturalistic 24, 30, 67, 158; ontic
Dahlem M. 186 74; personal 78–83;
Darden, L. 67, 113, 115 phenomenological 118–124; and
Davis, L. 175 phenomenology 120–122;
decomposability 4, 97–98, 109, reductive 30–31, 46, 113
167–168, 170–171 explanatory: autonomy 116–118;
decomposition 96–98, 109; functional demarcation 107; gap 31–32, 60,
85–86, 96; phenomenological 172; integration 110–116; pluralism
150–153; structural 96 112–114
Dennett, D. 24, 78–83, 90, 120 explication interview 64
Descartes, R. 94 extended mind hypothesis 99
de Warren, N. 20
dispositional: property 86, 149; Felleman D. 98
regularity 84; terms 78, 86–87 Fernandez, A. V. 55
distributed cognition 99–100 Fodor, J. 66, 85, 90, 97, 140, 142–145
Drayson, Z. 78 Foldvary-Schaefer, N. 175, 185
Dreyfus, H. 52, 139–145 Føllesdal, D. 141, 144
Drummond, J. J. 163n2 foundationalism 45
Dupré, J. 66, 111 fragmentation (Zerstückung) 151;
dynamical hypothesis 89–90 microphenomenology 181
dynamical line 182 free-energy principle 112
dynamical neural signature (DNS) 174 Friston, K. 112
dynamical stance 90 Fuchs, T. 55
Dynamical Systems Theory 50–51, function: constitutive 150, 156, 158,
60–62, 87–90, 169–172, 184, 194 161, 194; etiological 83–84;
intentional 16, 19–20, 122, 140,
ecological information 170 146–148, 150, 152, 154; procedural
Eliasmith, C. 91 146–147; proper 83–84, 87, 96;
Elicitation interview 64, 175–177, 180 representational 150, 152; systemic
embodiment 20, 27, 30, 37n2, 47, 153 83–84
emergence (and mechanism) 107 functional: phenomenology 7, 140,
empathy (Einfühlung) 26, 121–122 145–148; sketch 117, 161, 197;
enactivism 47 state 83, 139, 142
Engel, G. 34, 118 functionalism 6–7, 83, 85, 112, 129;
epilepsy 33, 174–180 and phenomenology 139–145
epoché 15–17, 146 functionalist naturalization 153–154
Index 203
Galileo 125 interictal phase 175–185
Gallagher, S. 37n2, 44, 46–48, 51, interlevel relations 106–107
53–54, 57, 119, 144, 195–197 interpretative phenomenological
generalization 73, 76, 93–94, 101, analysis 181
108–109, 119–120, 123–124; intersubjectivity 15, 37n2, 46
microphenomenology 181 introspection 24–26, 37n4, 80–81
generative passages 60–63, 172 intuition (Anschauung) 119, 122
Glennan, S. 4, 74–75, 96, 102–104, Irvine, E. 32–33
129n2 isomorphism 57–60
Godfrey-Smith, P. 42, 48, 77, 83, 114
Golonka, S. 170–171 Jack, A. 80
graph theory 169 Jackson, J. H. 177
Grobler, A. 74 Jacobs, L. 157
Gurwitsch, A. 163n2 Jedličková, L. 172
Haken, H. 91, 93, 103, 170 Kałwak, W. 33, 184

Halina, M. 107 Kaplan, D. M. 48, 108, 117, 171
Hall, H. 139–145 Kelso, J. A. S. 50, 88, 91, 93, 170
Hastings J. 56 Kendler, K. S. 35, 110, 118
Heidegger, M. 13, 34, 44 kinesthesis 19, 152–153
Heinz, A. 34, 110 Kockelmans, J. J. 17, 21–23, 37n7,
Hempel, C. 73, 75, 84 46, 143
Hensel, W. 114 Komorowska-Mach, J. 80
heterophenomenology 80–83 Kortooms, T. 20, 153
Hipólito, I. 48 Kriegel, U. 27
HKB model 91, 93, 103, 170 Krueger, J. 25
Hodgkin, A. 93–94, 104
Hodgkin and Huxley’s model of Lakatos, I. 112
neuron 93–94, 104 Lamb, M. 168
Høffding, S. 33, 55, 185 La Mettrie, J. O. de 94
Hohol, M. 112, 114 Landgrebe, L. 14, 153
Hohwy, J. 48, 112, 161 language of thought 90, 142
homeomorphism 63–66, 176–178 large-scale synchronization 59, 63,
homology 157 173, 177, 179
homomorphism 62 Larsen, R. R. 56
Hurlburt, R. 80 Lashley, K. 186
Husserl, E. 2–3, 13–37, 49–52, law: bridge 105, 111; ceteris
118–123, 139–153, 196–198 paribus 23, 101, 120; eidetic
Hutchins, E. 99 17–18, 23, 29, 50, 123–124;
formal 23, 51, 124; nature 35,
ictal phase 175–185 48, 73–76; phenomenological 51,
idealization 103, 107, 112 124, 148
ideal of completeness 107–108 le van Quyen, M. 65, 176–177
identification free self-reference levels of mechanisms 104–107
156, 158 Levine, J. 26, 31
imaginative variation 15–17, 21, Lingis, A. 122
25, 122 Lipton, P. 125–126
Ingarden, R. 67n3 Liu, Y. 175
integrated information theory 26–27 Livingston, P. 140
intentionality 18, 21, 33, 143, 145, localization: explanatory strategy
148, 150, 159 98–99; heuristic 109, 168
intentional stance 78, 81, 90, 121 locus of control 99
interfield theory 110, 113, 115 long-term potentiation 104, 114, 116
204 Index
Lopes da Silva, F. 178–180 naturalization 3, 6, 14, 44–49, 128,
Lotz, C. 37n2 153–156
Lutz, A. 8, 57–63, 173–177, 183, 194 near-decomposability 168
Neisser, J. 4, 27, 156–158
Madary, M. 7, 160–162 network neuroscience 169
Marbach, E. 54–55 neurophenomenology 32, 60–66,
Martiny, K. 33, 55, 185 171–178, 183–184
Maull, N. 67, 113, 115 Newell, A. 80
McDowell, J. 78–80 Nielsen, T. 53
McIntyre, R. 7, 139–145 Nisbett, R. 80
mechanism: model 102–104; noema 18–19, 140–145
nondeterministic 101; philosophy noesis 18–19, 140–145
94–95; scheme 102, 117; sketch nonlinear oscillator 172
102, 110, 112, 117–118; social
99, 101 Oizumi, M. 26
mechanistically mediated effects 107 olfactory spatial hypothesis 157
mechanistic property cluster 35, 110 Olivares, F. 63
Mendel’s first law 77 ontological neutrality 142, 154
mental causation 146 Oppenheim, P. 47, 66, 75, 105, 111
mental functioning ontology 56 Oullier, O. 93
mental representation 140–145, 152 Overgaard, S. 25
Merleau-Ponty, M. 13, 16, 37n2, 44, 53
mesolimbic system 158 Panksepp, J. 157
method: first-person 33, 59–60, 82; Parnas, J. 33, 55–56
reductive see reduction; second- passive synthesis 121–122, 152
person 33, 175–176 Petitmengin, C. 8, 33–34, 63–66,
methodological solipsism 142–143 163n4, 175–178, 180–184
micro-phenomenology 8, 43, 64–65, Petitot, J. 3, 43–44, 50
163n4, 180–185 phase portrait 88, 90, 183, 188
Miłkowski, M. 94, 100, 112, 114 phase scattering 63, 176
Millikan, R. 87, 96 phase space 174, 178–179, 183
Milner, B. 116 phase synchrony 173, 179
Milner, P. 186 phase transition 179, 183 see also
Mitchell, S. D. 66, 112–113 bifurcation
model: computational 51–52, 117, phenomenological: attitude 15,
168–169; dynamical 51, 88, 90–94, 18, 28, 45, 141, 150; cluster
100, 104, 171–174, 178–180; 174, 183; interview 33, 55, 61,
explanatory see explanation; 163n4, 176–177; method
how-actually 102–103; how- 14–17, 126 see also
possibly 102–103; how-roughly imaginative variation;
102, 104; mechanistic see psychiatry 55–56, 118;
mechanism; phenomenal 103, 107 understanding 124–127
Moran, D. 3, 14, 28, 45, 54, 66, 155 phenomenology: front-loaded 43,
Mormann, F. 175–177 52–54, 56, 119, 195; genetic
Moskalewicz, M. 33–34 19–20, 121, 150–153; hyletic 27,
motivation 22–23, 119–121, 146, 146–147; minimal 46;
148, 150, 152 naturalized see naturalization;
multiple realizability 108, 111 static 19, 150; transcendental 14,
mutual manipulability 100, 129n2 17, 20–23, 37n3, 45, 124, 128,
143–144, 153–155
Nagel, E. 72, 83–84, 118 Piccinini, G. 25, 80, 82–83, 85,
Nagel, T. 26 116–117, 139, 155
naturalism 28–31, 42–46 Piekarski, M. 162
Index 205
Płotka, W. 17, 37n3, 67n3, 154 Rosenthal, D. 74
Port, R. 50, 65, 88, 93, 172, 185 Roskies, A. 173
postictal phase 175–180, Roth, M. 86, 117
187–188, 194 Roux, S. 94
Pottkämper, J. 175, 180, 187 Roy J-M. 3, 14, 32, 43–44, 49–50, 57,
predictive coding 48, 112, 161–163 61–62, 154
predictive processing see predictive rubber hand illusion 184
coding Ruiz-Mirazo, K. 102
predictivism 93, 169 Rupert, R. D. 24
preictal phase 175–180, 187, 194
primal impression 20, 152 Salmon, W. C. 47, 72, 74–76, 83, 94
primary affective modes 157–158 Sass, L. A. 33–34
principle of all principles 119 Schacter, D. L. 32, 101
Prinz, J. 74 Schenk, F. 157
privatism 25 schizophrenia 33–35, 118
protention 20, 122, 126, 152–153, Schott, G. 185
157–161, 173 Schwitzgebel, E. 25, 80
Przyrembel, M. 184 Scoville, W. 116
psychologism 2, 45 Scriven, M. 73–74
psychology: ecological 99; eidetic 30, self-measurement 82–83
143, 148; empirical 6, 13, 17, Sellars, W. 43
20–22, 31, 42, 44, 120, 148; folk sense of agency 53–54, 197
53, 83, 121, 156; sense of ownership 53
phenomenological 13–15, 20–23, sensory substitution 33
37n4, 44–46, 124–125, 143, 145, Silberstein, M. 169
148, 154–155 Silberstein, S. 186
Putnam, H. 47, 66, 85, 105, 139 Simon, H. 80, 97, 168
small-world networks 169
qualia 26–28, 31–32, 81 Smith, B. 67n3
Quine, W. V. 42 Smith, J. 25, 81
Smith, J. A. 181
Raja, V. 93 Smith, L. 91
Ratcliffe, M. 33–34 Sober, E. 75
recomposition 98 Sowa, R. 123, 144, 198
reduction: eidetic 15, 21–23, 46, 144; Spivey, M. 188
inter-theoretic 47, 75; naturalistic Sporns, O. 169
42; nomological 75, 127; stable state 65, 88, 90, 172, 178
phenomenological 15–16, 18, 21, Stanghellini, G. 55
32, 143–144; transcendental 15, 21, state space see phase space
124, 143 Stich, S. 121
reductionism 3, 42, 47, 65–66, 112, Stinson, C. 168
195; mechanism 114 structure: generic 17, 21, 24–26, 60,
relevancy: causal 74, 150; constitutive 122; invariant 15–17, 61, 122;
79, 100, 150 neuro-dynamic 64–65, 176–177;
repellent 65, 178 pheno-dynamic 64, 176–177
representationalism 140 Stumpf, C. 148
representational theory of mind subjectivity 13, 15, 17, 27, 46–48,
140–141, 145 155–158
retention 20, 122–123, 126, 152–153, Suffczyński, P. 179–180
157–161, 173; longitudinal and
transverse 153; near and far 153 Taipale, J. 37n2
Reynolds, J. 3, 17, 21, 46, 67n1, 143 task analysis 85, 87, 149–150
Rodemeyer, L. 122, 153 Thinès, G. 17, 21, 143
206 Index
Thompson, E. 47, 57, 61, 66, Vincini, S. 15
109, 173 visual aura 185–187
time-consciousness 20, 50–52,
122–123, 153, 157–158, 172–173 Walmsley, J. 93
Timmermann, C. 184 Watkins, J. 73
tonic-clonic seizure 175 Watt’s governor 88–89
Tononi, G. 26–27, 31 Weber’s psychophysical law 77
topological map 157 Weiskopf, D. 86
Turing machine 85, 139 Welton, D. 19, 152
Wheeler, M. 43–44, 127
Ullah, G. 188 Williams, H. 120–121
underspecification 32–33 Wilson, A. 99, 170–171
understanding 73; constitutive 79; Wimsatt, W. 97, 168
enabling 127; phenomenological Woodward, J. 100, 103
124–127 Wright, L. 85
unification 75, 111–112
Unnwongse, K. 175, 185
ur-impression see primal impression Ylikoski, P. 99
Yoshimi, J. 29, 50–51
Valenzuela-Moguillansky, C. 184
van Gelder, T. 51–52, 88–89, 172 Zachar, P. 35
Varela, F. J. 32, 43, 46, 50–52, 60, Zahavi, D. 14–15, 17, 27–28, 37n2,
172–173, 183–184 45–47, 81, 119, 144, 162, 196
Vermersch, P. 64, 181 Zednik, C. 94, 100, 169, 171

(Routledge Research in Phenomenology) Marek Pokropski - Mechanisms and Consciousness - Integrating Phenomenology With Cognitive Science-Routledge (2021)

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

(Routledge Research in Phenomenology) Marek Pokropski - Mechanisms and Consciousness - Integrating Phenomenology With Cognitive Science-Routledge (2021)

Uploaded by

Copyright:

Available Formats

Mechanisms and Consciousness

This book develops a new approach to naturalizing phenomenology. The

Marek Pokropski is Assistant Professor at the Faculty of Philosophy,

Husserl’s Phenomenology of Intersubjectivity

Phenomenology of the Broken Body

Normativity, Meaning, and the Promise of Phenomenology

Levinas and Analytic Philosophy

The Bounds of Self

Towards a Phenomenology of Values

Mechanisms and Consciousness

For more information about this series, please visit: https://www.rout-

1 The Concept of Phenomenology 13

3 Models of Explanation in Cognitive Science 72

4 Phenomenology and Functionalism 139

5 Phenomenology and Dynamical Modeling 167

6 Conclusion: Toward Methodologically Guided

I.1 The fable of the blind men and an elephant 2

I wrote a significant part of this book in my hometown, Warsaw, during

For children in philosophy, this may be the dark corner haunted by

What is characteristic of philosophy is not a special subject-­matter,

Despite remarkable recent progress in neuroscience, such as research

Avoiding the difficulties inherent to the first-person perspective by ignor-

Outline of This Book

1.2.1 Reduction and Variation as Phenomenological Methods

Epoché can be seen as a necessary first step in initiating phenomenologi-

is based on the modification of an experienced or imagined objec-

In other words, the variation relies on imaginary modifications and

[i]n a certain way, we can therefore distinguish “explanatory” phe-

In other words, static phenomenology describes the contents of expe-

actually sufficient empirical science of the psychical in its relations to

The exploration of such psychic states, called perceptions, as states

Phenomenological psychology analyzes the eidos, the structure of percep-

Although Husserl claims that there is a “lawful regularity” in motivational

1.3 What Phenomenology Is Not

1.3.1 Phenomenology Is Not Introspection

Husserl not only opposed the introspectionist reading of phenomenol-

To use Husserl’s metaphor, a phenomenologist is like a geometer interested

1.3.2 Phenomenology Is Not About Qualia

we can say that a mental state is conscious if it has a qualitative

To put it differently, consciousness is a sort of inner awareness of qualia,

we will not have a complete scientific explanation of consciousness as

It seems to me that this sort of separation is also acknowledged in clas-

1.3.3 Phenomenology Is Not Anti-Naturalistic

when it is actually natural science that speaks, we listen gladly and

Husserl clearly acknowledges the importance of the natural sciences

my criticism of psychological method did not at all deny the value of

1.4 Why We Need Phenomenology in Explaining Consciousness

I agree with the latter view. Phenomenological models of a mental mal-

= integrating with cognitive science

2.2 Three Views on the Naturalization of Phenomenology

phenomenology cannot be naturalized because it tells the story of the

This, so to speak, orthodox approach to phenomenology sees in it a funda-

does not preserve a self-sufficient domain that is distinct from empiri-

Although minimal phenomenology gives up transcendentalism, it is, as

the terms “intentional arc,” “embodiment,” and “body schema,” like

To put it differently, the target reconceptualization of nature consists of

it is our general contention indeed that phenomenological descrip-

The general idea relies on translating phenomenological descriptions

let us make clear to ourselves the most universal peculiarities of

To put it differently, according to Husserl, phenomenology is not a

2.4 A Question of Constraints

it is quite possible that experimenting with phenomenology will lead

2.4.1.2 Formalization of Phenomenological Description

What is characteristic of philosophy is not a special subject-matter,

1.2.1 Reduction and Variation as Phenomenological Methods

1.3 What Phenomenology Is Not

1.3.1 Phenomenology Is Not Introspection

1.3.2 Phenomenology Is Not About Qualia

1.3.3 Phenomenology Is Not Anti-Naturalistic

1.4 Why We Need Phenomenology in Explaining Consciousness

2.2 Three Views on the Naturalization of Phenomenology

2.4 A Question of Constraints

2.4.1.2 Formalization of Phenomenological Description

2.4.1.3 Applied Ontology and Semantic Bridging

2.4.3 Neurophenomenology and Generative Passages

2.4.4 Neurophenomenology and Homeomorphism

(1) Analyzing the global dynamics of cerebral activity associated

(1) A temporal coincidence between the cerebral activity which is

3.3 Models of Explanation in Cognitive Science

i si si w s i 13,14

x 5 s13 13 s14 14

3.3.5.4 Limits of Mechanistic Explanation

3.4 Explanatory Integration of Cognitive Science

3.4.1 Theoretical Reduction, Unification, and Integration

3.4.2 Mechanistic Integration of Cognitive Science

3.5 Is Phenomenology Explanatory?

3.5.1 Types of Phenomenological Explanation

3.6 Conclusion: Toward Integration of Phenomenology With

4.2 Husserlian Phenomenology and Computational

4.3 The Notion of Function and the Idea of Functional

4.4 Functional Analysis and Phenomenological Decomposition

4.5 Toward Functional-Mechanistic Naturalization

4.5.1 Example 1: First-Person Perspective

4.5.2 Example 2: Vision Studies

5.2 Integrating the Dynamical and Mechanistic Frameworks

5.3 Neurophenomenology and Dynamical Systems Theory