Does Perception Have Content?

PHILOSOPHY OF MIND

Series Editor
David J. Chalmers, Australian National University and New York
University

Self Expression What Are We?

Owen Flanagan Eric T. Olson
Deconstructing the Mind Supersizing the Mind
Stephen Stich Andy Clark
The Conscious Mind Perception, Hallucination, and Illusion
David J. Chalmers William Fish
Minds and Bodies Cognitive Systems and the Extended Mind
Colin McGinn Robert D. Rupert
What’s Within? The Character of Consciousness
Fiona Cowie David J. Chalmers
The Human Animal Perceiving the World
Eric T. Olson Bence Nanay (editor)
Dreaming Souls The Contents of Visual Experience
Owen Flanagan Susanna Siegel
Consciousness and Cognition The Senses
Michael Thau Fiona Macpherson (editor)
Thinking Without Words Attention is Cognitive Unison
José Luis Bermúdez Christopher Mole
Identifying the Mind Consciousness and the Prospects of Physicalism
U.T. Place (author), George Graham, Derk Pereboom
Elizabeth R. Valentine (editors)
Introspection and Consciousness
Purple Haze Declan Smithies and Daniel Stoljar (editors)
Joseph Levine
The Conscious Brain
Three Faces of Desire Jesse J. Prinz
Timothy Schroeder
Decomposing the Will
A Place for Consciousness Andy Clark, Julian Kiverstein, and Tillmann
Gregg Rosenberg Vierkant (editors)
Ignorance and Imagination Phenomenal Intentionality
Daniel Stoljar Uriah Kriegel (editor)
Simulating Minds The Peripheral Mind
Alvin I. Goldman István Aranyosi
Gut Reactions The Innocent Eye
Jesse J. Prinz Nico Orlandi
Phenomenal Concepts and Phenomenal Does Perception Have Content?
Knowledge Edited by Berit Brogaard
Torin Alter, Sven Walter (editors)
Beyond Reduction
Steven Horst
Does Perception Have Content?
Edited by Berit Brogaard

1
3
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide.

Oxford New York

Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto

With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trademark of Oxford University Press

in the UK and certain other countries.

Published in the United States of America by

Oxford University Press
198 Madison Avenue, New York, NY 10016

© Oxford University Press 2014

All rights reserved. No part of this publication may be reproduced, stored in a

retrieval system, or transmitted, in any form or by any means, without the prior
permission in writing of Oxford University Press, or as expressly permitted by law,
by license, or under terms agreed with the appropriate reproduction rights organization.
Inquiries concerning reproduction outside the scope of the above should be sent to the
Rights Department, Oxford University Press, at the address above.

You must not circulate this work in any other form

and you must impose this same condition on any acquirer.

Library of Congress Cataloging-in-Publication Data

Does perception have content? : essays / edited by Berit Brogaard.
pages cm.—(Philosophy of mind series)
ISBN 978–0–19–975601–8 (hardcover : alk. paper) 1. Perception (Philosophy)
I. Brogaard, Berit, editor.
B828.45.D64 2014
121′.34—dc23
2013049316

98765432 1
Printed in the United States of America
on acid-free paper
CONTENTS

Contributors vii

1. Introduction: Does Perception Have Content? 1

BERIT BROGAARD

PART ONE Content Views

2. Empirical Problems with Anti-Representationalism 39
BENCE NANAY

3. Affordances and the Contents of Perception 51

SUSANNA SIEGEL

4. Looks, Reasons, and Experiences 76

KATHRIN GLÜER

PART TWO Against Strong Content

5. The Problem with the Content View 105
MARK JOHNSTON

6. The Preserve of Thinkers 138

CHARLES TRAVIS

7. Disjunctivism, Discrimination, and Categorization 179

DIANA RAFFMAN

PART THREE Reconciliatory Views

8. The Relational and Representational Character of Perceptual Experience 199
SUSANNA SCHELLENBERG

9. Experiential Content and Naïve Realism: A Reconciliation 220

HEATHER LOGUE

10. Love in the Time of Cholera 242

BENJ HELLIE
 Contents
vi   

PART FOUR Imagistic and Possible-Word Content

11. Image Content 265
MOHAN MATTHEN

12. What is the Content of a Hallucinatory Experience? 291

MICHAEL TYE

PART FIVE The Constituents of Perceptual Content and the

Role of Perception
13. What Does Vision Represent? 311
WILLIAM G. LYCAN

14. Phenomenal Intentionality and Secondary Qualities: The Quixotic

Case of Color 329
TERRY HORGAN

15. Which Causes of an Experience Are Also Objects of the Experience? 351
TOMASZ BUDEK AND KATALIN FARKAS

Index 371
CONTRIBUTORS

Berit Brogaard is Professor of Philosophy at University of Miami, Florida. She

is the author of Transient Truths: An Essay in the Metaphysics of Propositions and
numerous articles in the philosophy of perception and philosophy of language.
Tomasz Budek studied philosophy at the Central European University.
Katalin Farkas is a Professor of Philosophy at Central European University,
Budapest, and she has also been serving as Provost between 2010 and 2014.
She graduated in mathematics and philosophy at Eötvös Loránd University,
and earned a PhD from the Hungarian Academy of Sciences. Her main area of
research is the philosophy of mind, and this is the topic of her book The Subject’s
Point of View (OUP, 2008). She held the Kerstin Hesselgren Visiting Professorship
at the University of Stockholm; she was a Junior Fellow at the Collegium Budapest,
a visitor at the University of Sydney, The Center for Subjectivity Research in
Copenhagen, and the RSSS at the Australian National University in Canberra.
Kathrin Glüer is Professor of Theoretical Philosophy at Stockholm University.
She works mainly in the philosophy of mind and language. Recent work
includes: Donald Davidson. A Short Introduction (OUP 2011), “General Terms and
Relational Modality”, with Peter Pagin, in Noûs 46, 2012: 159–199, “In Defence of
a Doxastic Account of Experience”, in Mind & Language 24, 2009: 297–373, and
"Against Content Normativity”, with Åsa Wikforss, in Mind 118, 2009: 31–70.
Benj Hellie is Associate Professor of Philosophy at the University of Toronto. He
has published on perception and consciousness—and, more recently, on action—in
regard to matters of metaphysics, rational psychology, language, and philosophical
historiography. His current project, Out of This World: A Copernican Revolution
in Modal Space, attempts to give formal expression to broadly Kantian themes in
‘M&E’ by reassigning various duties from semantics to discourse pragmatics.
Terry Horgan is Professor of Philosophy at the University of Arizona. He is au-
thor (with John Tienson) of Connectionism and the Philosophy of Psychology
(MIT, 1996), (with Matjaž Potrč) Austere Realism: Contextual Semantics Meets
Minimal Ontology (MIT, 2008), and (with David Henderson) The Epistemological
Spectrum: At the Interface of Cognitive Science and Conceptual Analysis (OUP,
2011). He has published (often collaboratively) on various aspects of philoso-
phy—including philosophy of mind, metaphysics, philosophy of language, epis-
temology, and metaethics.
vii
 Contributors
viii   

Mark Johnston is the Walter Cerf Professor of Philosophy at Princeton University

and the author of many widely reprinted articles in philosophy of mind, met-
aphysics and epistemology. He is also the author of Saving God (Princeton
University Press, 2009) and Surviving Death (Princeton University Press, 2011).
Heather Logue is a Lecturer in Philosophy at the University of Leeds. Her research
focuses on issues in metaphysics and epistemology, and particularly on issues con-
cerning perceptual experience. She has published and has forthcoming papers on
Naïve Realism, disjunctivism, and skepticism about the external world, and she
co-edited (with Alex Byrne) Disjunctivism: Contemporary Readings (MIT Press,
2009). Before coming to Leeds, Heather completed her PhD at MIT in 2009 and
her bachelor's degree at the University of Pittsburgh in 2003.
William G. Lycan is William Rand Kenan, Jr. Professor of Philosophy at the
University of North Carolina. He is author of eight books, including Logical Form
in Natural Language (1984), Knowing Who (with Steven Boër, 1986), Consciousness
(1987), Judgement and Justification (1988), and Real Conditionals (2001).
Mohan Matthen is a senior Canada Research Chair in the Philosophy Department
at the University of Toronto. He is the author of Seeing, Doing, and Knowing (OUP,
2005), a philosophical treatment of perceptual processes. He also works in phi-
losophy of biology, with a special focus on evolution, selection, and species.
Bence Nanay is Professor of Philosophy and BOF Research Professor at the Centre
for Philosophical Psychology at the University of Antwerp, and Senior Research
Associate at Peterhouse, Cambridge University. He is the author of Between
Perception and Action (OUP, 2013) and Aesthetics as Philosophy of Perception (OUP,
forthcoming), and editor of Perceiving the World (OUP, 2010). He has published
papers in various journals on philosophy of mind, aesthetics, and philosophy of
biology.
Diana Raffman is a professor of philosophy at the University of Toronto. She is the
author of a number of papers in the philosophy of mind, primarily about percep-
tual experience, and in the philosophy of language, primarily about vagueness. She
has recently completed a book, Unruly Words: A Study of Vague Language (OUP,
2014), which advances a new theory of linguistic vagueness.
Susanna Schellenberg is Associate Professor of Philosophy at Rutgers University.
Her articles include “Perceptual Content Defended” (Noûs, 2011), “Ontological
Minimalism about Phenomenology” (Philosophy and Phenomenological Research,
2011), “The Situation-Dependency of Perception” (Journal of Philosophy, 2008),
and “Action and Self-Location in Perception” (Mind, 2007).
Susanna Siegel is Edgar Pierce Professor of Philosophy at Harvard University. She
is author of The Contents of Visual Experience and numerous articles in the phi-
losophy of perception.
 Contributors   ix

Charles Travis is professor of philosophy in King’s College, London. Prior to that,

he was professor in Northwestern University (Evanston, Illinois), before that pro-
fessor in the University of Stirling (Scotland). He has published two books on
Wittgenstein, one on propositional attitudes, and one volume of collected essays
(on the theme of sensitivity of representing to the occasion for it). He works on
general themes in epistemology and metaphysics, with an increasing emphasis on
Frege, and, over the last decade, on what he sees as the fundamental problem of
perception: How can perception make the world bear, for the perceiver, on what
he is to think and do? For further details, or access to forthcoming work, see his
personal webpage: http://sites.google.com/site/charlestraviswebsite.
Michael Tye encountered philosophy at Oxford, and taught at Temple University,
St. Andrews, and the University of London, before coming to the University of
Texas at Austin in 2003. He is the Dallas TACA Centennial Professor in Liberal Arts.
1

Introduction
DOES PERCEPTION HAVE CONTENT?
Berit Brogaard

1.1. The Strong and Weak Content Views

The title of this volume takes the form of a question, a question which exerts a con-
siderable hold on contemporary philosophers of mind, particularly those working
on perception. This seems to be a fairly new trend. Not so long ago the question
would not even have been considered. Perhaps it would not have seemed intelli-
gible. But things have changed, and there is now a considerable number of articles,
theses, and books aimed at answering it, positively or negatively. What are the fac-
tors responsible for this topic becoming a “live” one? Why is it only now receiving
so much attention?1
I believe the answer to this question is largely historical. On the face of it, tra-
ditional debates about perception were typically concerned with a different ques-
tion, viz., that of whether we perceive the external world directly or indirectly. In
Perception: A Representative Theory, for example, Frank Jackson argues that when
we see things in the environment, we see them in virtue of perceiving something
else. The things that we perceive without having to perceive something else are
sense data. Jackson thought that sense data are something we literally perceive and
the only things we are directly perceptually aware of.
Though the debate about whether we perceive the external world directly or
by virtue of perceiving something else is orthogonal to the debate about whether
perceptual experience has content, it may be argued that the two debates concern
some of the same issues. As we will see below, particular ways of understanding
perceptual content may, at least at first glance, appear to imply that if perception
has content, then the content is an intermediary between the perceiver and the ex-
ternal world, and the perceiver experiences the world by being acquainted with the
content. Things are not quite as simple as this, of course. But it does raise the fol-
lowing question: If the debates about the directness of perception and perceptual

1
Thanks to the anonymous reader for encouraging me to consider these issues and questions. 1
2   Introduction

content are intermingled, what caused the relatively sudden interest in whether
perceptual experience has content?
The notion of perceptual content is not new, of course. In Perception:
A Representative Theory, for example, Jackson casually refers to perceptual con-
tent (e.g., 1977, p. 40), but his endeavors are not aimed at answering the question
of whether perception has content. I believe the recent considerable interest in
the question of whether perception has content may have been a result of the
rise of cognitive science and its focus on the idea of a representational state of
the mind. It seems that the debates in cognitive science have sparked analogous
debates in philosophy of perception about what it means to say that percep-
tual experience has content and whether perceptual experience has content in
the first place. The chapters in this volume deal with these two issues. The first
half of the volume directly addresses the question of whether perception has
content, whereas the second half addresses the question of what the content of
perceptual experience is like.
The question of whether perceptual experience has content may, at first
glance, seem rather trivial. A simple argument for the view would run as follows.
Perceptual experience is accurate or inaccurate. If it’s accurate, it’s accurate in
virtue of some proposition p being true. If it’s inaccurate, it’s inaccurate in virtue of
some proposition p being false. But that proposition p just is the content of percep-
tual experience. So perceptual experience has content. While this argument has
something to be said for it, it doesn’t quite get to the core of the debate.
As Susanna Siegel points out in her book The Contents of Visual Experience,
one flaw in the argument from accuracy conditions is that it does not require that
the accuracy conditions had by experiences are conveyed to the subject by her
experience (Siegel 2010, p. 43). But accuracy conditions that are not conveyed to
the subject by the experience are not suitable to serve as experiential contents. For
example, all experiences are accurate if and only if Mother Nature corroborates
the experience. But no typical experiences convey Mother Nature corroborates my
experience to the subject and the proposition Mother Nature corroborates my expe-
rience ought not normally count as the experiential content. So, the general move
from accuracy conditions to contents is invalid.
Siegel (2010) distinguishes between weak and strong perceptual content. On
the weakest acceptable formulation of the view that perceptual experience has
content, experiential content is its accuracy conditions properly conveyed to the
subject.
Weak Content View
Experience e has the proposition p as a content iff p is conveyed to the subject
of e, and necessarily if e is accurate, then p is true.
It is the question of whether perceptual experience has content in this mini-
mally acceptable sense that directly or indirectly has inspired the topics of most of
the chapters in this volume.
 Introduction   3

As we will see below, it takes substantial argument to establish this view with
some degree of plausibility, and many thinkers reject the view, including naïve
realists and enactivists.
There are also historical views not discussed in this volume that, at least prima
facie, are at odds with the weak content view. On one such view, perceptual ex-
perience is a strictly non-intentional (or non-representational) kind of sensation
that requires further interpretation in order to be fully graspable by the perceiver.
We might call this type of view the ‘raw sensations view’. Thomas Reid’s two-state
view can be interpreted along these lines. Because raw sensations are strictly
non-intentional, they do not have accuracy conditions at all. So unless they are
subjected to further interpretation, they cannot properly be said to have content
even in a weak sense.
Visual form agnosia, a brain condition in which perceivers can see that there
is something in front of them but cannot identify what is in front of them, may
shed some light on what raw sensations are. Visual agnosia patients sometimes de-
scribe the “something” in front of them as a blob without clear boundaries, color,
shape, or texture. It’s not implausible to think that only raw sensations are avail-
able to these patients’ conscious visual system. The information they consciously
possess about their environment does not represent any particular thing but just
a “something.” Arguably, the accurate/inaccurate dichotomy does not apply to the
visual experience of visual agnosia patients. Their visual experiences certainly are
not like normal, accurate perceptual experiences but nor is it the perceptions (in
the standard sense) that have gone wrong.
Though the weak content view is already very contentious, some thinkers be-
lieve that being contentful is essential to perceptual experience. On this view, it’s
a fundamental feature of perceptual experience that it has representational con-
tent, i.e., content that is suitable to serve as the content of a propositional attitude
(Siegel 2010).
Strong Content View
Experience e has the proposition p as a content iff necessarily, the subject of e
bears a propositional attitude towards p.
The strong content view, if true, appears to rule out a number of familiar views
about perceptual experience. On the face of it, views that take perception to be
a perceptual relation to an external object (and features of that object) appear to
be at odds with the strong content view. Naïve realists often hold that perception
is fundamentally characterized by its relational properties but not fundamentally
characterized by its representational properties (see, e.g., Fish 2009). Veridical (or
non-hallucinatory) perception is a perceptual relation to an object that does not
involve representational properties in any interesting sense. On this view, verid-
ical perception has content only in the weak sense: when a perceiver stands in the
perceptual relation, she is in a position to acquire knowledge on the basis thereof.
Disjunctivism adds to this view that cases of hallucinations (Langsam 1997;
4   Introduction

Snowdon 1980/81; Smith) and in some cases also illusions (Hinton 1973; McDowell
1982, 1986, 2008; Martin 2002; Fish 2008, 2009) are not of the same fundamental
kind as veridical perceptions.2 Some hold that hallucinations are thoughts or
beliefs (e.g., Fish), others that they are more closely related to mental imagery
in that they are relations to imaginary objects (e.g., Smith 2002), and yet others
that nothing more can be said about hallucinations than that they are internally
indiscriminable from veridical perception but that the perceptual relation fails to
obtain (Martin). Naïve realists can thus hold that hallucinations, and perhaps also
illusions, have representational content and yet deny that it is a fundamental fea-
ture of perceptual experiences that they have representational contents.
Though naïve realists with traditionalist leanings adamantly deny that percep-
tual experiences have any type of content, some thinkers are more all-embracing.
As we will see, several of the contributors to this volume argue that naïve realism,
initial appearances to the contrary, can accept the strong content view.
There are also historical views that appear to be at odds with the strong content
view. One is adverbialism. The name ‘adverbialism’ derives from the word group
made up of adverbs. Adverbs modify the verb that they grammatically are adja-
cent to. They may lead to further specification of features of the denotation of the
verb or an alteration of the meaning of what was said. For example, if I am told
that John made a sandwich, I might be interested in knowing whether he made it
slowly, gracefully, angrily, or carelessly. ‘Slowly’, ‘gracefully’, ‘angrily’, and ‘carelessly’
are adverbs that provide information about features of the action picked out by the
verb. Adding these adverbs to the original sentence does not cancel out the original
meaning. If John made a sandwich angrily, then he made a sandwich. However,
other adverbs will alter the meaning of the original sentence when added. For ex-
ample, the sentence ‘John barely made a sandwich’ doesn’t entail that John made a
sandwich. Adverbialism holds that features of perception play a role analogous to
the first group of adverbs (Chisholm 1957). They specify a way in which the subject
perceives the world. For example, if John has a perceptual experience of a red cat,
then John is appeared to redly and cat-wise. On this view, perception is neither a
relation nor a mental state with content in the strong sense. It just is a way of per-
ceiving. Roderick Chisholm (1957) is a legendary defender of the adverbial theory.
Wylie Breckenridge (n.d.) is one of its contemporary defenders.
The sense-datum view, originally defended by Russell (1912), Broad (1925),
Price (1950), Ayer (1956), and Jackson (1977), also appears compelled to deny that
perception has content in the strong sense.3 On the sense-datum view, perceivers
don’t perceive the world directly; instead they perceive sense-data by standing in
a perceptual relation to these sense-data. Sense-data are proxies for objects in the
external world. They have colors, shapes, textures, and so on. While sense-data

2
Fish holds the intermediate view that illusions are sometimes good cases of perception and
sometimes bad. See Fish (2009) and Brogaard (2011).
3
More recent defenders include: Robinson (1994), and Casullo (1987).
 Introduction   5

represent objects and features in the external world, sense-data are not them-
selves perceptual experiences. Perceptual experiences fundamentally consist in
being related to sense-data.
The aforementioned views are among the best known views that commonly
are thought to deny that perceptual experience has content in a strong sense.
What of views that embrace the strong content view? One candidate to be a view
of this kind is representationalism (or intentionalism, as it’s sometimes called).
Strong representationalism is the view that for a perceptual experience to have
a certain content just is for it to have a certain phenomenal character (Chalmers
2004). A weaker version merely holds that the contents of perceptual experience
supervene on the phenomenal character of the experience. Though representa-
tionalism makes more commitments than the strong content view, the two are
compatible. Representationalism, as well as the strong content view, holds that
perceptual experience fundamentally consists in representing externally instanti-
ated properties and/or objects.
In the following I will look in some detail at the contributions to this volume
and, along the way, point to some potential issues raised by each contribution.

1.2. Content Views

Bence Nanay: In “Empirical Problems with Anti-Representationalism” Bence Nanay

outlines some empirical challenges for what he calls ‘anti-representationalism’, the
view that ‘perception is not a process of constructing internal representations’ (cf.
Noë 2004, p. 178).
Anti-representationalist views fall into two categories, enactivism and rela-
tionalism. According to enactivism, we are equipped with the ability to perceive,
not for the sake of our pleasure, but because we need to act to survive. But we do
not need to construct perceptual representations in order to get around in the
world. The information needed to act is already out there in the world—externally
stored, as it were. Perception, enactivists hold, is an active and dynamic process
between the agent and her information-packed environment. Relationalists also
deny that perceptual states are representations. They take perceptual states to be
relations, not between a perceiver and an abstract perceptual content, but rather
between a perceiver and an external concrete particular (Brewer 2006; Martin
2004, 2006).
The first empirical challenge to anti-representationalism that Nanay presents
turns on the observation that our perceptual system sometimes attributes incom-
patible properties to the same object. Nanay invites us to consider the empirically
supported view that humans have two visual subsystems that are anatomically and
functionally separated, also known as the ventral and dorsal streams. The ventral
stream, which starts in the visual cortex and then runs into the temporal gyri and
ends in the prefrontal cortex, is in charge of object identification and recognition.
6   Introduction

FIGURE 1.1 The Ebbinghaus illusion Studies have shown that this illusion leads to a
misperception of the size of the central circle but only marginally affects grasping behavior
directed at the central circle.

The dorsal stream, which starts in the visual cortex and then runs upwards into
the parietal cortex and ends in the sensorimotor cortex, is in charge of guiding
ongoing action.
In neurotypical humans, the ventral and dorsal visual systems need not
ascribe the same properties to objects. One example of a case in which the two
pathways ascribe different properties is the Ebbinghaus illusion (­figure 1.1), in
which a circle that is surrounded by smaller circles looks larger than a circle of
the same size that is surrounded by larger circles. There is empirical evidence
that the two visual streams attribute different size properties to the inner circle.
While the circle that is surrounded by the smaller circles looks larger than a
same-sized circle surrounded by larger circles, our grip-size is not influenced
by the illusion. Our hand aperture when attempting to reach to and grasp the
middle circle is the same in the two cases.
The representational view, Nanay argues, has a straightforward way of
explaining the different property attributions. The two visual pathways represent
the inner circle as having two different size properties.
According to Nanay, the anti-representationalist cannot offer the same expla-
nation. If perception is a perceptual relation between a perceiver and perceptible
property instances, then there is a single perceptual relation between the perceiver
and the property instances regardless of how the brain processes the relevant in-
formation. In a three-dimensional model of the Ebbinghaus illusion, the percep-
tual relation is a relation between the perceiver and the property instances of a
poker chip. According to the enactivist, the information that is relevant in percep-
tion is stored in the external world. But, Nanay argues, it is unclear which of the
properties is relevant to perception in the case of the Ebbinghaus illusion.
It would thus seem that the anti-representationalist must deny that there are
two perceptual episodes in the Ebbinghaus scenario. However, denying this has dire
consequences. The size property the middle poker chip appears to have plays a role
in the justification of our beliefs, and the property that our hand aperture is tracking
plays a role in our ongoing actions. So, neither can be dismissed as irrelevant.
 Introduction   7

The second empirical problem for anti-representationalism that Nanay consid-

ers turns on the multimodality of perception. Information in one sensory modality
can influence information processing in another, even at very early stages of infor-
mation processing. For example, if there is a flash in the visual scene and two beeps
are played during the flash, then the visual information is experienced as two flashes
(Shams et al. 2000). A further example of this is presented by blind individuals who
are taught to navigate using echolocation. These individuals typically develop a visual
phenomenology that represents objects and their features in their environment.
Representational views, Nanay argues, can explain multisensory percep-
tion as an attribution of properties by different sense modalities to the same
perceived scene. Different sensory modalities represent the same scene as hav-
ing different properties by matching two modality-specific representations.
Anti-representationalists appear to have greater difficulties explaining multisen-
sory perception. The enactivists’ talk about perception as an active exploration of
the environment presupposes multimodal perceptual abilities. Likewise, the rela-
tionalists’ perceptual relation appears to be the product of integrated information.
So, Nanay argues, it appears that brain representations are required to explain the
integration of multisensory information and hence play a crucial role in percep-
tion, contrary to what anti-representationalists claim.
Susanna Siegel: In her contribution to this volume Susanna Siegel brings the con-
tent debate to bear on Gibson’s notion of an affordance. On Gibson’s view, an affor-
dance is a kind of possibility of action. Siegel focuses on experiences of affordances in
which actions seem to the subject be pulled out of her directly by the environment. Here
is an example from Siegel. Suppose you are passing someone on an otherwise empty
sidewalk. There are lots of different ways you can respond. You can continue walking
until you collide with the passerby, you can step to the right, to the left, etc. Typically
only some of the many possibilities in this type of situation are salient. Siegel describes
cases in which one feels as if the situation mandates a specific course of action, such as
adjusting your position so you and the passerby don’t collide. Siegel calls experiences of
affordances in which a course of action is felt to be mandated ‘experienced mandates’.
Siegel finds the seeds of arguments against content views in enactivist
thinkers such as Hubert Dreyfus, who discuss experienced mandates in on-
going action (though not under that label). In the passerby scenario, if you
decide to step to the right and initiate the action, you cannot easily stop in the
middle. So the part of the action that lies ahead of the part of the action you
are currently completing is an experienced mandate. The current experience
determines the next. So, it would seem that there is no need for experiential
representations to guide action in these kinds of cases. Something like this view
is explicit in the following passage from Dreyfus.
In our skilled activity we move to achieve a better and better grip on our situ-
ation. For this movement towards maximum grip to take place, one does not
need a mental representation of one’s goal. Rather, acting is experienced as a
steady flow of skillful activity in response to one’s sense of the situation. Part
8   Introduction

of that experience is a sense that when one’s situation deviates from some
optimal body-environment relationship, one’s activity takes one closer to that
optimum and thereby relieves the ‘tension’ of the deviation. One does not
need to know what that optimum is. One’s body is simply solicited by the situ-
ation to get into equilibrium with it. (Dreyfus 2002, p. 12)
Siegel considers several challenges to the weak content view: either experienced
mandates do not involve any content at all, or experienced mandates do involve
content but the content does not play any explanatory role in action.
Siegel argues that experienced mandates have contents, and claims to identify
the content that could reflect the distinctive character of experienced mandates.
She offers the weak content view as a premise in one of her arguments against the
idea that perceptual content plays no significant role in explaining behavior. A
simplified version of Siegel’s argument for the weak content view can be articu-
lated as follows (see Siegel 2010 for the full argument):
Siegel’s Argument for Weak Content
1. All visual (perceptual) experiences present clusters of properties F as
being instantiated.
2. So, necessarily, things are as E presents them to be only if F is
instantiated. (from 1)
3. So, E has a set of accuracy conditions C, conveyed to the subject of E.
(from 2)
Conclusion: All visual experiences have content.

Below we will consider some of Heather Logue’s concerns about the argument. As
I see it, one potential problem with the argument is that it doesn’t make it explicit
what ‘present’ and ‘convey’ mean and how the two are connected. As I understand
Siegel, a perceptual experience presents a property as instantiated only if it percep-
tually seems that way to the subject. Likewise, F’s being instantiated is conveyed to
a subject only if things perceptually seem that way to her. So, the argument could
be formulated as follows:
The Argument from Appearances
1. All visual (perceptual) experiences make it perceptually seem to the
subject that clusters of properties F are instantiated.
2. So, necessarily, things are as they perceptually seem only if F is
instantiated. (from 1)
3. So, E has a set of accuracy conditions C that correspond to how things
perceptually seem to the subject. (from 2)
Conclusion: All visual (perceptual) experiences have content.

One potential issue with this version of the argument lies with the step from
premise 2 to premise 3. The accuracy conditions for an experience should be the
complete condition C such that the experience is accurate if and only if C obtains.
However, we don’t get that from premise 2. Premise 2 only mentions properties
 Introduction   9

that perceptually appear to be instantiated. If, however, one doesn’t pay much at-
tention to a tiny leaf on a tree in one’s visual field, then it may not perceptually
seem to one that properties of that leaf are instantiated. Whether or not the leaf ’s
properties are instantiated, however, might impact the accuracy of the experience.
So, we may not be able to get the accuracy conditions, or the experimental con-
tent, from how things perceptually appear to the subject. Of course, if perceptual
appearances just are perceptual experiences, then this objection has no traction.
Kathrin Glüer: In “Looks, Reasons, and Experiences,” Kathrin Glüer starts off with
an argument for the strong content view. According to Glüer, contents either are, or
are essentially such that they determine, accuracy conditions. Accuracy conditions, she
says, are conditions the world must satisfy in order for the content in question to be true.
Contents have their truth-conditions essentially. They play a role in accounts of how
different mental states relate to each other. We assign contents to psychological states
to model a specific kind of structure among them and to predict and explain how one
state can lead to another state. According to Glüer, the overarching reason perceptual
experiences have contents of this sort is that they are propositional attitudes. Perceptual
experiences, she holds, are beliefs. Since beliefs uncontroversially have contents, and
perceptual experiences are beliefs, perceptual experiences have contents.
To say that perceptual experiences are beliefs is not to say that there are no
differences between perceptual experiences and other beliefs. For example, per-
ceptual experiences have a phenomenology that is notably different from the phe-
nomenology of beliefs that are not perceptual experiences. Perceptual experiences
have a distinctive sensory phenomenology.
One challenge for the view that perceptual experience is belief is to explain
cases in which we don’t believe what we experience. When we dip a stick in water
and it looks bent, we don’t come to believe that it is bent. We know that it’s not.
Alex Byrne (2009) has argued that perceptual experiences are primitive beliefs.
According to him, we do indeed believe that the stick is bent on a very primitive
level. But on a more rational level, we do not believe that the stick is bent.
This is not Glüer’s strategy. According to Glüer, when we have a perceptual ex-
perience, things look a certain way to us. Gluer takes these looks to constitute the
content of perceptual experience. If I am looking at a blue car, and the car looks blue
to me, then the content of my perceptual experience is ‘Look(the car is blue)’, where
‘look’ is an operator on the embedded material. When I look at the stick in the
water, I come to believe that it looks bent but I don’t come to believe that it is bent.
Glüer considers and replies to a potential problem for her appearance account
of perceptual experiences. The envisaged counter-argument runs as follows. The
phenomenal notion of look cannot be used to specify the very content of visual ex-
perience because ‘look’ is a propositional attitude operator and thus cannot occur
in the content of any first-order propositional attitude. Glüer calls this the ‘Attitude
Operator Argument’. The argument can be summarized as follows:
The Attitude Operator Argument
1. Perceptual experience is a first-order propositional attitude.
 Introduction
10   

2. If perceptual contents contain ‘look’ operators, then perceptual

experience is not a first-order propositional attitude.
Conclusion: Perceptual contents do not contain ‘look’ operators.

The first premise is widely held to be true: second-order propositional attitudes are
not typically considered perceptual states. For example, most views of perception
would hold that you cannot perceptually experience having a belief or a desire.
That process would count as introspection, not perception. Of course, those who
take introspection to be a form of perception would resist drawing this distinction.
But even people in this camp can admit that the standard forms of perception that
Glüer is interested in are first-order propositional attitudes and not second order.
Glüer rejects premise 2. ‘Look’, she argues, is not a propositional attitude oper-
ator. So, perceptual contents can be first-order, even if they contain a ‘look’ operator.
The reason, she argues, is simple: Propositional attitude operators create hyperinten-
sional contexts. But ‘look’ does not. So, ‘look’ is not a propositional attitude operator:
Glüer’s Counterargument
1. If Φ is a propositional attitude operator, then it creates a hyperintensional
context.
2. ‘Look’ does not create a hyperintensional context.
Conclusion: ‘Look’ is not a propositional attitude operator.

The first premise in Glüer’s reply is relatively uncontroversial: if an operator is a

propositional attitude operator, then it generates hyperintensional contexts (e.g.,
Lois Lane desires Superman but not Clark Kent). The premise in need of justifica-
tion is the second one, viz., the premise that ‘look’ does not create hyperintensional
contexts. It is at least initially plausible that ‘look’ does elicit hyperintensional con-
texts. Consider:
1. It looks to Lois Lane as if Superman is flying by.
2. It looks to Lois Lane as if Clark Kent is flying by.

‘Superman’ and ‘Clark Kent’ are necessarily co-referential but substituting one for
the other appears to elicit a change in truth-value. This indicates that ‘look’ gener-
ates a hyperintensional context. However, Glüer thinks this appearance is illusory.
According to Glüer, ‘look’ satisfies the following substitution principle:
Substitution Principle
Co-phenomenal expressions can be substituted salva veritate in ‘look’-contexts.
I take it that Glüer takes co-phenomenal expressions to be expressions that refer
to entities that look the same to a perceiver in normal viewing conditions. The
Substitution Principle implies that if two expressions are not co-phenomenal,
then substituting one for the other will change the truth-value of sentence.
Glüer notes that a Superman-look is very different from a Clark Kent-look.
So, ‘Superman’ and ‘Clark Kent’, though co-referential, are not co-phenomenal.
 Introduction   11

That seems right. But Glüer then argues that it follows from this that ‘look’-
contexts are not hyperintensional. This, however, could be questioned. On the
standard definition of ‘hyperintensionality’, an operator is hyperintensional just
in case substituting an expression for a logically (or metaphysically) equivalent
expression under the operator can change the truth-value of the whole. On that
definition, ‘look’ generates hyperintensional contexts.
However, there might be an independent problem with the second premise of
the Attitude Operator Argument:
Premise 2
If perceptual contents contain ‘look’ operators, then perceptual experience is
not a first-order propositional attitude.
This premise appears to rest on the assumption that hyperintensional operators
are propositional attitude operators. This latter assumption, however, is mistaken.
Fictional operators, such as ‘according to the Sherlock Holmes stories’ are hyper-
intensional yet they are not propositional attitude operators. So, if premise 2 is
true, then it is not because ‘look’ is hyperintensional. For the Attitude Operator
Argument to be effective against Glüer’s position, then, one would need to show
on independent grounds that ‘look’ operators are propositional attitude operators.

1.3. Against Strong Content

Mark Johnston: In his contribution to this volume, Mark Johnston argues against the
view that perceptual experience has content. One problem with the view, Johnston
argues, is that it fails to adequately capture the distinction between illusory and
non-illusory experience. As there can be veridical illusions, the distinction be-
tween veridical/non-veridical experience is orthogonal to the distinction between
illusory/non-illusory experience. Johnston considers a number of optical illusions
that bear on the latter distinction. In the larger ball illusion, for example, two balls
are presented as located behind each other in a three-dimensional room. Because
the two balls actually have the same size, the ball that is presented as being further
away looks much larger than the other ball. Johnston points out that there is no
non-illusory counterpart of the illusion. If we could somehow override the standard
perceptual mechanisms that present relative sizes, we might experience the two balls
as having the same size. But it would still be an illusion because the perceptual sys-
tem then wouldn’t function normally. But the content view fails to account for this.
If we were to ask “why is the experience illusory?” the answer could not be that the
content misrepresents, because it doesn’t. The explanation would have to appeal to
the failed causal relation between the object and the perceiver. But on the content
view, the causal relation is not a constituent of the content of the experience, which is
to say, the content plays no role in the explanation. So, even if the content view is able
 Introduction
12   

to model the veridical/non-veridical distinction in terms of the truth or falsehood

of the proposition experienced, the view does not correctly model the non-illusory/
illusory distinction.
Johnston then considers whether the view that the content of perception is
object-dependent fares better in terms of capturing the relevant distinction. Such
a view may seem to guarantee that perceptual experience “puts us in touch with
worldly items.” However, Johnston argues that this view fails to model the distinc-
tive epistemic role of perceptual experience. If it’s merely the content of experience
that puts in touch with worldly items, then we cannot account for how experience
justifies perceptual judgments in a significant way.
A natural response to the first concern is to take content to represent a causal
relation between the worldly items and the perceiver, as has been argued by John
Searle. If one is seeing a red tomato over there, then the content of the percep-
tion is not just that there is a red tomato over there but also that it is causing
the very experience in question. The content view can then correctly model the
illusory/non-illusory distinction and not just the veridical/non-veridical distinc-
tion. However, the view faces independent problems. Johnston considers a case in
which you are looking into cars at dusk. When looking at the upholstery of other
people’s cars, the fabric is presented to you in gray-tones. But when you peek into
your own car, the back seat looks red, because memory fills in. In the memory case
the redness of the seat does in part cause the appearance that the seat is red. But
presumably this is not the right kind of causation. So, Searle would have to amend
his view by building in something like causation of the right sort, which seems
like an implausible component of perceptual content. Searle’s view, like all content
views, also fails to account for the epistemic significance of perceptual experience.
Johnston considers several other possible content views and shows why they
fail in one of the two ways. His own view is a form of naïve realism but one that
is distinct from disjunctivism. He thinks at least some disjunctivists mistakenly
think that we perceive facts. This, however, turns perceptual experience into a kind
of propositional attitude. But perceptual experience, Johnston argues, is an objec-
tual attitude.
Johnston’s argument is not easily refuted, as it seems right that content by it-
self cannot play an epistemically significant role and cannot account for illusions
and hallucinations. To refute the argument, one might argue that perceptual expe-
rience is also fundamentally relational, as Schellenberg and Logue do (see below).
One might also attempt to show that experience must be treated as representa-
tional in order to account for cases in which the objects or properties presented
are different from the external object and its property instances without the expe-
rience counting as illusory or hallucinatory. Color experience may be an example
of this.
Charles Travis: In “The Preserve of Thinkers,” Charles Travis argues against
the view that mental states represent. Travis starts out by distinguishing among
three notions of representation: effect-representation, auto-representation, and
 Introduction   13

allo-representation. Effect-representation is a two-place relation between two his-

torical circumstances. For example, teetering rock effect-represents aeons of wind
erosion, a child’s footstep in the sand effect-represents the child who left it behind,
and a pig’s snout effect-represents a pig. Both agents and objects can effect-represent.
Unlike effect-representation, auto-representation and allo-representation are re-
served for thinkers. They are three-place relations. To auto-represent is to assume
that things are a certain way or to passively take things to be a certain way. Unlike
auto-representing, allo-representing is not simply a condition one is in, it is something
one does when one sends a message and takes responsibility for its accuracy. Since
perceptual experiences are not agents, Travis argues, they can only effect-represent.
But this relation is not very interesting in the present context because it’s a purely
causal relation that may obtain between many other states of an organism and its
environment.
One of Travis’ arguments against thinking that perceptual experience is rep-
resentational turns on perceptual looks. The representational view depends on the
claim that there are perceptual looks that determine a particular representational
content. But perceptual looks cannot play this role. Travis offers the example of a
chrome yellow Porsche. A chrome yellow Porsche would normally look yellow in
daylight. But suppose Pia’s Porsche, which has been painted chrome yellow, is cov-
ered in baked-on beige mud. Should we say that the Porsche looks chrome yellow
or beige? According to Travis, this is a question to be settled by rational discussion
by agents who have grasped how a Porsche ought to look in daylight. One cannot
expect there to be a pre-determined answer to the question of when it is true to
say that a Porsche looks chrome yellow. As only rational agents can settle the ques-
tion of what a Porsche looks like in daylight, perceptual looks cannot play the role
of determining a unique representational content. So, perceptual experience does
not have content even in a minimal sense.
A second argument against the idea that perceptual experiences can do more
than effect-represent is that representing requires performing certain selection
tasks. For example, whether an object counts as a knife depends on context. Suppose
Sid is making the table and wants to know where the knives are. Pia replies with
‘There are knives in the third drawer’. Even if the third drawer contains matte knives
among other art supplies, she has conveyed something false to Sid. In the present
context, a matte knife does not count as a knife. What the word ‘knife’ means in
the present context depends on the selection tasks the conversationalists (or the
speaker) are performing. Performing a selection task requires cognitive abilities that
only agents can possess. Since mental states cannot perform these selection tasks,
they cannot allo-represent. But we cannot derive content from a sign and the thing
it effect-represents. So if mental states cannot allo-represent, then they do not have
content even in a minimal sense.
One might wonder which part of Siegel’s (2010) argument for thinking that
perceptual experience has content in a minimal sense Travis would reject. As we
saw above, Siegel’s argument can be summarized as follows.
 Introduction
14   

The Argument from Appearances

1. All visual (perceptual) experiences E make it seem to the subject that
clusters of properties F are instantiated.
2. So, necessarily, things are as they seem only if F is instantiated. (from 1)
3. So, E has a set of accuracy conditions C that correspond to how things
seem to the subject. (from 2)
Conclusion: All visual (perceptual) experiences have content.

Even if we assume that the environment is hospitable (e.g., there is no fake lighting
or deceitful plastic surgery), Travis would probably balk at the move from 2 to
3. This is because seemings that determine a unique content rely on the agent’s
rational abilities. But seemings that depend on the agent’s rational abilities cannot
serve as the source of content for visual experiences.
One might question Travis’ use of appear words. An agent’s cognitive system
may determine that a Porsche painted chrome yellow but covered in mud ought to
be said to look yellow. In these circumstances it would be true to reply to someone
who asks ‘What does Pia’s Porsche look like?’ with ‘Yellow’. This, however, is what
Roderick Chisholm (1957) calls the comparative use of ‘look’ (see also Jackson 1977).
‘Pia’s Porsche looks yellow in daylight’ cashes out to ‘Pia’s Porsche looks one of the
ways yellow Porsches look in daylight’. Following standard linguistic analysis, this
can be analyzed as ‘there is an x such that x is a way yellow Porches look, and Pia’s
Porsche looks x’ (Brogaard 2012a, b). This analysis makes unreduced appeal to a no-
tion of ‘look x’. This latter notion of ‘look’ is the non-comparative (perceptual) use.
In the circumstances in which Pia’s Porsche is covered in mud, it non-comparatively
looks beige. So, one might argue, the non-comparative looks do determine a par-
ticular set of accuracy conditions, independently of the agent’s rational abilities.
However, this reply does not get to the heart of the matter. Consider a per-
ceptual experience of a white wall that is partly illuminated by sunlight. On a very
restricted notion of non-comparative looks, the wall does not look uniformly col-
ored. However, thinkers who hold that the experiences represent would want to
say that the experience of the wall represents the wall as white. More generally, if
perceptual experiences represent, they ought to represent color constancies as well
as size and shape constancies. But it follows, then, that perceptual seemings cannot
be associated with one set of accuracy conditions independently of how our brains
happen to calculate these color-, size-, and shape-constancies.
Diana Raffman: In “Disjunctivism, Discrimination, and Categorization”,
Diana Raffman starts out by exposing some misconceptions of the perceptual
indiscriminability relation as it figures in recent treatments of disjunctivism by
Martin (2004, 2006) and Siegel (2004). As Siegel points out, one needs a way to
characterize the relevant pairs of perceptions and hallucinations to state the debate
between disjunctivism and its opponents. Introducing the notion of indiscrim-
inability has been the standard way of meeting this need. Raffman, however, is
skeptical that this relation can satisfy the need.
 Introduction   15

Like many other philosophers, Martin takes the relation of indiscriminability

to be nontransitive. Raffman questions this claim on empirical grounds. Raffman
completed a study in which subjects were asked to report on the indiscriminability
of neighboring patches in a phenomenal continuum and actively adjust the set-
tings on a wavelength measure. The study showed that while subjects report that
the pairs of patches have the same color, they nonetheless adjust the color measure
as they proceed along the continuum. According to Raffman, this indicates that
the participants fail to notice the color changes despite perceiving them.
After taking issue with the nontransitivity of the indiscriminability rela-
tion Raffman argues that the standard conception of the relation is incoherent.
Only external stimuli, she argues, can stand in relations of indiscriminability
to one another. Mental states or phenomenal characters cannot. Yet external
stimuli, even if physically identical and viewed under identical conditions, do
not invariably appear the same. Indiscriminability thus cannot be understood
as a relation of sameness but must be treated as a statistical notion. To a first
approximation: stimuli are indiscriminable just in case they appear the same in
75 percent of same/different comparisons, or some other percentage, depending
upon the experimenter’s explanatory goals.
This insight, Raffman argues, enables a new conception of phenomenal con-
tinua. She proposes to treat them as cases in which subjects asked whether pairs
of color patches are the same fail to notice color changes that they do perceive.
This new conception of phenomenal continua resolves the air of mystery sur-
rounding them. The phenomenology of phenomenal continua appear enigmatic
because, even given perfectly constant viewing conditions, the first and last
items appear different and yet nowhere between the two do we notice any local
difference in appearance. Distinguishing between a noticed and a perceived
color change explains this phenomenon.
Though Raffman is skeptical of formulations of disjunctivism in terms of the
notion of indiscriminability, she ultimately offers a positive proposal on behalf of the
disjunctivist. Raffman believes the difficulties disjunctivists are facing when attempt-
ing to formulate the very position they are defending rest on a conflation of two types
of judgments that are importantly different: categorical judgments and mere same/
different judgments. A categorical judgment is a type-identification that takes place
via recognition. Mere same/different judgments merely specify whether two stimuli
are the same or different in a given respect without type-identifying them. For ex-
ample, a perceiver may be able to tell that two ripe tomatoes have different colors
without being able to type-identify them as, say, red-32 as opposed to red-33. Raffman
could characterize the commonality between veridical experience and hallucinations
in terms of mere same/different judgments. On this view, an event is an experience
as of Φ just in case the subject cannot tell it apart from a veridical experience of Φ by
introspection.
One potentially contentious point in Raffman’s argument is her interpreta-
tion of the empirical study. The study showed that while subjects report that the
 Introduction
16   

pairs of patches have the same color, they nonetheless adjust the color measure as
they proceed along the continuum. Raffman suggested that this indicates that the
participants fail to notice the changes despite perceiving them. However, if the
changes were consciously perceived, then this proposal seems to rest on a con-
troversial assumption to the effect that one can have perceptual experience in the
absence of attention.
Even if perceptual experience can overflow selective attention, Raffman’s in-
terpretation could be questioned. The subjects in the study report that they don’t
perceive a change in color but nonetheless adjust the color measure as they move
about in the perceptual continuum. Their adjustment of the color measure could
reflect a sensitivity to wavelength in the visual dorsal stream. Milner and Goodale
originally took dorsal stream processes to be responsible for blindsight—the ability
had by people with V1 lesions to reliably report color and other attributes of visual
stimuli that are not consciously seen (Goodale et al. 1991; Goodale and Milner 1992).
This hypothesis has later been countered. However, the question of the extent to
which the dorsal pathway can discriminate among color stimuli remains largely
unexplored.

1.4. Reconciliatory Views

Susanna Schellenberg: In her contribution to this volume Susanna Schellenberg

offers a reconciliatory position that takes perceptual experience to be fundamen-
tally a matter of representing the environment in a certain way and being percep-
tually related to objects in the environment. She articulates the two views she is
committed to as follows:
Representational View
Perceptual experience is fundamentally a matter of representing the environ-
ment as being a certain way.

Relational View
Perceptual experience is fundamentally a matter of being perceptually related
to objects in the environment.
The two views are commonly thought to be in opposition, because it is assumed that
if perception is fundamentally characterized by its representational properties, then it
cannot be fundamentally characterized by its relational properties. It is this common
belief that Schellenberg takes issue with in her chapter. While Schellenberg defends
both views, others reject the whole package. Sense-data theorists, for example, take
perception to be a relation between a perceiver and sense-data, and adverbialists
take perception to be an act of perceiving in a certain way. So, sense-data theorists
and adverbialists would have to reject both views as stated.
 Introduction   17

Schellenberg distinguishes the representational view from what she calls the
‘Association Theory’.
Association Theory
Every experience can be associated with (propositional) content in the sense
that sentences can be articulated that describe how the environment seems
to the subject, without the content expressed being a proper part of the
experience.
According to Schellenberg, virtually any theory of perceptual experience could
accept the association theory, as it merely requires that we can use language to
partially describe our perceptual appearances. The association theory is akin to the
weak content view. The main difference is that the association theory specifically
mentions how content comes to have accuracy conditions, viz., through descrip-
tion sentences.
After presenting an argument for the association theory, Schellenberg goes on
to defend the representational view. In broad outline, the argument can be sum-
marized as follows. In order for things to perceptually seem a certain way to us, we
need to employ discriminatory selective capacities that constitute the seeming. For
example, if it seems to me that the fire truck is red, I must be able to discriminate
red from green. But employing these capacities in this way just is to represent the
environment in virtue of using these capacities. So, Schellenberg argues, percep-
tual experience is fundamentally a matter of representing the environment in a
certain way, which is the first principle.
The step from the last premise in this simplified presentation of her argument
to the conclusion rests on a particular epistemic view about perceptual experi-
ence. It seems incontestable that to use certain discriminatory selective capacities
that constitute seemings just is to represent one’s environment in a certain way.
However, one might question that this shows that it’s the perceptual experience as
opposed to the state of seeming itself that represents the environment. To get to the
conclusion that perceptual experience fundamentally is a matter of representing
the environment a certain way, it could be argued that a further premise is needed,
viz., the premise that if certain discriminatory selective capacities are required in
order for a certain state of perceptual seeming to obtain, then these discriminatory
selective capacities are also required in order for the underlying perceptual experi-
ence to obtain.
One might object to this last implicit premise. On the liberal view, perceptual
experience can represent high-level properties such as being a vine leaf maple,
being a cork screw, being sad, and being English. If the liberal view is false, then
I cannot have a perceptual experience that represents a vine leaf maple. But it can
nonetheless still perceptually seem to me that the tree is a vine leaf maple. Certain
discriminatory selective capacities, then, are required in order for a certain state
of perceptual seeming to obtain. But, if the liberal view is false, then it doesn’t
 Introduction
18   

follow that the same discriminatory selective capacities are required in order for
the underlying perceptual experience to obtain.
A simple move to avoid this consequence would be to adopt a liberal view
with respect to perceptual properties. While some would be happy with this as-
sumption (Siegel 2005), others might think the liberal view is too controversial to
assume up front (see, e.g., Brogaard 2013). Siegel (2005) offers an argument for the
liberal view. However, her argument does not show that it’s perceptual experiences
as opposed to perceptual seemings that represent high-level properties. So it is not
conclusive in the present context. I return briefly to the debate about high-level
properties below when discussing William Lycan’s chapter.
Heather Logue: In her contribution to this volume, Heather Logue defends a
version of the weak content view as well as the claim that three different content
views of varying strengths are compatible with naïve realism. The three content
views can be characterized as subsets of the following four assumptions:
1. There is a proposition associated with E, and
2. This proposition specifies the way things perceptually appear to the
subject in virtue of having E.
3. Perceptual experience consists in the subject perceptually representing her
environment as being a certain way.
4. Perceptual experience fundamentally consists in the subject perceptually
representing her environment as being a certain way.

According to Logue, (1) and (2) are the bare minimum that one is committed to when
one says that experience has content. Assumptions (1)‒(3) constitute the medium con-
tent view and (1)‒(4) constitute the strong content view. Assumption (4)is stronger
than (3) because it presupposes that some element of perceptual experience, for ex-
ample, its phenomenal character or some aspect of its epistemological role supervenes
on the subject perceptually representing her environment as being a certain way.
Logue argues that one of the main reasons naïve realists are reluctant to accept
the view that perceptual experience has content originates in their belief that accept-
ing this position would make naïve realism redundant. They seem to think that the
assumption that perceptual experience has content (even in a minimal sense) is suf-
ficient to explain its epistemological, behavioral, and phenomenal features. Because
naïve realists believe that accepting even the weak content view makes naïve realism
redundant, they are willing to bite the bullet and say that there is no proposition asso-
ciated with perceptual experience that specifies the way things perceptually appear.
It may seem that beliefs grounded in perceptual experience ought to have
contents that specify the way things perceptually appear. If this is so, and all per-
ceptual experiences give rise to (implicit or explicit) beliefs, then we can say that
the contents of beliefs grounded in perceptual experiences compose the contents
of those experiences. This would be a variation on Reid’s (1983) position.
However, as Logue points out, beliefs based on perceptual experience bear
not just on experience but also on background beliefs. So, we cannot read off the
 Introduction   19

content of a perceptual experience from the contents of the beliefs to which it

naturally gives rise. For example, seeing water pouring down outside may give
rise to the belief that it’s raining. But the inference from the content of the per-
ceptual experience to the content of the belief rests on the assumption that if
water is pouring down outside, then it’s raining. So, the content of the belief, viz.,
that it’s raining, does not reflect the content of the experience.
Even if we set aside this concern, the idea that the content of perceptual
experience just is the content of an inferred belief faces challenges. In the
Müller-Lyer illusion, for example, two equal-sized line segments perceptually
appear to be unequal-sized (­figure 1.2). People familiar with the illusion don’t
believe that the line segments have different lengths because they have defeat-
ers stating that the line segments are equal-sized. So, there is no belief that is
grounded in the perceptual experience of the line segments being unequal-sized.
A further problem with the view that the content of perceptual experience just
is the content of an inferred belief is that perceptual experience is rich in a way that
belief is not. Because of the richness of experience, it’s unlikely that we have beliefs
corresponding to all elements of every perceptual experience. As the view that ex-
perience has content concerns all experiences and not just a subset, the content of
experience is not just the same as the content of belief originating in the experience.
Logue goes on to consider Siegel’s (2010) argument for the weak content
view. She argues that the argument is unlikely to convince naïve realists who deny
that perceptual experience can be inaccurate. If perceptual experience cannot be
inaccurate, then it cannot really be assessable for accuracy. So, denying that per-
ceptual experience is assessable for accuracy is not as hard a bullet to swallow as it
may initially seem. As Logue points out, the view that experiences are assessable
for accuracy is not an intuitive dictum but a theoretical presupposition.
According to Logue, however, there is a different argument for the weak con-
tent view that she thinks that naïve realists ought to accept, viz., the argument
from belief generation. The argument can be summarized as follows: at least some
perceptual experiences automatically give rise to beliefs. But this raises the ques-
tion of why perceptual experiences give rise to particular beliefs and not others.

FIGURE 1.2 The Müller-Lyer illusion In the Müller-Lyer illusion you believe the lines are of the
same lengths but no matter how long you look, your continue to experience them as having
different lengths. This illustrates a case in which perceptual information is encapsulated from
belief influence.
 Introduction
20   

For example, why does my experience of a blue car in good viewing conditions
give rise to the belief that the car is blue and not to the belief that the car is black?
The most natural answer to this question is that perceptual experience is associ-
ated with a proposition that captures how things perceptually appear to be, and
that this proposition constrains the content of my belief. According to Logue,
naïve realists are more likely to be willing to accept that some perceptual experi-
ences give rise to beliefs than they are to accept that perceptual experiences are
assessable for accuracy.
Logue doesn’t think that accepting the weak content view shows that there
is no room for naïve realism. According to her, the assumption that perceptual
experience has content (even in a minimal sense) is unlikely to be sufficient to
explain its epistemological, functional, and phenomenal features. For a strong rep-
resentationalist, the content of experience is sufficient to explain the phenomenal
features of experience. But another option is to say that its phenomenal features
derive from the subject standing in certain perceptual relations to entities in her
environment. If this is so, then veridical experience fundamentally consists both in
the subject’s perceptually representing her environment as being a certain way and
as being perceptually related to entities in her environment. According to Logue,
this last option amounts to a reconciliation of naïve realism with the strong con-
tent view.
Logue’s argument from belief generation, as formulated, only shows that
the weak content view is correct. However, it might be extended to provide
evidence for the strong content view: If things perceptually appear to the sub-
ject a certain way, we can say that the subject perceptually represents her en-
vironment as being a certain way, perhaps in virtue of having certain selective
discriminatory capacities, as Schellenberg argues. For example, if a blue car
perceptually appears to me to be blue, then I perceptually represent a car as
being blue in virtue of my abilities to discriminate between blue and other
colors. Moreover, if things perceptually appear to the subject a certain way,
then she is more likely to form a particular belief rather than another. For
example, if a blue car perceptually appears to me to be blue, then I am more
likely to form the belief that the car is blue than the belief that the car is some
other color. But an aspect of the epistemological role of experience is to con-
strain the content of belief originating in the experience. So, this aspect of the
epistemological role of experience is grounded in the way the subject percep-
tually represents her environment as being. In other words, if the content of
perceptual experience is what constrains the contents of beliefs originating
in the experience, an aspect of the epistemological role of perceptual experi-
ence supervenes on the subject perceptually representing her environment as
being a certain way. So, as Logue cashes out the meaning of ‘fundamentally’,
perceptual experience fundamentally consists in the subject perceptually rep-
resenting her environment as being a certain way. So, the strong content view
is correct.
 Introduction   21

There is an implicit assumption in Logue’s argument that is not stated in her

chapter. Though it is incontestable in my opinion that perceptual experience some-
times gives rise to belief, it is not obvious that it elicits belief in a direct way. Visual
experience may give rise to a visual seeming that then elicits belief. If visual seem-
ings and visual experiences come apart, what constrains the content of belief is
the visual seeming and not the visual experience. It follows that visual seeming
has content but not that perceptual experience does. The issue of whether visual
experiences and visual seemings come apart is a tricky one. Ernie Sosa has offered
the following argument for thinking that visual experiences and visual seemings
come apart:
Take for instance the look of an empty chessboard as it is viewed up close
in bright light. The array that then gives content to the subject’s experience
involves 64 alternating black or white squares. Yet if this is A’s first encounter
with a chessboard, the proposition that she faces such an array may hold no
attraction for her. Someone B familiar with chessboards will of course be
attracted to assent to that proposition, and the attraction will presumably be
prompted by the experience shared with A. So, the experience should be dis-
tinguished from the seeming. (2009, 137)
The argument is this. We cannot perceptually distinguish between 64 entities
and a number of entities in the close vicinity of 64. So, if I have a perceptual
experience of a chessboard but I am completely unfamiliar with chessboards,
then it will not perceptually seem to me that the thing in front of me has 64
alternating black and white squares. If, on the other hand, I have a perceptual
experience of a chessboard and I am familiar with chess boards, then it will
perceptually seem to me that the thing in front of me has 64 alternating black
and white squares. The perceptual experience could be the very same in the two
cases. But the perceptual seemings come apart. So, perceptual experiences and
perceptual seemings come apart.
One simple reply to this argument would be to deny that the perceptual
experiences are the same in the two cases. For example, one could hold that the
experienced subject has a perceptual experience that, among other things, repre-
sents the number 64, whereas the inexperienced subject does not have a percep-
tual experience that represents the number 64. If one simultaneously holds that
perceptual seemings are aspects of perceptual experiences, then one can straight-
forwardly explain the difference between the perceptual seemings of the subjects.
Even if perceptual experiences and perceptual seemings come apart, how-
ever, no irreversible damage is done to Logue’s argument. The argument can be
re-stated as an argument to the best explanation for the conclusion that the con-
tent of perceptual experience constrains the content of perceptual seemings. The
argument runs as follows. At least some perceptual experiences automatically
give rise to perceptual seemings (e.g., the experience of the chessboard gives rise
to the perceptual seeming that the thing perceived has 64 alternating black and
 Introduction
22   

white squares). But this raises the question of why perceptual experiences give
rise to particular perceptual seemings and not others. The most natural answer
to this question is that perceptual experience is associated with a proposition that
constrains how things perceptually appear. Adapting the further argument for the
strong content view requires the further premise that perceptual seemings, un-
like perceptual experiences, are rationally based. But this is a natural stance to
take when maintaining that perceptual seemings and perceptual experiences come
apart (Sosa 2009a, 2009b).
Benj Hellie: In “Love in the Time of Cholera,” Benj Hellie argues for a dif-
ferent type of reconciliatory view that’s naïve-realism friendly in some but not all
respects.4 He takes perceptual states to be ecological states rather than inner, or
psychological, states but holds that all perceptual states have content in a minimal
sense. It is in this latter respect that his view is unfriendly to naïve realism, or at
least to those among naïve realists who deny the content thesis.
Hellie is not a big fan of the phrase ‘perceptual experience’. But if we need
to use it, he suggests that we take it to refer to a certain aspect of the stream
of consciousness, viz., that aspect which is one’s course of attentive stances.
The attentive stance delimits the perceptual part or aspect of the stream of
consciousness.
One of Hellie’s aims in his chapter is to refute a content-based case against
naïve realism. The argument against naïve realism that Hellie takes issue with can
be summarized as follows. Regardless of what naïve realists take hallucinations
to be, most are willing to admit that it can perceptually seem that one is engaged
with the environment. This seeming, however, can be inaccurate. So, the seeming
has content. As both veridical and non-veridical experience have content if one of
them does, naïve realism is false.
Hellie replies to this argument that it mistakenly assumes that perceptual
appearances are an aspect of perceptual experience. One of the arguments he pro-
vides is an argument from linguistics. According to Hellie, the perceptual verbs
‘look’, ‘feel’, ‘taste’, ‘smell’, and ‘sound’ are copular verbs just like ‘be’ and ‘become’.5
These verbs take an adjectival predicate as its their syntactic complements, as in:
(1)
(a) Andrea is/becomes/looks tall.
(b) Kim is/becomes/looks similar to a cat.
(c) Luke is/becomes/looks like a dog.

4
Hellie talks about direct realism, but I use the phrase ‘naïve realism’ here in the interest of a
uniform terminology.
5
Note that subject-raising verbs and copular verbs are sometimes all classified as copular verbs.
I don’t have a problem with this standard classification. If, however, ‘look’ is a copular verb, then it
should be more similar semantically to quintessential copular verbs, such as ‘is’ and ‘become’, than to
subject-raising verbs.
 Introduction   23

The perceptual copular verbs, Hellie points out, resist taking ‘that’ clauses as their
complements and only reluctantly take non-finite verb phrases (e.g., ‘to have had
a good time’) as their complements. In this respect they behave differently from
subject-raising verbs, such as ‘seem’, ‘appear’, ‘believe’, ‘prove’, ‘expect’, ‘turn out’,
‘find’, ‘deem’, and ‘assume’.
(2)
(a) It appears/seems that Sam is running for office.
(b) #It looks that Sam is running for office.
(c) My shoes seem to have been left out in the rain.
(d) ?My shoes look to have been left out in the rain.
Hellie takes this to suggest that the perceptual copular verbs do not operate syn-
tactically on clauses but on predicates, which means that they do not operate se-
mantically on propositions but on properties. Hellie proceeds from these linguistic
considerations to an account of perceptual copular verbs as expressing special
kinds of beliefs.
In previous work I have argued that ‘seem’, ‘look’, and ‘appear’ all function as
subject-raising verbs (see, e.g., Brogaard 2012b). Though I grant that ‘look’ does
not take ‘that’-clauses as its complement, I do not think this observation suffices
for treating it as semantically different from ‘seem’ and ‘appear’. I take this be-
havior of ‘look’ to be an irregularity of the verb. Even if Hellie is right that ‘look’
is a true copular verb, however, he does not provide a knockdown argument for
treating perceptual seemings or appearances as beliefs. The main reason for this is
that even though perceptual seemings can be expressed in terms of ‘look’, they are
equally well expressed in terms of ‘seem’ or ‘appear’, which uncontroversially are
subject-raising verbs.
Though the reconciliatory views defended by Schellenberg, Logue, and Hellie
are sympathetic to the content views, they may still face some of the same chal-
lenges as the strong versions of naïve realism. By insisting that perceptual experi-
ence fundamentally is a matter of being perceptually related to objects and features
in the environment, they may face problems along the lines of those raised by
Bence Nanay (this volume). By holding that perceptual experience fundamentally
is a matter of being related to the environment, they need to specify in which ways
ventral and dorsal stream representations differ in this regard. The visual ventral
and dorsal pathways appear to attribute different properties to objects in the envi-
ronment. Does this mean that there are two ways of being related to the environ-
ment: A dorsal way and a ventral way? Or is one type of vision not fundamentally
a matter of being related to the environment?
There are also ways in which the reconciliatory accounts may fare better than
their naïve realist cousins. For example, they can agree with Nanay that standing
in perceptual relations to the environment presupposes integrating multisensory
information in a representational form. Given a reconciliatory view, it is unprob-
lematic to grant that this is so.
 Introduction
24   

1.5. Imagistic and Possible-World Content

Michael Tye: So far I have spoken somewhat abstractly of perceptual content

without saying what the nature of this content might be. Many of those who defend
the strong content view think that perceptual content is imagistic in the sense that
it is more similar to the content of an image than to the content of a sentence. On
one way of construing image content, image content represents an array of features
located in a visual field relative to the perceiver (Peacocke 1983, 1998, 2001; Matthen,
this volume) The features, however, are not represented merely as features. They
are represented as belonging or not belonging together. Those that belong together
form figure and ground that possess the features. Some of the figures are material
objects. Others are shades, shadows, and patches of light. When the features that
group together to form figures change, the figures are seen as changing or moving
against the background.
This view is consistent with an understanding of content as sets of possible
worlds. In his contribution to this volume, Tye argues against his earlier gappy
content view (Tye 2007) and defends a possible-world account of perceptual con-
tent. His earlier view was inspired by the problem posed by hallucinations. If the
content of veridical perception is object-dependent, it contains concrete particu-
lars as constituents of the content or supervenes on such concrete particulars. In
hallucinations, however, there is no object corresponding to what is perceived. So,
it’s natural to think that hallucinatory content must be gappy.
In philosophy of language, a similar strategy has long been employed to deal
with the problem of empty names. Russellians, who think that the semantic value
of a proper name is its referent, must provide some explanation of how to deal
with proper names, such as ‘Odin’, ‘Santa Claus’, and ‘Sherlock Holmes’, that do not
refer to an actual concrete particular. One commonly employed strategy is to say
that the Russellian propositions semantically expressed by sentences containing
empty names contain a gap. For example, where the sentence ‘Obama is President’
expresses a proposition of the form <a, President>, the sentence ‘Sherlock Holmes
is a fictional detective’ expresses the proposition <___, fictional detective>.
Russellians typically supplement this view with a story about how these sentences
pragmatically imply descriptive propositions.
Following the lead of Russellians, some philosophers of mind have taken hal-
lucinations to have gappy content (Chalmers 2004; Tye 2007; Schellenberg 2011;
see also Siegel 2011). For example, if I am having a hallucination of a red spider
crawling on my arm, part of the content would be of the form <___, red>.
In his contribution to the present volume Tye retracts his early gappy-content
view. His skepticism about this view derives from difficulties making sense of
what a gap is supposed to be. A structured Russellian proposition is an n-tuple
of constituents. For example, the proposition Mont Blanc is 4,810 meters high is
an n-tuple containing Mont Blanc and all its snowfields and the property of being
 Introduction   25

4,810 meters high. If, however, there is no concrete particular to serve as the first
constituent of the pair, what takes its place?
There are two possible answers to this question. Either nothing takes its place
or something takes its place. If nothing takes its place, then it would seem that
the propositions expressed by sentences containing empty names and hallucina-
tions are properties. However, we can quickly rule out this possibility, as properties
and propositions are of different semantic types. If, on the other hand, some sort
of placeholder takes its place, then the proposition cannot be the content of the
mental state or sentence in question. For example, suppose the proposition has
the form of a propositional function Fx. Theorists who hold that sentences express
propositional functions, such as Irene Heim (2006; see also Brogaard 2012a), take
the functions to be existentially closed by default. So, we would get ∃x(Fx). But
this says that something is F. Surely, the content of my hallucination of a red spider
is not the proposition that something is red. If, on the other hand, x is a variable,
a concrete-particular hole or a concrete-particular spatio-temporal location, then
the content says that the variable or hole or spatio-temporal location x is red. This,
too, cannot be the content of my hallucination of a red spider.
Based on these types of considerations Tye dismisses his old view that the
contents of perception are Russellian propositions and proposes two alternatives.
The first is that the contents of perception are sets of possible worlds. This view
is very similar to the view that perception has scenario content (Peacocke, 1983,
1992). Scenario content is normally thought to be a set of centered worlds, spe-
cifically worlds that include a perspective. On this view, the content of a veridical
perception of a particular red spider is the set of possible worlds that contains that
particular red spider. In the case of a hallucination of a red spider, there is no red
particular red spider. So, there is no set of worlds that can constitute the content.
The content, then, is the empty set.
The other suggestion Tye makes is that propositions are structured but that
they are not composed out of properties and/or objects. Instead they are com-
posed out of Kaplanian characters. David Kaplan (1989) proposes that expressions
have a (semantic) character that determines their semantic value. Characters are
functions from a context of utterance to a semantic value. For example, indexicals
such as ‘I’ and ‘now’ are functions from a context of utterance to the speaker and
the time of the context and proper names are functions from a context to the ref-
erent of the proper name.
If we apply this idea to the case of perception, we can say that the content of
a perceptual experience of an object that has a certain property would consist of a
function from the perceptual situation to the object that corresponds to the phe-
nomenal character and a function from the perceptual situation to the property
that corresponds to the phenomenal character. The content of a veridical or hallu-
cinatory perception of a red spider, then, is a structured proposition that consists
of a function from a perceptual situation to the object that corresponds to the
 Introduction
26   

phenomenal character and a function from a perceptual situation to the property

that corresponds to the phenomenal character.
Note here that the Kaplanian view requires that the constituents of the content
of perceptual experience depend on the phenomenal character of the experience.
It would not do, for example, to take the constituents of the content of a halluci-
nation of a red spider to be a function from a perceptual situation to a perceived
spider and a function from a perceptual situation to a property the spider is per-
ceived to have. After all, no concrete particular spider is perceived and introducing
a purely perceptual entity (a spider percept) to account for the contents of halluci-
nation would make the Kaplanian twist redundant.
Though Tye does not ultimately choose one of the two views, he points out
that the Kaplanian view has the virtue that it’s compatible with a version of rep-
resentationalism that he favors, viz., the view that phenomenal character super-
venes on content. The first view predicts that the content of all hallucinations is the
empty set. So, the content of my hallucination of a red spider has the same content
as my hallucination of a blue worm; viz., the empty set. But the two hallucinations
have different phenomenal characters. So the first view is at odds with the version
of representationalism Tye favors. The second view appears to be doing better in
this regard. Since the constituents of the propositions are functions from percep-
tual situations to entities corresponding to the phenomenal character, hallucina-
tions with different phenomenal characters will inevitably have different contents.
So, my hallucination of a red spider and my hallucination of a blue worm have
different contents.
Tye does not show how the perceiver’s perspective comes to set its mark on
perceptual experience. One possibility is to take the worlds that form elements of
the content of perceptual experience to be centered worlds (see, e.g., Brogaard,
forthcoming b). A centered world is a world in which certain features are marked,
typically an individual and a time. On this proposal, perceptual content would be a
set of centered worlds, which would have a truth-value relative to the actual world.
For example, if you see a blue car in front of you, the content of your perceptual
experience would be a set of centered worlds in which there is a blue car in front
of the marked individual. This content is true only if there is a blue car in front of
the actual perceiver.
Mohan Matthen: In his contribution to this volume Mohan Matthen offers
an account of perceptual content that avoids certain arguments against the strong
content view. Matthen takes the content of a sensory image (e.g., a visual image) as
well as perceptual experience to consist of existentially quantified propositions that
can be represented with the schema <sortal S, feature F, location L>. This schema
is to be read as ‘some S with feature F occupies a viewpoint-dependent location L’.
This structure is logically simple and cannot express negation or quantification,
with the exception of an existentially quantified form of the feature structure. So
no image corresponds to ‘there is a man somewhere or other’ or ‘there are more
 Introduction   27

than three men on the bench’. The objects and their features are presented as spa-
tially and temporally located relative to each other.
One virtue of taking perceptual experience to have imagistic content is that it
avoids certain arguments against the strong content view. According to Matthen,
linguistic reports of what was perceived may be theory-laden in mentioning only
how things perceptually look to the perceiver. For this reason, linguistic reports
need not reflect what was really perceived. Unlike the content of linguistic reports,
imagistic content is not theory-laden, as it differs from the content of the sentences
used to describe what we perceive. For example, if an art critic is looking at an
artistic garden, she may describe it as ‘beautiful’. The image itself, however, does
not represent the garden as beautiful. Whether it is beautiful or not requires an
application of a theory about artistic gardens.
According to Matthen, the view that the content of perception is imagistic
may appear to raise independent problems. In imagistic memory objects and fea-
tures are presented as located in the past relative to the perceiver, and in visual
imagery objects and features may not be temporally or spatially located relative to
the perceiver at all. Perceptual experience, on the other hand, represents objects
and features as present here and now. But if the contents of both mental imagery
and perception are imagistic, how do we explain the fact that only perceptual expe-
rience essentially represents objects and features as located relative to the here and
now of the perceiver?
It may be thought that perceptual content contains temporal and spatial re-
lations to the here and now of the perceiver. However, Matthen thinks there are
reasons to deny this. His argument runs as follows:
Shared Content Argument
1.  The content of an imagination does not contain relations to the here and
now of the perceiver.
2.  But a perceptual experience and an imagination can have a shared image.
Conclusion: The content of perception does not contain relations to the
here and now of the perceiver.
But if the content of perception does not contain relations to the here and now of
the perceiver, then how can perceptual experience represent objects and features
as intimately tied to the here and now of the perceiver? Matthen suggests that
there is a cognitive feeling that marks the image as presented as an image of the
here and now. Similarly, there is a feeling of pastness that accompanies the images
of episodic memory.
Both Tye and Matthen assume that perceptual experience has content and
focus on making a positive proposal about the nature of this content. As we have
seen, it’s not possible to draw any definitive conclusions about whether perceptual
experience has content without saying something about what the content is like, if
indeed perception has content.
 Introduction
28   

1.6. Constituents of Perceptual Content and the Role of Perception

William Lycan: As we have seen, several of the arguments for and against the view
that perception has content make particular assumptions about the constituents of
perceptual content and the role of perception. As such, it doesn’t seem that debates
about whether perception has content can be settled in isolation from debates
about the role of perception and the constituents of perceptual content, if indeed
perceptual experience has content.
The question of the constituents of perceptual content, in turn, depends on
other debates in philosophy of mind. One important question is whether percep-
tion can be meaningfully separated from cognition.
According to the modularity of mind hypothesis, systems involved in pro-
ducing particular mental states or mental abilities are modular (Fodor 1983;
Sperber 1994, 2002; Pinker 1997; Carruthers 2006). Beyond its core premise
of modularly describing what different regions of the brain do, the modularity
hypothesis also states that certain kinds of information are encapsulated from
influences from other regions (Fodor 1983). For example, the Müller-Lyer illu-
sion illustrates that perceptual information is encapsulated from belief influence.
It may be understood and believed that the lines are the same length between
the two images, but nevertheless they continue to perceptually appear to be of
different lengths.
For information to be encapsulated it does not suffice that there are no
top-down influences on producing it; it must also be free of influences from other
modules. So the Müller-Lyer illusion does not demonstrate that perception is
modular under a strict interpretation of ‘modularity’. However, it does suggest that
there is some meaningful distinction to be drawn between perceptual states and
cognitive states.
William Lycan grants a weak version of the modularity hypothesis: Even if
there is a gradual transition from truly perceptual states to truly cognitive states,
he thinks it makes sense to talk about perceptual states as distinct from cognitive
states. But he doesn’t believe there is any way to settle the question of which fea-
tures perceptual states represent. If he is right about this, this might have conse-
quences for the debate about whether perception has content.
After considering and rejecting various ways that one might settle what
perceptual states represent, Lycan turns to Siegel’s argument from phenomenal
contrast (2005). Let E1 be a visual experience of someone who has the ability to
recognize elm trees (expert) and who is looking at an elm tree, and let E2 be the
visual experience of someone who does not have the ability to recognize elm trees
(novice) and who is looking at the same tree in the same viewing conditions. The
expert finds the tree familiar, the novice does not. So there is a difference in the
overall phenomenal character of their (perceptual or non-perceptual) experiences.
Where ‘K-properties’ is short for ‘high-level kind properties’, Siegel’s argument can
be articulated as follows:
 Introduction   29

The Argument from Phenomenal Contrast

1. The overall phenomenology of which the phenomenology of E1 is a part
differs from the overall phenomenology of which the phenomenology of
E2 is a part (familiarity effects).
2. If the overall phenomenology of which the phenomenology of E1 is a part
differs from the overall phenomenology of which the phenomenology of
E2 is a part, then there is a phenomenological difference between E1 and
E2 (cognitive penetration).
3. If there is a phenomenological difference between E1 and E2, then E1 and
E2 differ in perceptual content (weak representationalism).
4. If there is a difference in perceptual content between E1 and E2, it’s a
difference with respect to the K-properties represented by E1 and E2.
Conclusion: There is a difference in the K-properties represented by E1
and E2.

Unlike other strategies, Lycan thinks Siegel’s argument holds some initial
promise. However, he thinks that it ultimately fails to settle the debate. One
problem with Siegel’s argumentative strategy, according to Lycan, is that it seems
to over-generate. As formulated, it is restricted to natural kind properties, but the
very same argument can be used to argue that the constituents of perception in-
clude high-level natural kind properties can also be used to argue that the constit-
uents of perception include artificial kind properties (e.g., being a clock radio),
mental state properties (e.g., being depressed), aesthetic properties (e.g., being
gloomy), moral properties (e.g., being a virtuous agent), personal taste properties
(e.g., being attractive), mathematical properties (e.g., being 64 alternating black
and white squares), and event properties (e.g., being a car crash).
To illustrate consider a six-year-old who has not had any art classes and a
skilled art critic. Let E1 be a visual experience of the skilled art critic who is looking
at Edvard Munch’s painting The Scream, and let E2 be the visual experience of the
six-year-old who is looking at the same painting. The art critic has the recogni-
tional abilities to pick out the painting as being Edvard Munch’s The Scream, being
an oil on cardboard painting and being completed in Oslo in 1893. The child does
not. So there is a difference in the overall phenomenal character of their (percep-
tual or non-perceptual) experiences. By running through the argument, we can
presumably get to the conclusion that the overall difference in phenomenology
between E1 and E2 is a difference with respect to their kind-properties, in this
case being Edvard Munch’s The Scream, being an oil on cardboard painting and
being completed in Oslo in 1893. Siegel’s argument for a moderately liberal view
thus seems equally supportive of an extremely liberal view that grants that we per-
ceive extremely high-level properties, such as being Edvard Munch’s The Scream or
being created in 1893. The main problem with the extremely liberal view is that it
doesn’t seem to allow for a non-trivial distinction between perceptual and cogni-
tive states. But if aesthetic properties and other extremely high-level properties are
 Introduction
30   

not among the constituents of perception despite contributing to the differences

in phenomenal character between our overall experiences, then Siegel’s argumen-
tative strategy cannot be used to settle the debate about whether perception repre-
sents any high-level properties.
Terry Horgan: In “Phenomenal Intentionality and Secondary Qualities: The
Quixotic Case of Color,” Terry Horgan offers an extension of his phenomenal
intentionality view to color (see, e.g., Horgan and Tienson 2002; Graham et al.
2007, 2009). According to Horgan, consciousness is richly and pervasively repre-
sentational, or intentional. While various different vehicles represent, including
thought and inner speech, the most fundamental kind of mental intentionality is
phenomenal intentionality.
Phenomenal consciousness, the qualitative and subjective character of ex-
perience, is a narrow and intrinsic feature of mentality. Phenomenal conscious-
ness does not constitutively depend on anything outside the skull or outside itself.
Horgan grants that phenomenal consciousness may metaphysically supervene
on physical facts. Physical stuff, however, is not a constituent of phenomenal
consciousness.6
Phenomenal intentionality, Horgan argues, is entirely constituted by phe-
nomenology. Because phenomenal intentionality is narrow, there is an exact
match of phenomenal intentionality between you and your envatted brain phys-
ical duplicate. So, there is no difference between your perceptual experiences and
those of your envatted brain. The difference between the mental life of you and
your envatted brain lies elsewhere. The intentionality of belief, for example, is not
entirely made up by phenomenal intentionality. The intentionality of beliefs con-
stitutively depends both on phenomenal intentionality and the external environ-
ment. Because there is not a good match between the phenomenal intentionality
of your envatted brain and the external environment, your envatted brain’s beliefs
largely fail to be satisfied. All beliefs with object-dependent contents fail to refer
and hence are false.
There is thus a constitutive difference between phenomenal intentionality
and other types of intentionality. Phenomenality, even if non-veridical, provides
reference-constituting experiential acquaintance with the properties and relations
that make up its content. Beliefs do not provide this type of acquaintance with
the properties and relations that constitute their content. Beliefs come to refer
to external objects through what Horgan calls ‘grounding presuppositions’. If,
for example, you have a belief that we could express with ‘that picture is hanging
crooked’, the object-dependent content is associated with phenomenally consti-
tuted grounding presuppositions that are satisfied by the external environment.
These include the presupposition that there is an object at a certain location, that

6
Some use ‘constitutive dependence’ and ‘metaphysical supervenience’ synonymously. Here
I take it that Horgan distinguishes between the essence of phenomenal consciousness and what it
depends on metaphysically.
 Introduction   31

the object is a picture, that there is no other picture at that general location and
that the object is the cause of the crooked-picture experience.
Horgan extends this framework to color. His phenomenal intentionality proj-
ect requires that there are color properties that do not constitutively depend on
the external environment. This is so, because we are experientially acquainted
with color properties. Even your envatted brain has perceptual experiences of
red tomatoes. However, it is a further question whether these color properties are
also located where they perceptually appear to be located, viz., on the surfaces of
things, and whether these color properties are the constituents of the contents of
our beliefs and judgments.
Horgan follows David Chalmers (2006) in thinking that while the color prop-
erties we are directly acquainted with in experience could have been located on the
surface of things, objects in the actual world do not actually possess these kinds of
primitive color properties. So, our perceptual experiences are non-veridical.
However, Horgan doesn’t think that our beliefs and judgments about colors
and color discourse are false. Beliefs about concrete particulars can come to have
object-dependent contents. Likewise, our beliefs can come to have contents that
involve the grounds of our color experiences. These grounds, Horgan argues, are
dispositional color properties in Locke’s sense.
Though Horgan’s proposal at first glance seems rather radical in the content
direction, it appears to meet some of the concerns voiced by thinkers leaning in
the other direction. One of Charles Travis’ concerns about representational views
is that they leave the agent out of the equation. Travis should be less concerned
about an approach along these lines. According to Horgan, the most fundamental
kind of phenomenology is agent-centered. All representation derives from this
kind of intentionality, as far as Horgan is concerned.
Horgan’s view offers a potential reconciliation with some of the opponents of
the content views. The phenomenal intentionality thesis takes the agent to be a cen-
tral source of intentionality and offers an explanation of how this agent-centered
intentionality may be a source of both perceptual content and thought content. The
main objection from the Travis camp to this sort of approach probably wouldn’t be
that it construes representation independently of the agent who is doing the repre-
senting but rather that it treats phenomenology as representational independently
of any higher-order epistemic states.
Tomasz Budek and Katalin Farkas: Like most of the literature on perception,
the majority of chapters in this volume have focused on visual perception. In their
contribution to this volume, however, Tomasz Budek and Katalin Farkas con-
sider the more general question of the nature of the constituents of perceptual
experience in different sense modalities. Their main conclusion is that there is no
uniform answer to the question of what the objects of perception are, not even
within a given sense modality. Among the sense modalities discussed by Budek
and Farkas is the modality of olfaction. Budek and Farkas argue that the object
of olfaction sometimes is a chemical and sometimes is a substance or a thing.
 Introduction
32   

For example, they think that when we walk into a room in which coffee is being
brewed and take a big sniff, we can correctly be said to have smelled coffee odor
in some circumstances and coffee in others. In the envisaged case coffee odor and
coffee are different elements in the causal chain leading to the experience.
This observation is particularly interesting with respect to the debate about
perceptual content because the identity of the object of perception will determine
the accuracy conditions for the experience. If I have a perceptual experience of
coffee odor, then my experience can be accurate even if there is no coffee in the
room but only a chemist experimenting with odor chemicals. If, on the other
hand, I have a perceptual experience of coffee, then my experience can be accurate
only if there is coffee in the room. If, indeed, the accuracy conditions of perceptual
experience reflect the contents of perception, then my sniffing the very same odor
molecules in the same narrow (or internal) circumstances can amount to percep-
tual experiences with different contents.
Olfaction (and other sense modalities such as touch), Budek and Farkas
argue, differ in this respect from visual perception. While it’s plausible that the
objects of olfaction sometimes are odor chemicals, it is plainly implausible that
the object of visual perception sometimes is light absorbed by the cones of the eye.
That is simply not what we mean by the term of art ‘object of perception’. As the
expression is ordinarily used in the perception literature, an object of perception
has to be phenomenally present to the subject, which the light absorbed by the
cones of the eye clearly is not.
One could object to Budek and Farkas’ proposal that an element in the causal
chain leading to smells sometimes serves as an object of olfaction that it focuses too
much on causation. Just as light is not an object of visual perception, so odor chemi-
cals may not be objects of olfaction. Perceived odors, like perceived color properties,
may be largely brain constructs. Further phenomenological and empirical studies
would be needed to settle this question.
Budek and Farkas do not take a direct stance on the content debate. However,
their contribution raises the question of whether the heavy focus on vision in the
literature on perception is partially responsible for the resistance to the idea that
perception has content.
Consider, again, the case of blind individuals who have been taught to navigate
and detect objects via echolocation. A good number of these individuals develop
visual imagery that veridically represents objects in the individual’s environment. In
this case, the proximal causes outside the head are sound-related. But it is far-fetched
to think that these sound-related proximal causes are the objects of the visual experi-
ence. A more plausible assumption is that the visual experiences of these blind indi-
viduals represent visible features, making the weak content view seem appealing.7

7
I am grateful to Susanna Siegel and an anonymous reader from Oxford University Press for
detailed comments on an earlier draft of this paper.
 Introduction   33

References

Ayer, A. J. (1956). The Problem of Knowledge. London: Macmillan.

Breckenridge, W. (n.d.). A New Defence of the Adverbial Theory. Manuscript.
Brewer, B. (2006). Perception and content. European Journal of Philosophy, 14, 165–181.
Brogaard, C. D. (1925). The Mind and Its Place in Nature. London: Routledge & Kegan Paul.
Brogaard, B. (2011). Are there unconscious perceptual processes? Consciousness and
Cognition, 20, 449–463.
Brogaard, B. (2012a). What do we say when we say how or what we feel? Philosophers
Imprint, 12(11), June 2012.
Brogaard, B. (2012b). Perceptual reports. In Mohan Matthen (Ed.), Oxford Handbook of the
Philosophy of Perception. Oxford: Oxford University Press.
Brogaard, B. (2013). Do we perceive natural kind properties? Philosophical Studies, 162(1),
35–42.
Brogaard, B. “Vision for Action and the Contents of Perception”, Journal of Philosophy,
Volume 109, Issue 10, October 2012, 569–587.
Brogaard, B. “It’s not what it seems. A semantic account of ‘seems’ and seemings” Inquiry
56/2–3 (2013), 210–239
Brogaard, B., Marlow, K., & Rice, K. (2013). The long-term potentiation model for
grapheme-color binding in synesthesia. In David Bennett & Chris Hill (Eds.), Sensory
Integration and the Unity of Consciousness. Cambridge, MA: MIT Press.
Byrne, A. (2009). Experience and content. Philosophical Quarterly, 59, 429–451.
Carruthers, P. (2006). The Architecture of the Mind. Oxford: Oxford University Press.
Casullo, A. (1987). A defense of sense-data. Philosophy and Phenomenological Research, 48,
45–61.
Chalmers, D. (2004). The representational character of experience. In B. Leiter (Ed.), The
Future for Philosophy (pp. 153–181). Oxford: Oxford University Press.
Chalmers, D. (2006). Perception and the fall from Eden. In T. Gendler & J. Hawthorne
(Eds.), Perceptual Experience (pp. 49–125). Oxford: Clarendon Press.
Chisholm, R. (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University
Press.
Fish, W. (2008). Disjunctivism, indistinguishability, and the nature of hallucination. In
F. Macpherson & A. Haddock (Eds.), Disjunctivism: Perception, Action, Knowledge
(pp. 144–167). Oxford: Oxford University Press.
Fish, W. (2009). Perception, Hallucination, and Illusions. New York: Oxford University Press.
Fodor, J. A. (1983). The Modularity of Mind. Cambridge, MA: MIT Press.
Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action.
Trends in Neurosciences, 15, 20–25.
Goodale, M. A., Milner, A. D., Jakobson, L. S., & Carey, D. P. (1991). A neurological dissocia-
tion between perceiving objects and grasping them. Nature, 349, 154–156.
Graham, G., Horgan, T., & Tienson, J. (2007). Consciousness and intentionality. In M.
Velmans & S. Schneider (Eds.), Blackwell Companion to Consciousness (pp. 468–484).
Oxford: Basil Blackwell.
Graham, G., Horgan, T. & Tienson, J. (2009). Phenomenology, intentionality, and the unity
of mind. In B. McLaughlin, A. Beckermann, & S. Walter (Eds.), The Oxford Handbook
of Philosophy of Mind (pp. 512–537). Oxford: Oxford University Press.
 Introduction
34   

Heim, I. (2006). Remarks on Comparative Clauses as Generalized Quantifiers. Manuscript,

MIT.
Hinton, J. M. (1973). Experiences. Oxford: Clarendon Press.
Horgan, T., & Tienson, J. (2002). The intentionality of phenomenology and the phenom-
enology of intentionality. In D. J. Chalmers (Ed.), Philosophy of Mind: Classical and
Contemporary Readings (pp. 520–533). Oxford: Oxford University Press.
Jackson, F. (1977). Perception: A Representative Theory. Cambridge: Cambridge University
Press.
Kaplan, D. (1989). Demonstratives. In J. Almog, J. Perry, & H. Wettstein (Eds.), Themes from
Kaplan (pp. 481–563. New York: Oxford University Press.
Langsam, H. (1997). The theory of appearing defended. Philosophical Studies, 87, 33–59.
Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17, 376–425.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 120, 37–89.
Martin, M. G. F. (2006). On Being Alienated. In T. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 354–410). Oxford: Oxford University Press.
McDowell, J. (1982). Criteria, defeasibility, and knowledge. Proceedings of the British Academy,
68, 455–479; cited as reprinted in T. Baldwin and T. Smiley (Eds.), Studies in the Philosophy
of Logic and Knowledge (pp. 7–29). Oxford: Oxford University Press, 2005.
McDowell, J. (1986). Singular thought and the extent of inner space. In P. Petit & J. McDowell
(Eds.), Subject, Thought, and Context (pp. 137–168 ). Oxford: Clarendon Press.
McDowell, J. (2008). The Disjunctive conception of experience as material for a transcen-
dental argument. In F. Macpherson & A. Haddock (Eds.), 4Disjunctivism: Perception,
Action, Knowledge (pp. 376–389). Oxford: Oxford University Press.
Noë, A. (2004). Action in perception. Cambridge, MA: MIT Press.
Peacocke, C. (1983). Sense and Content. Oxford: Oxford University Press.
Peacocke, C. 1992. A Study of Concepts, Cambridge MA: MIT Press.
Peacocke, C. (1998). Nonconceptual content defended. Philosophy and Phenomenological
Research, 58, 381–388.
Peacocke, C. (2001). Does perception have a nonconceptual content? Journal of Philosophy,
98, 239–264.
Pinker, S. (1997). How the Mind Works. New York: W. W. Norton & Company.
Price, H. H. (1950). Perception. 2nd edn. London: Methuen.
Reid, Thomas (1983). Inquiry and Essays. Edited by Ronald Beanblossom and Keith Lehrer.
Indianapolis, IN: Hackett.
Russell, Bertrand. [1912] (1997). The Problems of Philosophy. New York: Oxford University
Press.
Schellenberg, S. (2011). Ontological minimalism about phenomenology. Philosophy and
Phenomenological Research, 83, 1–40.
Shams, L., Kamitani, Y., & Shimojo, S. (2000). What you see is what you hear. Nature, 408,
788.
Siegel, S. (2004). Indiscriminability and the phenomenal. Philosophical Studies, 120(1–3),
91–112.
Siegel, S. (2005). Which properties are represented in perception? In T. Szabo Gendler &
J. Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Siegel, S. (2010). The Contents of Visual Experience. New York: Oxford University Press.
 Introduction   35

Siegel, S. (2011).The contents of perception. In Edward N. Zalta (Ed.), The Stanford

Encyclopedia of Philosophy. Winter 2011 edn. http://plato.stanford.edu/archives/
win2011/entries/perception-contents/.
Smith, AD. (2002). The Problem of Perception. Cambridge, Mass.: Harvard University Press.
Snowdon, P. (1980/81). Perception, vision and causation. Proceedings of the Aristotelian
Society, 81, 175–192.
Sosa, E. (May 2009a). Précis of A Virtue Epistemology (Oxford University Press, 2007).
Philosophical Studies, 144(1), 107–109.
Sosa, E. (May 2009b). Replies to commentators on A Virtue Epistemology (Oxford
University Press, 2007). Philosophical Studies, 144(1), 137–147.
Sperber, D. (1994). The modularity of thought and the epidemiology of representa-
tions. In L. A. Hirschfeld & S. A. Gelman (Eds.), Mapping the Mind (pp. 39–67).
Cambridge: Cambridge University Press.
Sperber, D. (2002). In defense of massive modularity. In I. Dupoux (Ed.), Language, Brain,
and Cognitive Development (pp. 47–57). Cambridge, MA: MIT Press.
Tye, M. (2007). Intentionalism and the argument from no common content. In J. Hawthorne
(Ed.), Philosophical Perspectives 21: Philosophy of Mind (pp. 589–613).
PART ONE

Content Views
2

Empirical Problems with

Anti-Representationalism
Bence Nanay

2.1. Anti-Representationalism

Philosophers, psychologists, and cognitive scientists often talk about percep-

tual experiences, or perceptual states in general, as representations. Many of our
mental states are representational. Most of our emotions, for example, are about
something: we are afraid of a lion, fond of chocolate mousse, etc. The same goes
for beliefs, desires, and imaginings. It seems natural then to suppose that percep-
tual states are also representations: when I see a cat, my perceptual state is about
this cat: it refers to this cat. My perceptual state represents this particular as having
a number of properties and the content of my perceptual state is the sum total of
these properties (see Peacocke 1992; Nanay 2010a, 2013; Pautz 2010; Siegel 2010a,
2010b). I call this view representationalism.
Anti-representationalism is the view according to which “perception is not a
process of constructing internal representations” (Noë 2004, p. 178; see also Campbell
2002; Travis 2004; Martin 2006; Brewer 2006; Ballard 1996; O’Regan 1992). As
anti-representationalism is a negative view, which really is just the rejection of the
idea of perceptual representations, different approaches reject this idea for different
reasons and they also replace the theoretical role perceptual representations are sup-
posed to play with different alternatives. I will sort these anti-representationalist
arguments and theories into two very broad categories (acknowledging that they
themselves have many different versions): enactivism and relationalism.1

2.1.1. ENACTIVISM

The main enactivist claim is that we have all the information we need in order
to cope with our environment in the world out there. So we do not need to con-
struct representations at all and, more specifically, we do not need perceptual

1
Enactivism and relationalism often combine, see Noë (2004) and Hellie (forthcoming). 39
 Content Views
40   

representations either. As Dana Ballard put it: “The world is the repository of the
information needed to act. With respect to the observer, it is stored ‘out there’,
and by implication not represented internally in some mental state that exists
separately from the stimulus” (Ballard 1996, p. 111; see also Brooks 1991; Ramsey
2007).
In short, perception is an active and dynamic process between the agent
and the environment and this dynamic interaction doesn’t have to be (or maybe
couldn’t even be) mediated by static entities like representations (Chemero 2009;
Port & Van Gelden 1995). Another version of the positive claims that enactivism
makes is the following: when we see a scene, it is not the case that the whole scene
in all its details is coded in our perceptual system. Only small portions of it are: the
ones we are attending to. The details of the rest of the scene are not coded at all, but
they are available to us all along—we just have to look (O’Regan 1992; Noë 2004,
esp. pp. 22–24).2
The enactivist version of anti-representationalism covers a wide range of views
that differ from one another in important ways: behavior-based AI, Gibsonian
ecological psychology (Gibson 1966, 1979; Chemero 2009), embodied and dis-
tributed cognition (Hutto & Myin 2013), dynamical systems theory (Port & Van
Gelden 1995), and non-classical connectionism (Ramsey 2007), just to name a few.
I will lump them together nonetheless under the label of enactivism as the objec-
tion I will raise applies to all of them as they all share the premise that there are no
perceptual representations.

2.1.2. RELATIONALISM

The starting point of the relationalist version of anti-representationalism is that

perceptual states are not representations: they are constituted by the actual per-
ceived objects. Perception is a genuine relation between the perceiver and the
perceived object—and not between the agent and some abstract entity called ‘per-
ceptual content’ (Travis 2004; Brewer 2006; Martin 2004, 2006; but see also Byrne
& Logue 2008’s criticism).
One of the arguments in favor of this ‘relational view’ is that if we assume that
perception is representational, then we lose the intuitively plausible assumption
that the object of perception is always a particular token object. The charge is that
the representational view is committed to saying that the content of perceptual
states is something general. Although this claim may not be justified in the case of
certain versions of the representational view (ones that hold that perceptual states

2
One may be tempted to point out that these two claims, even if they are true, may only give
us reason to conclude that perceptual representations are not static or not detailed, but they give us
no reason to give up perceptual representations per se. My aim here is not to criticize the arguments
in favor of various versions of anti-representationalism, but to raise a general problem for all
anti-representationalist accounts.
 Empirical Problems with Anti-Representationalism   41

have object-involving, or maybe gappy, content), it does pose an important ques-

tion. If the content of a perceptual state is taken to be the conditions under which it
represents the world correctly (Peacocke 1992), then how can this content specify
a token object? It is likely to specify only the conditions a token object needs to
satisfy. And then any token object that satisfies these conditions would equally
qualify as the object this perceptual state represents. Suppose that I am looking at
a pillow. Replacing this pillow with another, indistinguishable, pillow would not
make a difference in the content of my perceptual state. On these two occasions
the content of my perceptual state is identical and the phenomenal character of
my perceptual state is also identical (the two pillows are indistinguishable, after
all). Thus, it seems that according to the representational view, the two perceptual
states themselves are identical. But their objects are very different (see Soteriou
2000 for a good summary on the particularity of perception and Nanay 2012b for
a representationalist answer).
The relational view, in contrast, insists that perceptual states are about some-
thing particular. Replacing the pillow with another, indistinguishable, pillow
would give rise to an entirely different (but maybe indistinguishable) perceptual
state. We have to be careful about what is meant by the identity or difference of our
perceptual states, as one clear disagreement between the relational and the repre-
sentational view is whether these two perceptual states are identical or different.
The disagreement between the representationalists and the relationalists is about
whether seeing the first pillow and seeing the second, indistinguishable, pillow are
mental processes of the same type, then this disagreement no longer seems very
clear, as there are many ways of typing mental processes. Even the relationalists
would agree that we can type these two instances of seeing in such a way that they
would both belong to the same type, say, the type of perceptual states in general.
And even the representationalists could say that there are ways of typing these two
perceptual states so that they end up belonging to different types.
It has been suggested that the real question is whether these two perceptual
states belong not just to the same type but whether they belong to “the same fun-
damental kind” (Martin 2004, pp. 39, 43). The representational view says they do;
the relational view says they don’t. Belonging to a ‘fundamental kind’ is supposed
to “tell what essentially the event or episode is” (Martin 2006, p. 361). Whether or
not we find these considerations compelling (see Byrne and Logue 2008, espe-
cially section 7.1, for a thorough analysis of the ‘fundamental kind’ version of the
relational view and Nanay 2014 for criticism), the argument from the particularity
of perception in favor of the relational view can be rephrased in the following
manner: the representationalist does not have any principled way of differentiat-
ing the two perceptual states in this example. The relationalist does.
If we dispose of the very idea of perceptual representation, we need to find
an alternative way of talking about perception. If we cannot say that my percep-
tual representation represents x as having property F, what should we say if I see
a as F? Different anti-representationalists give different answers to this question.
 Content Views
42   

The relationalists say that there is a relation between the perceiver and the token
perceived object (as well as its properties: a and F) (Campbell 2002; Martin 2006;
Brewer 2006). The enactivists use a variety of metaphors: what happens when I see
a as F is that I fixate on a’s property F (Ballard 1996). Yet another alternative would
be to say that I pick up a’s F-ness in my ambient optic array (Gibson 1966, 1979;
Chemero 2009).
Some of these positive suggestions of anti-representationalism may be more
promising than others, but I will argue that there are empirical problems with the
very idea of disposing of perceptual representations: it is inconsistent with empir-
ical findings about dorsal perception and about the multimodality of perception.
I will analyze these two problems in the next two sections.

2.2. The First Empirical Problem: Dorsal Perception

The first reason to doubt anti-representationalism is that sometimes our percep-

tual system seems to attribute two incompatible property-instances to the same
object. If we accept that there are perceptual representations, this is easy to accom-
modate: we have two perceptual representations, each representing the object as
having a property-instance. But it is unclear how the anti-representationalist can
describe these cases. Here is the most famous example.
Humans (and other mammals) have two visual subsystems that use dif-
ferent regions of our central nervous system, the ventral and dorsal streams. To
put it very simply, the ventral stream is responsible for identification and recogni-
tion, whereas the function of the dorsal stream is the visual control of our motor
actions. In normal circumstances, these two systems co-function, but if one of
them is removed or malfunctioning, the other can still function relatively well (see
Milner & Goodale 1995; Goodale & Milner 2004, for overview).
If the dorsal stream is malfunctioning, the agent can recognize the objects in
front of her, but she is incapable of manipulating them or even localizing them in
her egocentric space (especially if the perceived object is outside the agent’s fovea).
This happens if a patient is suffering optic ataxia. If the ventral stream is malfunc-
tioning, the agent can perform actions with objects in front of her relatively well,
but she is incapable of even guessing what these objects are. This happens in the
case of visual agnosia.
The philosophical implications of this physiological distinction are not at all
clear. Some argued that ventral visual processing is conscious, whereas dorsal is
unconscious (see esp. Milner & Goodale 1995; Goodale & Milner 2004), but this
view has been criticized both on empirical and on conceptual grounds (see, e.g.,
Dehaene et al. 1998; Jeannerod 1997; Jacob & Jeannerod 2003; see also Brogaard
forthcoming a and forthcoming b for summaries). It has also been suggested that
dorsal processing gives rise to nonconceptual content, whereas ventral processing
 Empirical Problems with Anti-Representationalism   43

gives rise to conceptual content (see Clark 2001 for a summary on the literature on
this). I do not need to take sides in either of these questions.
In healthy humans, the way the dorsal and the ventral stream works can come
apart in some circumstances, as in the case of the three-dimensional Ebbinghaus
illusion. The two-dimensional Ebbinghaus illusion is a simple optical illusion.
A circle that is surrounded by smaller circles looks larger than a circle of the same
size that is surrounded by larger circles. The three-dimensional Ebbinghaus illu-
sion reproduces this illusion in space: a poker chip surrounded by smaller poker
chips appears to be larger than a poker chip of the same diameter surrounded by
larger ones. The surprising finding is that although our judgment and experience
of the comparative size of these two chips is incorrect as we judge the first chip
to be larger than the second one, if we are asked to pick up one of the chips, our
grip-size is influenced by the illusion in a much smaller degree (Aglioti et al. 1995;
cf. Gilliam 1998; Franz et al. 2003; Haffenden & Goodale 1998). The usual way of
explaining this finding is that our dorsal stream is not fooled (or, more precisely,
only mildly fooled) by the illusion but our ventral stream is.
The same results can be reproduced in the case of other optical illusions. In
the Müller-Lyer illusion, while we (mistakenly) see the two lines as having different
length, our eye- and pointing movements represent them (correctly) as being the
same (Goodale & Humphrey 1998; Gentilucci et al. 1996; Daprati & Gentilucci
1997; Bruno 2001). Similarly, in the case of the ‘Kanizsa compression illusion’ and
the ‘hollow face illusion’, our perception is deceived but our action is much less so
(Bruno & Bernardis 2002; and Króliczak et al. 2006, respectively). Thus, some-
times our ventral visual subsystem attributes a different property to an object from
the one the dorsal subsystem does.
This is the representationalist way of describing the three-dimensional
Ebbinghaus case: we have two perceptual representations, a dorsal and a ventral
one, and they represent the chip as having different size properties. But what can
the anti-representationalist say? If perception is a relation between the perceiver
and the perceived token object’s properties, then we have one perceptual relation
here: the one between the perceiver and the perceived token poker chip. But then
which property of the perceived object constitutes the other one of the two relata
of this relation? The property we experience the chip as having or the one that our
grip-size seems to be tracking? These two perceptual episodes are both relations
to the very same token object: the same poker chip and the properties of this same
poker chip. And two different perceptual episodes cannot be constituted by the
very same perceptual relation.
If, on the other hand, as the enactivist says, “the world is our external mem-
ory,” then what serves as our external memory here: the property we experience
the chip as having or the one that our grip-size seems to be tracking? It is difficult
to see what would even be meant by having two different ‘worlds as our external
memory.’
 Content Views
44   

The anti-representationalist needs to choose. If she is relationalist, she needs

to choose because these two perceptual episodes are both relations to the very
same token object: the same poker chip. And two different perceptual episodes
cannot be constituted by the very same perceptual relation. And if she is enactivist,
then it is difficult to see what would even be meant by having two different ‘worlds
as our external memory.’
In short, if we endorse anti-representationalism, we need to deny that
there are two perceptual episodes in this scenario. One of them has to go. How
one of them is exiled (and which one is) depends on the specific version of
anti-representationalism. But this seems wrong: the property we experience the
chip as having plays a clear role in our perception: it justifies our beliefs and other
mental states, for example. And the property that our grip-size appears to be
tracking also plays a clear role: it guides our goal-directed action of grasping the
chip. The fine details of our bodily movements could not be explained without ap-
pealing to this property playing a role in the perceptual processing.
One way of resisting this argument would be to argue that we should only con-
sider a size-property in the Ebbinghaus case to play a role in our perception if it is
consciously experienced. Thus, only one property plays a role in our perception: the
one that we experience the chip as having. The other one is irrelevant. This seems
to be the route most relationalists would take as they very often characterize the
relation that constitutes perception as a relation between the token perceived object
and the perceiver’s experience. And this general approach also seems to be part of
at least some versions of the enactivist package (see esp. O’Regan 2011; and Hutto
& Myin 2013): when perception is taken to be the dynamic and active exploration
of the environment, this exploration is to be understood as a conscious process—in
fact, the main interest of many enactivists concerns the phenomenal character of
perception (O’Regan 2011; Noë 2004).
There are two problems with this strategy, one more serious than the other.
The less serious problem is that if the anti-representationalist claims that the only
property that plays a role in our perception is the one that we experience the chip
as having, then what should we say about the other property: the one our grip-size
is tracking? It plays an obvious and important role in guiding our goal-directed
action but the anti-representationalist is forced to say that it is not represented in
our perception—how can these two claims be made consistent? One popular way
of doing so would be to say that although our perceptual system does not represent
these properties, it carries information about them. So the information of the size
of the chip that guides my grasping movement is coded in the perceptual system,
but it is not represented. What this suggestion amounts to clearly depends on how
one interprets the concept of information-carrying. It needs to be different from
representing, but it cannot be too different as the information of the size-property
of the chip needs to be available to other parts of our brain (that would guide our
goal-directed actions). The classic concept of information-carrying (à la Dretske
1981) will not do as x carrying information about y does not imply that y is somehow
 Empirical Problems with Anti-Representationalism   45

coded in x in such a way as to make y available to other systems. The more recent
concept of information-carrying (à la Dretske 1995) will not do either as the differ-
ence between information-carrying and representation according to Dretske 1995
is supposed to be that representations have the function to carry information. But
regardless of how we interpret the concept of function in this definition (Millikan
1984; Neander 1991; Bigelow and Pargetter 1987; Walsh 1996; Nanay 2010b, 2012a),
our perceptual system does seem to have the function to carry information about
the size of the chip that would then help us to approach it with the right grip size.
But then it would follow that the perceptual system does represent this property.
In short, appeal to the distinction between information-carrying and representing
does not seem to help the anti-representationalist.
The second, even more serious problem with this anti-representationalist re-
sponse is that if one takes this route, it seems inevitable that she needs to deny that
perception can be unconscious. If what is constitutive of perception is, rather than
perceptual representations, conscious fixation or a relation between an object and
a conscious experience, then perception must be conscious by definition. But this
is a dangerous conclusion to draw: there seems to be a lot of examples of uncon-
scious perception, from visual agnosia and neglect patients to subliminal priming
and blindsight.
Some anti-representationalists will undoubtedly bite the bullet and embrace
the idea that perception is necessarily conscious, but then they have to give us a
way of analyzing those perceptually guided actions that, like the grasping of the
chip in the Ebbinghaus case or like the goal-directed actions of visual agnosia
patients, are not guided by consciously experienced properties of objects. But this
would go against the consensus in cognitive science, where it is generally assumed
that these episodes are unconscious.

2.3. The Second Empirical Problem: The Multimodality of Perception

There is a lot of recent evidence that multimodal perception is the norm and not
the exception—our sense modalities interact in a variety of ways (see Spence &
Driver 2004 for a summary and O’Callaghan 2008, forthcoming for philosoph-
ical overviews). Information in one sense modality can influence the information
processing in another sense modality at a very early stage of perceptual processing
(often in the primary visual cortex in the case of vision, e.g., see Watkins et al.
2006). A simple example for this is ventriloquism, where vision influences our
audition: we experience the voices as coming from the dummy and not from the
ventriloquist (see Bertelson 1999). But there are more surprising examples: if there
is a flash in your visual scene and you hear two beeps while the flash lasts, you ex-
perience it as two flashes (Shams et al. 2000).
What is the most important for us from this literature is that the multimodal-
ity of perception presupposes that information from two different sense modalities
 Content Views
46   

is unified in a shared framework (see, e.g., Vroomen et al. 2001; Bertelson and de
Gelder 2004). Noise coming from above and from the left and visual information
from the upper left corner of my visual field are interpreted by the perceptual sys-
tem as belonging to (or bound to) the same sensory individual (whatever that may
be). This is easy for the representationalist to analyze: vision attributes a property
to a part of the perceived scene and audition attributes a different property to the
same perceived scene. The two different sense modalities represent the same scene
as having different properties.
To put it very simply, multimodal perception seems to require matching
two representations, a visual and the auditory one. If we cannot talk about
perceptual representation, how can we talk about what is being matched? The
auditory sense modality gives us a soundscape and vision gives us a visual
scene and our perceptual system puts the two together. It is difficult to explain
this without any appeal to representations. The enactivist arsenal seems insuf-
ficient: they can appeal to the active exploration of the multimodal environ-
ment, but this is unlikely to help here: we are actively exploring the world that
is given to us in both sense modalities—but this in itself requires multimodal
integration. In short, the active exploration of the environment presupposes
multimodal integration, which, in turn, seems to presuppose representations.
Enactivists could insist that the active exploration of the environment happens
separately in each sense modality—but this is in conflict with the findings
about multimodal integration very early in perceptual processing (as early as
the primary visual cortex, see Watking et al. 2006).
The relationalist version of anti-representationalism also seems powerless as
the relation between the perceiver and the token perceived object that constitutes
perception seems to be the outcome of this process of unifying multimodal in-
formation: our experience of the perceived token object (thus, presumably, the
perceptual relation) is brought about by this unification process. The argument
from multimodality seems to show that the phenomena anti-representationalists
emphasize, be it the active and dynamic exploration of the environment or the
relation to a token object, presuppose the coordination of information in the dif-
ferent sense modalities, but this can only be accounted for in representational
terms.
The anti-representationalist has a further option: they can bite the bullet
and admit that there are sub-personal perceptual representations, but when
it comes to the personal level, there aren’t any. The relationalist version of
anti-representationalism may find this response more palatable than the enactiv-
ists, who are often explicit about not limiting their attention to personal level phe-
nomenon (see esp. Ballard 1996; Noë 2004, pp. 28–32). But John McDowell, for
example, explicitly argued that while a representationalist picture is the correct
one for the sub-personal level, we should accept Gibson’s claims with regards to
the personal level, which would make his view (at least in this respect) a version of
enactivism (McDowell 1994).
 Empirical Problems with Anti-Representationalism   47

Even if we accept the personal/sub-personal distinction as unproblematic

(see Bermúdez 2000 for some serious doubts about this), it is difficult to see how
this strategy would work. The claim of the anti-representationalists is that there
are no perceptual representations. One natural way of understanding this is that
there are no representations in the perceptual system, but this would make the
anti-representationalist claim one about the sub-personal level. There may be a
way of understanding the claim that there are no perceptual representations in
such a way that it is about the personal level, but I do not see how this could be
done without appealing to consciousness. Although it is often emphasized that the
personal/sub-personal distinction is not the same as the conscious/unconscious
distinction, it is hard to see how the personal level claim that there are no per-
ceptual representations would not amount to saying that no representations are
involved in perceptual experience.
And here the anti-representationalist’s claim boils down to the suggestion that
perceptual experience is a relation or that perceptual experience is the active explo-
ration of our environment. No doubt, some proponents of relationalism would be
perfectly happy with this claim. But this is not a claim about perception in general.
It is a claim about one specific way of perceiving: conscious perception. And, as we
have seen in the previous section, conscious perception is just one sub-category
of perception: not all perception is conscious and a theory of perception should
not be a theory of conscious perception. The anti-representationalist’s claim, un-
derstood as a claim about conscious perception, may be true, but nothing follows
from it for perception per se.

2.4. Conclusion

I argued that in spite of recent efforts to exile the concept of representation from
the discussion of perception, there are empirical reasons why we should hold onto
this concept: we are unlikely to be able to account for and explain dorsal vision and
the multimodality of perception. If this is true, then we should keep perceptual
representations as one of the most important concepts in philosophy of percep-
tion. The real question then is not whether there are perceptual representations
but what kind of representations they are.

References

Aglioti, S., DeSouza, J. F. X., & Goodale, M. A. (1995). Size-contrast illusions deceive the eye
but not the hand. Current Biology, 5, 679–685.
Ballard, D. H. (1996). On the function of visual representation. In Kathleen Akins (Ed.),
Perception (pp. 111–131). New York: Oxford University Press.
Bermúdez, Jose Luis (2000). Personal and subpersonal: A difference without a distinction.
Philosophical Explorations, 3(1), 63–82.
 Content Views
48   

Bertelson, P. (1999). Ventriloquism: A case of cross-modal perceptual grouping. In G.

Aschersleben, T. Bachmann, & J. Müsseler (Eds.), Cognitive Contributions to the
Perception of Spatial and Temporal Events (pp. 347–362). Amsterdam: Elsevier.
Bertelson, P., & de Gelder, B. (2004). The psychology of multimodal perception. In C.
Spence & J. Driver (Eds.), Crossmodal Space and Crossmodal Attention (pp. 141–177).
Oxford: Oxford University Press.
Bigelow, J., & Pargetter, R. (1987). Functions. Journal of Philosophy, 84, 181–197.
Brewer, Bill (2006). Perception and Content. European Journal of Philosophy, 14, 165–181.
Brogaard, B. (forthcoming a). Are there unconscious perceptual processes? Consciousness
and Cognition.
Brogaard, B. (forthcoming b). Conscious vision for action versus unconscious vision for
action? Journal of Philosophy.
Brooks, R. A. (1991). Intelligence without representation. Artificial Intelligence, 47, 139–159.
Bruno, Nicola (2001). When does action resist visual illusions? Trends in Cognitive Sciences,
5, 385–388.
Bruno, Nicola, & Bernardis, Paolo (2002). Dissociating perception and action in Kanizsa’s
compression illusion. Psychonomic Bulletin & Review, 9, 723–730.
Byrne, Alex, & Logue, Heather (2008). Either/or. In Adrian Haddock & Fiona MacPherson
(Eds.), Disjunctivism: Perception, Action, Knowledge (pp. 54–97). Oxford: Oxford
University Press.
Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press.
Chemero, A. (2009). Radical Embodied Cognitive Science. Cambridge, MA: MIT Press.
Clark, Andy (2001). Visual experience and motor action: Are the bonds too tight?
Philosophical Review, 110, 495–519.
Daprati, E., & Gentilucci, M. (1997). Grasping an illusion. Neuropsychologia, 35, 1577–1582.
Dehaene, S., Naccache, L., Le Clec'H, G., Koechlin, E., Mueller, M., Dehaene-Lambertz, G.,
van de Moortele, P. F., & Le Bihan, D. (1998). Imaging unconscious semantic priming.
Nature, 395, 597–600.
Dretske, Fred (1981). Knowledge and the Flow of Information. Cambridge, MA: MIT Press.
Franz, V. H., Bülthoff, H. H., & Fahle, M. (2003). Grasp effects of the Ebbinghaus illu-
sion: Obstacle avoidance is not the explanation. Experimental Brain Research, 149,
470–477.
Gentilucci, M., Cheiffe, S., Daprati, E., Saetti, M. C., & Toni, I. (1996). Visual illusion and
action. Neuropsychologia, 34, 369–376.
Gillam, Barbara (1998). Illusions at century’s end. In Julian Hochberg (Ed.), Perception and
Cognition at Century’s End (pp. 95–136). San Diego, CA: Academic Press.
Goodale, Melvyn A., & Humphrey, G. Keith (1998). The objects of action and perception.
Cognition, 67, 181–207.
Goodale, M. A., & Milner, A. D. (2004). Sights Unseen. Oxford: Oxford University Press.
Haffenden, A., & Goodale, M. A. (1998). The effect of pictorial illusion on prehension and
perception. Journal of Cognitive Neuroscience, 10, 122–136.
Hellie, B. (forthcoming). Conscious Life. Oxford: Oxford University Press.
Hutto, D., & Myin, E. (2013). Radicalizing Enactivism. Cambridge, MA: MIT Press.
Jacob, Pierre, & Jeannerod, Marc (2003). Ways of Seeing: The Scope and Limits of Visual
Cognition. Oxford: Oxford University Press.
Jeannerod, M. (1997). The Cognitive Neuroscience of Action. Oxford: Blackwell.
 Empirical Problems with Anti-Representationalism   49

Króliczak, Grzegorz, Heard, Priscilla, Goodale, Melvyn A., & Gregory, Richard L. (2006).
Dissociation of perception and action unmasked by the hollow-face illusion. Brain
Research, 1080, 9–16.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 120, 37–89.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). Oxford: Oxford University Press.
McDowell, J. (1994). The content of perceptual experience. Philosophical Quarterly, 44,
190–205.
Millikan, Ruth G. (1984). Language, Thought and Other Biological Categories. Cambridge,
MA: MIT Press.
Nanay, Bence (2010a). Attention and perceptual content. Analysis, 70, 263–270.
Nanay, Bence (2010b). A modal theory of function. Journal of Philosophy, 107, 91–109.
Nanay, Bence (2012a). Function attribution depends on the explanatory context. Journal of
Philosophy, 109, 623–627.
Nanay, Bence (2012b). Perceiving tropes. Erkenntnis, 77, 1–14.
Nanay, Bence (2013). Between Perception and Action. Oxford: Oxford University Press.
Nanay, Bence (2014). The representationalism versus relationalism debate: Explanatory
contextualism about perception. European Journal of Philosophy, in press.
Neander, Karen (1991). Functions as selected effects. Philosophy of Science, 58, 168–184.
Noë, Alva (2004). Action in Perception. Cambridge, MA: MIT Press.
O’Callaghan, C. (2008). Seeing what you hear: Crossmodal illusions and perception.
Philosophical Issues, 18, 316–338.
O’Callaghan, C. (forthcoming). Perception and multimodality. In E. Margolis, R. Samuels, &
S. Stich (Eds.), Oxford Handbook to Philosophy and Cognitive Science. Oxford: Oxford
University Press.
O’Regan, K. (1992). Solving the “real” mysteries of visual perception: The world as an out-
side memory. Canadian Journal of Psychology, 46, 461–488.
O’Regan, K. (2011). Why Red Doesn’t Sound Like a Bell. New York: Oxford University Press.
Pautz, Adam (2010). An argument for the intentional view of visual experience. In Bence
Nanay (Ed.), Perceiving the World (pp. 254–309). Oxford: Oxford University Press.
Peacocke, Christopher (1992). A Study of Concepts. Cambridge, MA: MIT Press.
Port, R., & van Gelder, T. J. (Eds.) (1995). Mind as Motion: Explorations in the Dynamics of
Cognition. Cambridge, MA: MIT Press.
Ramsey, W. M. (2007). Representation Reconsidered. Cambridge: Cambridge University
Press.
Shams, L., Kamitani, Y., & Shimojo, S. (2000). What you see is what you hear. Nature, 408,
788.
Siegel, Susanna (2010a). Do visual experiences have contents? In Bence Nanay (Ed.),
Perceiving the World (pp. 333–368). Oxford: Oxford University Press.
Siegel, Susanna (2010b). The Contents of Visual Experience. New York: Oxford University
Press.
Soteriou, M. (2000). The particularity of visual perception. European Journal of Philosophy,
8, 173–189.
Spence, C., & Driver, J. (Eds.) (2004). Crossmodal Space and Crossmodal Attention.
Oxford: Oxford University Press.
Travis, C. (2004). The silence of the senses. Mind, 113, 57–94.
 Content Views
50   

Vroomen, J., Bertelson, P., & de Gelder, B. (2001). Auditory-visual spatial interac-
tions: Automatic versus intentional components. In B. de Gelder, E. de Haan, & C.
Heywood (Eds.), Out of Mind (pp. 140–150). Oxford: Oxford University Press.
Walsh, D. M. (1996). Fitness and function. British Journal for the Philosophy of Science, 47,
553–574.
Watkins, S., Shams, L., Tanaka, S., Haynes, J. D., & Rees, G. (2006). Sound alters activity in
human V1 in association with illusory visual perception. NeuroImage, 31, 1247–1256.
3

Affordances and the Contents of Perception*

Susanna Siegel

A hole in the ground protects some creatures but endangers others. Dry ground
is passable by creatures who walk, but fatal for a fish. These environments provide
different possibilities for different creatures.
J. J. Gibson invented the word “affordance” to denote possibilities of action
for a creature that are given by the environment.1 He proposed that we perceive
affordances, and that the paradigmatic perceptions are byproducts of action plans.
These proposals inspired an “action-first” approach to visual perception, which
foregrounds the role of the perceiver as an actor.
The action-first approach to visual perception can be contrasted with the
“spectator-first” approach, which foregrounds the role of the perceiver as an
observer. This approach is heir to David Marr’s computational theory of vision,
and like Marr’s theory, it gives a central role in perception to belief-like repre-
sentations. Here the paradigmatic perceptions are observations of scenes with
which one does not necessarily interact, such as watching a sunset.
In recent years, both of these approaches have been used to investigate the
nature of perceptual experience, leading to a divide over the centrality of repre-
sentation in analyzing perception.2 On the surface, the two approaches are easily
reconciled by the hypothesis that affordances are on par with color and shape as
properties represented in experience.3 But even if affordances could in principle
be represented in experience, it is reasonable to ask whether they have to be so

For extensive discussion and criticism, thanks to audiences at Harvard, Rice, Geneva, Charleston,
Madison, Oslo, Glasgow, Copenhagen, Princeton, and Universidad Autónoma de México, as well as to
John Bengson, Ned Block, Berit Brogaard, Alex Byrne, Jeremy Dolan, Anya Farennikova, Grace Helton,
Sheridan Hough, Zoe Jenkin, Sean Kelly, Fiona Macpherson, Eric Mandelbaum, Farid Masrour, Laura
Perez, Álvaro Peláez, Sebastian Rödl, and Charles Siewert, and to Miguel-Angel Sebastian and Anna
Bergqvist for writing comments on several drafts. Thanks most of all to Sebastian Watzl for many
illuminating conversations about every aspect of this material.
1
Gibson 1977.
2
Proponents of the action-first approach include Hurley (1998), Noë (2006), Kelly (2006, 2010),
Orlandi (forthcoming). Proponents of spectator-first approach include Byrne (2009), Chalmers (2005),
Pautz (2010), Peacocke (1995), Siegel (2010).
3
For developments of this idea, see Bengson, n.d. and Nanay (2011). 51
 Content Views
52   

represented—or whether instead we simply experience affordances without rep-

resenting them. If any representations of affordances would be an idle wheel in
explaining the function and character of perceptual experience, that would go
some way toward vindicating the action-first approach. And, if most properties
presented in experience could be shown to involve affordances, that would suggest
that representation in general is an idle wheel in perception.4
In this chapter, I draw out a challenge to the centrality of representation in
perceptual experience that arises from an important class of experiences of affor-
dances. These are experiences of the environment as compelling you to act in a
way that is solicited or afforded by the environment. I call such experiences expe-
rienced mandates. They are generally structured by how you are already acting in
a situation—not only by how you can act or are disposed to act in it. From your
point of view, the environment pulls actions out of you directly, like a force moving
a situation, with your actions in it, from one moment to the next. Experiences like
these were discussed by Merleau-Ponty in the 1960s, and they figure prominently
in recent discussions by Hubert Dreyfus, Adrian Cussins, and Sean Kelly, among
others.5 These experiences help us make sense of the idea that affordances could be
experienced without being represented: the environment invites or solicits an ac-
tivity, and you experience these affordances by doing, and by feeling moved to do,
what they invite you to do. From these experiences we can reconstruct a challenge
to the primacy of representation in perception.
We can use experienced mandates to ask three questions about the role of
representation in perception.
Q1. Do experienced mandates have any representational contents?
Q2. Do experienced mandates have any representational contents relevant
to explaining why it makes sense to the subject to perform the action
she experiences as mandated?
Q3. Is there any proposition that could reflect the mandate that the subject
experiences, if that proposition were the content of the experience?

I’m going to assume that experienced mandates have perceptual experiences as a

part, and that an experienced mandate counts as having representational contents
if its component perceptual experience does. Questions Q1, Q2, and Q3 then bear
directly on the role of representation in perceptual experience.
Regarding Q1, if experienced mandates have no representational content, then
their component perceptual experiences do not either, and this conclusion would
challenge the primacy of representation in perceptual experience. Regarding Q2,
if experienced mandates have no contents relevant to explaining the associated

4
Kelly (2006) suggests that experiences of color and shape are always experienced mandates, in
which colors and shapes solicit us to view them in certain ways, and that none of these experiences
involve representation.
5
Merleau-Ponty (1962); Kelly (2005); Cussins [1990] (2003); Dreyfus (2002), (2005).
 Affordances and the Contents of Perception   53

actions, then no component perceptual experiences do either, and this conclusion

would challenge the explanatory relevance of any representational contents per-
ceptual experiences may have. Regarding Q3, if we could identify a proposition
that would characterize the mandate that the subject experiences, then representa-
tional contents could in principle help to illuminate this distinctive phenomenon.
To assess these questions, I explain more fully what experienced mandates are
and what challenges they raise in section 3.1. I address all three questions in sec-
tion 3.2, where I argue that the answer to all of them is Yes.
Once these questions are on the table, it is natural to wonder whether mandates
are represented in certain perceptual experiences—not just whether any represen-
tational contents could in principle reflect the feeling of an action being solicited
and mandated by an object or situation. Since the stronger thesis that mandates are
represented in experienced mandates entails positive answers to all three questions,
one could reach the conclusions I defend here in one fell swoop by defending the
stronger thesis directly. But it is useful to consider questions Q1‒Q3 on their own.
By proceeding this way, we can distinguish more easily between challenges to the
primacy of representation in perception that arise at different levels of generality,
and we can examine the role of representation in experienced mandates more fully.
In addition, my route to identifying contents that could in principle reflect man-
dates suggests a strategy for defending the stronger claim. Without offering a full
defense of the claim, I’ll outline the argumentative strategy at the end.

3.1. Experienced Mandates

3.1.1. AFFORDANCES AND EXPERIENCED MANDATES

Experienced mandates are a kind of experience of a type of affordance. We can

characterize them more fully, by locating them in relation to three kinds of affor-
dances that are good candidates for being experienced.
Proto-affordances are possibilities unrelated to agency, either because they are
possibilities for objects that lack agency, or they are possibilities to which a sub-
ject’s agency is irrelevant. Suppose you see a ball with its edge resting on the stalk
of a plant on a hill, a rock teetering on the edge of a cliff, and a path with two people
walking toward each other. The proto-affordances here include the rollability of
the ball down the hill, the possibility that the teetering rock could fall off the cliff,
and the fact that the two hikers could pass one another without stopping or col-
liding. If a subject perceptually experienced them, these proto-affordances would
characterize how the ball, the rock, the path, and the pedestrians look to that sub-
ject. One might associate various actions with the proto-affordances, such as free-
ing the ball, tipping over the rock, or moving aside to let an oncoming person pass.
But in principle, a subject need not experience these actions as possible, simply in
virtue of experiencing the proto-affordances that prompt the associations.
 Content Views
54   

Unlike proto-affordances, affordances proper involve possibilities of action

for a creature. Suppose you see a bed that you could plop down on to rest—it is not
flimsy cardboard, or full of nails underneath, or a cavernous pit in disguise. If you
experience this affordance, then you would experience the bed as a place to stop
and rest. If the bed really was a cavernous pit in disguise, then your experience
would be falsidical. Similarly, consider a tuft of hair that covers your interlocutor’s
left eye but could be moved aside, or a thick forest that would shelter you from
heavy wind and rain. Here, the forest affords protection, and the tuft of hair affords
being moved out of the way to provide better eye contact. If you experienced these
affordances, you would experience the hair as covering the eye like a curtain that
you could open to improve your eye contact, and the forest as a place that could
shelter you. If the forest turned out to harbor monsters, or the interlocutor had
only one eye, then these experiences would be falsidical.
The experiences of affordances I’ve described so far are fully characterized
even if the subject does not feel invited, solicited, or prompted to enter the forest,
move the hair, or plop down on the bed. We can say that an affordance by X (hair,
forest, bed) of phi-ability is a non-soliciting affordance for a subject S, if X is expe-
rienced by S as affording phi-ability, but X is not experienced as soliciting, inviting,
or otherwise prompting S herself to phi. So if X’s affordance is non-soliciting for S,
it is experienced by S, and it is not experienced by S as soliciting.
In contrast, suppose that in the midst of an important conversation the tuft of
hair keeps falling over your interlocutor’s eye, obstructing proper communication
by interfering with eye contact. You might well experience the hair as an obstacle
that should be moved away to allow for fuller eye contact. Or suppose you see a
perfectly moist, frosted piece of chocolate cake resting on a plate with a fork on
a napkin next to it. You might well feel solicited by the cake to eat it. Similarly,
you might feel invited by the forest to enter it if you need shelter. We can say that
an affordance by X (hair, forest, cake) of phi-ability is a soliciting affordance for a
subject S, if it is experienced by S as soliciting, inviting, or otherwise prompting S
herself to phi. 6 So if X’s affordance solicits a subject S, then X is experienced by S
as soliciting.7
One and the same thing could be experienced as affording different things on
different occasions. A forest might be experienced as a shelter if you need protec-
tion, but as an obstacle to be circumvented if it stands between point A where you
are and point B where you’re going. Similarly, the forest might be experienced as

6
We can also be moved to action by unconscious perception of things. In a broader sense than the
one used here, those perceptions would be perceptions of soliciting affordances as well.
7
I’ve called these affordances “soliciting affordances” and “non-soliciting affordances,” and it is
natural to call the corresponding experiences “soliciting experiences” and “non-soliciting experiences.”
But please disregard any suggestion by these locutions that the affordances or the experiences are (or
are felt as) the things that do the soliciting. According to a soliciting experience, the bearer of the
affordance solicits you to phi, and according to a non-soliciting experience, something affords phi-ing.
 Affordances and the Contents of Perception   55

soliciting entrance on one occasion but not soliciting entrance on another, even if
it is experienced as proto-affording shelter both times.
Among soliciting experiences, we could distinguish between increments of
felt solicitation. A piece of cake might look perfectly positioned to be eaten, but
not so appealing that you experience it as something that is calling out to be eaten.
The hair over the eye might look easily movable, but you might have reconciled
yourself to the hairstyle, so that you don’t experience the hair as calling loudly to
be moved aside. These are differences in increments of felt solicitation.
Experienced mandates belong to the category of soliciting experiences and
have a high degree of felt solicitation. With the ball coming toward you in a tennis
game, the felt solicitation to swing your racket and hit it might be so strong that no
other option enters your mind.
But what exactly is it to feel solicited to do something (to phi) by a slice of
cake, a forest, or a tennis ball? In particular, how does this feeling relate to a moti-
vation or urge to phi? Conceptually, solicitation in the sense used here and moti-
vation can come apart. Suppose you hear music that (you can tell) is designed to be
danced to, but you feel completely unmoved by it to dance.8 The music, as you hear
it, is telling you to dance, but you don’t feel its pull. It is as if the music is trying to
make you dance, but you are not cooperating and have no inclination to cooperate.
Similarly, the neatly stacked colorful packs of candy thoughtfully placed at eye
level in the grocery store check-out line are supposed to make us disposed to buy
them. The exhortative tone of many advertisements on radio and television has an
analog in visual displays, and we can feel that they are designed to propel us into
purchasing the things advertised, even when we feel not at all moved to buy the
allegedly indispensable item. It might seem natural to call any experience of ex-
hortation an experienced mandate. If seeing a piece of cake moves you to want to
eat it when you weren’t even hungry to begin with, you are mobilized from scratch
in just the way that advertisers dream of. But advertisers’ coercive intentions to
generate desires are often detectable, even when the coercion does not succeed. I
reserve the label “experienced mandate” for soliciting experiences involving some
increment of motivation to do what is solicited.
The phenomenal aspect of experienced mandates whereby they are motivated
constitutes what I’ll call a feeling of answerability. This feeling also normally results
from hearing one’s name called or from meeting another person’s gaze. (Given
the developmental importance of joint attention, we should expect special sensi-
tivity to the direction of other people’s gaze.) We say that a person answers to one
name but not to another (e.g., “Julia” but not “Julie”). To “feel answerable to ‘Julia’ ”
means not that one can answer to “Julia,” but that one does answer to Julia.
Like answerability itself, the feeling of answerability is structurally similar to
responsibility: you can shirk a responsibility, but you won’t thereby cease to be

8
I thank Farid Masrour for this example.
 Content Views
56   

responsible. Similarly, so long you feel answerable to something, even ignoring it

is a response. But the feeling of answerability falls far short of answerability itself.
If you are answerable to someone else, let us suppose, you take them to be a source
of normative constraint on you. In feeling answerable to the dance music’s solici-
tation to dance, or to the cake’s solicitation to eat it, you do not thereby experience
the music or the cake as a source of normative constraint on you.
We can zero in further on the feeling of answerability by examining delu-
sions of reference.9 Consider the difference between the experience of someone
looking at you when you’re in the same room, and the experience of someone
on television looking at you by looking at the camera. In both cases, you feel
looked at. But there is also a difference that is brought into focus by the deluded
subjects who assimilate both cases to the live case. The deluded subject feels
that the anchorman is addressing her—not just whoever might be listening.
What do the deluded subjects feel, when they feel addressed by the anchorman
on TV? They feel a need to negotiate social space. The negotiation involves un-
avoidable response, either in answering directly or ignoring. Ignoring someone
looking at you at least initially often feels different from being unaware of them of
looking at you. The deluded subject feels moved by the anchorman in the way that
others normally feel moved only by other people who are in the same immediate
surroundings.
When the cake merely solicits you without leaving you feeling motivated
to eat it, the cake looks as if it is to-be-eaten. The cake is like the anchorman,
talking to you by talking to whoever is listening. In contrast, when the cake
solicits you in a way that leaves you feeling compelled to eat it, the cake may
again look as if it is to-be-eaten. But in addition, the cake is like a person looking
at you. You feel answerable to it, the way the deluded subject feels answerable
to the anchorman, or normals feel answerable to other people who meet their
gaze.
We can contrast experienced mandates with another superficially similar but
ultimately distinct phenomenon. Suppose you are flying in an airplane and sud-
denly realize that you should have already cancelled the electricity service at your
previous address and must do it as soon as possible. It can seem natural to describe
your conscious state at this moment as “experiencing a mandate to cancel your
electricity service.” You may feel poised and ready to make the requisite phone call,
just as soon as the plane lands. But to the extent that your immediate surroundings
are not ones in which you can or will cancel your electricity service—you don’t
have a phone, it wouldn’t work anyway in flight, and for these reasons you don’t
intend to call until later on—this type of experience is not an experienced mandate
in our sense. Whatever feeling of “mandate” you may have in this situation is not
integrated into the situation you’re currently in. It has no dynamic that unfolds

9
Bortolotti 2013.
 Affordances and the Contents of Perception   57

in that situation and does not absorb your perceptual attention.10 Whereas with
the dance music you were solicited to do something but not motivated to do it,
here you are motivated to something, without anything in your immediate envi-
ronment soliciting you to do it. Experienced mandates involve affordances by the
perceived environment that you’re currently in. In experienced mandates, solicita-
tion and motivation go together.
A final observation about experienced mandates is that the thing that one
experiences as soliciting can either be localized in an object (cake, hair), or it can be
an entire situation. Suppose you are alone on a narrow sidewalk, walking. The side-
walk affords following its path. Then someone turns a corner and begins walking
toward you. They are still far off, and no one else is in between. In this completely
ordinary situation, without having to think about it all, you assume that the person
is going to continue walking toward you until you pass. And now the affordance
of travel is more complicated. When your paths cross, the part of the sidewalk tra-
versed by them will not be passable. But you don’t yet know exactly which part
this will be. The space has to be negotiated by adjusting your relative positions.
Traversing the sidewalk’s path is afforded, but it is not afforded simply by the side-
walk. It is afforded by the sidewalk together with the passerby, contingent on their
cooperation. It remains to be seen what form the cooperation will take. The same
possibilities are open to both of you: step aside to let the other pass, or continue on
whatever path they open up for you? Make clear gestures designed to acknowledge
the other person, or play down the fact that there’s any interaction? Acknowledge
any adjustment they make, or just carry on? Besides these possible modes of full
cooperation, there is also the possibility of grudging cooperation, borderline
non-cooperation (barely move out of the way), or at an extreme, collision.11
So experienced mandates are motivated, soliciting experiences of affordances
of things in the immediate environment, including entire situations. These are the
experiences that our questions Q1‒Q3 are about.
We can distinguish between three temporal relationships an experience can
bear to the mandated action that it presents. First, one might experience as man-
datory an action not yet undertaken. Second, one might experience as mandatory
an action now being completed as mandated. Third, one might retrospectively
experience as having been mandated an action just completed.

10
An experience like the one described could unfold in the subject’s immediate environment,
if the subject was in an agitated state in which everything in the plane came to look like an obstacle
to making the crucial phone call. Perhaps such a subject would experience their surroundings in the
plane as anti-affording cancelling their electricity service. (In principle, though probably not in fact,
a subject’s agitation could even make oblong items start to look to them like phones). But to create an
illuminating foil for experienced mandates, in the plane example I’m assuming that the subject is not
even experiencing the environment as either affording or anti-affording the task of cancelling their
electricity service.
11
These possibilities are made vivid in T. G. Seuss’ story The Zax, in which a north-going Zax and
a south-going Zax are stopped in their tracks because neither will move out of the other’s way.
 Content Views
58   

These relations form a dynamic phenomenon, which can be described as

sequence of experiences. I’ll describe them in terms of an action A:
(i) an experience of: A being such that you must undertake it, because the
immediate situation demands it.
(ii) an experience of: being about to undertake A, because the immediate
situation demands it.
(iii) an experience of: now carrying out A, because the immediate situation
demanded it.

In experience (i), you experience the mandate without executing the mandated
action right then and there—before executing it, or without ever executing it.12 In
principle, a situation could present the same affordance as mandated, first in way
(i) and then subsequently in ways (ii) and (iii). But we shouldn’t assimilate (i) to
an anticipation of executing the action. Just as you might receive a mandate from
someone else that you go on to ignore, so too you might experience a potential
action (such as moving the hair aside) as mandatory in way (i), without going on
to execute it, though possibly only at the cost of some dissonance.
I’m going to use “experienced mandates” to denote dynamic experiences
spanning types (i)‒(iii). But in discussing how experienced mandates relate to rep-
resentation, I’ll focus mainly on experiences of type (i).

3.1.2. HOW EXPERIENCED MANDATES CHALLENGE THE CENTRALITY OF

REPRESENTATION IN PERCEPTION

Due to their combination of solicitation and motivation, experienced mandates

are better candidates for showing representation in perceptual experience to be
explanatorily dispensable, compared to experiences of non-soliciting affordances,
or experiences of proto-affordances, or experiences that don’t involve affordances
at all.
In an experienced mandate, the current situation seems to determine the sub-
sequent one. Since the subsequent situation seems already to be on the way when
one has the current experience, one might think there is no need for the current
experience to represent the possibility of acting in the way the situation mandates,
and more generally, no need for experiential representations to guide the subject
in planning and executing the mandated action. With both solicitation and moti-
vation built in to experienced mandates, it might seem natural to assimilate these
experiences to action that is fueled by dynamic interaction with the environment.

12
Kelly (2006) suggests that experiences characterized in (i) are typical in perceiving shape and
color constancies. In perceiving the color or shape of something, he argues, we are sensitive to an
optimal point from which it could be viewed. He does not claim, however, that every experience of
perceptual constancies is one in which we actually optimize our bodily position vis-à-vis the thing
we’re seeing.
 Affordances and the Contents of Perception   59

We find something like this challenge in Dreyfus’ description of experienced

mandates. He does not discuss such experiences under that description, but he
uses the metaphor of solicitation to describe them. At the start of one of his many
discussions of skillful action, Dreyfus connects the central behaviorist idea that
representations are not needed to mediate between the environment and behavior
to Merleau-Ponty’s notion of the “intentional arc”:
Existential phenomenologists hold that the two most basic forms of intelli-
gent behavior, learning, and skillful action, can be described and explained
without recourse to mind or brain representations. . . . The intentional arc
names the tight connection between the agent and the world, viz. that, as
the agent acquires skills, these skills are “stored”, not as representations in the
mind, but as more and more refined dispositions to respond to the solici-
tations of more and more refined perceptions of the current situation.
Maximum grip names the body’s tendency to respond to these solicitations
in such a way as to bring the current situation closer to the agent’s sense of an
optimal gestalt. I will argue that neither of these abilities requires mental or
brain representations. (2002, p. 1)
(I’ve highlighted the part about solicitations). In further explicating the kind of
solicitation involved in Merleau-Ponty’s notion of the intentional arc, Dreyfus ap-
peals to the idea that skilled action is guided by a feeling of “tension” that tracks
the status of the situation relative to one’s goal.13 Dreyfus suggests that when a
feeling of tension is sensitive to one’s goal, it might obviate any need for repre-
sentations to guide one’s responses to the environment. When movement toward
‘maximum grip’ takes place, according to Dreyfus,
. . . acting is experienced as a steady flow of skillful activity in response to one’s
sense of the situation. Part of that experience is a sense that when one’s situ-
ation deviates from some optimal body-environment relationship, one’s ac-
tivity takes one closer to that optimum and thereby relieves the “tension” of
the deviation. One does not need to know what that optimum is. One’s body is
simply solicited by the situation to get into equilibrium with it. (2002, p. 378)
Dreyfus illustrates further by discussing tennis:
[C]‌onsider a tennis swing. If one is a beginner or is off one’s form one
might find oneself making an effort to keep one’s eye on the ball, keep
the racket perpendicular to the court, hit the ball squarely, etc. But if one
is expert at the game, things are going well, and one is absorbed in the

13
The notion of tension comes from Merleau-Ponty, who uses it in Phenomenology of Perception
to describe the phenomenon of perceptual constancies. He writes “The distance from me to the object
is not a size which increases or decreases, but a tension which fluctuates around a norm.” Quoted in
Kelly 2010.
 Content Views
60   

game, what one experiences is more like one’s arm going up and its being
drawn to the appropriate position, the racket forming the optimal angle
with the court—an angle one need not even be aware of—all this so as
to complete the gestalt made up of the court, one’s running opponent,
and the oncoming ball. One feels that one’s comportment was caused by
the perceived conditions in such a way as to reduce a sense of deviation
from some satisfactory gestalt. But that final gestalt need not be repre-
sented in one’s mind. Indeed, it is not something one could represent. One
only senses when one is getting closer or further away from the optimum.
(2002, p. 379)
Now, much of what we do is in a broad sense skilled action—a fact easily observed
by watching toddlers in the midst of learning how to open doors or put on their
shoes. In contrast, Dreyfus focuses on specialized skilled action (tennis), rather
than on sensory-motor habits that nearly everyone eventually develops (tying
shoes, opening doors). But experienced mandates can be found in the domain of
sensory-motor habits as well. In the passerby example, one can imagine feeling
that stepping aside to let the other person pass is mandatory, in the sense that
among the possible modes of passing another person on the path, moving aside
to let the other person pass is the only one experienced as possible, and in that
sense, mandatory. The type of mandate can vary with the situation. Perhaps the
passerby is frail, or moves only with difficulty, and the felt mandate stems from
moral sensitivity.14 Or perhaps one has cultivated a habit of always letting the
other person pass, out of politeness, or because one enjoys determining how
such micro-interactions with the public unfold. Here too, the mandatory aspect
stems not from specialized motor skill, but from a broadly social sensitivity. In
yet other cases the felt mandate might be broadly aesthetic, as it is in the example
involving the tuft of your interlocutor’s hair that falls just in front of their left
eye, making it harder to read their expression, and producing in you a strong
impulse to move the hair out of the way. Or in an exhausted state, a fluffy bed in
an empty room might be experienced as inviting you to plop down on it for rest.
The felt mandate does not come from specialized skilled action or its dynamics
of execution in any of these examples.
The challenge posed by experienced mandates to the primacy of represen-
tation is potentially quite powerful. Experienced mandates pervade much of our
conscious lives, arising both in habitual action and in specialized skilled action.
In light of the broad challenge, let us ask: what is the relationship between expe-
rienced mandates and representation?

14
In an excellent discussion of similar phenomena, Bengson n.d., following Mandelbaum (1969)
discusses an example of this sort, where someone gives up their seat on the bus to someone else who is
visibly tired. A similar example occurs in Murdoch (1958).
 Affordances and the Contents of Perception   61

3.2. Experienced Mandates and Representation

3.2.1. DO EXPERIENCED MANDATES HAVE CONTENTS?

We can begin our inquiry into this relationship by asking whether experienced man-
dates are experiences with accuracy conditions, where the accuracy conditions char-
acterize how the environment seems to the subject to be. If an experience has accuracy
conditions, then it is accurate only if those conditions are satisfied. When questions
Q1‒Q3 ask about the role of representational contents in experienced mandates, they
are asking about the kind of accuracy condition that would characterize the experi-
ence from the subject’s point of view. Discerning just which accuracy conditions do
that is no small task.15 But some candidates naturally suggest themselves. For instance,
if an experience of a tennis ball hurtling toward you has an accuracy condition, it
might include the condition that the ball is green and coming toward you. A prelimi-
nary question, however, is whether experiences have any representational contents at
all. And a version of that question concerns experienced mandates:
Q1. Do experienced mandates have any representational contents?
So far, I have glossed over the exact relationship that experienced mandates bear
to perceptual experience, leaving open that they might simply be perceptual expe-
riences, or alternatively they might have perceptual experiences as components. It
seems obvious that perceptual experiences are related to experienced mandates in
one of these two ways. You couldn’t very well play tennis without seeing the ball, or
seek protection by entering a forest without perceiving where the forest is in relation
to you.
The central observation favoring the idea that perceptual experiences in expe-
rienced mandates are contentless is that the subject feels immediately solicited by
their environment in ways that move them to action. There might then seem to be
no explanatory role for contents of experience to play.
In reply, the fact that while acting easily in an environment, one is seemingly
propelled by it does not undermine the general considerations about perception
that suggest that such experiences have contents. Consider the case of visual expe-
riences, since all of our examples involve vision. According to the Content View,
all visual perceptual experiences have contents. The central motivation for the
Content View is phenomenological. When you see things, they look to you to be a
certain way. And when they look to you to be a certain way, they look to have cer-
tain properties. I won’t repeat a full defense (given elsewhere) of the Content View
here, but the key transition in that defense moves from ‘X looks to have property
F’ to ‘The experience of X’s having F is accurate only if X has F’.16

15
We can’t read off directly from introspection which contents experiences have. For discussion,
see Siegel 2010, ch. 4.
16
Siegel 2010, ch. 2.
 Content Views
62   

On the starting point of the transition, the fact that Dreyfus’ tennis player can
so easily navigate the tennis court does not entail that the things the player sees on
the court—the player, the oncoming ball, the spaces between himself and net—fail
to look any way to him at all. Without seeing the tennis court, it would be hard to
play tennis and hard to perceive the gestalt to be completed. As Dreyfus observes, in
the experience of playing tennis, there are some “perceived conditions” in response
to which one adjusts one’s movement. Since the adjustments are made to “complete
the gestalt” of the tennis game, these perceived conditions presumably include the
components of the gestalt, such as the positions of the opponent and the ball. Which
ways these things look will depend on many factors, including what you’re doing—
such as whether you’re playing the game, watching the players, or studying the court
in order to draw it. But if the court, or the things in it, or the situation on the court
didn’t look any way to you at all, then you would not be seeing them. (Even if you
are hallucinating rather than seeing, the same basic phenomenological point holds.)
In all of these cases, properties characterize the way things look to us, when we
see them. And if things look to have certain properties, then, it seems, the experience
is accurate, only if things have the properties that they look to have. Of course, posi-
tive reason is needed to think that in general, when things you see look to you to have
certain properties, the experience is accurate only if the things are the way they look.
A full defense of this transition is the core of my case for the Content View.
What’s important here is that nothing specific to experienced mandates forces any
departure from the starting point of this general argument.
Although it may seem obvious from our examples so far that experienced man-
dates involve perceptual experiences, different examples might call this assumption
into question. And if there are experienced mandates that extinguish all percep-
tual experiences as they unfold, then trivially, no contentful perceptual experiences
would play any important role analyzing experienced mandates. That result would
open the possibility of denying that experienced mandates have representational
contents, challenging the significance of the Content View, while tolerating its truth.
One might hypothesize that the extinction of perceptual experiences is a spe-
cial case of an ordinary occurrence, in which we exercise skills or habits without
much guidance, if any, from perceptual experience. So it is worth considering
whether some experienced mandates plausibly extinguish perceptual experiences
we may start out having in earlier stages of habit or skill-formation, but which
fade out completely by the time the habit or skill is well-established. Perhaps in
some such cases, we form beliefs about what’s around us, without basing those
beliefs on any experience. One often doesn’t need to look carefully, or at all, to see
where to reach for a familiar doorknob, because one’s body ‘knows’ already, out
of sensory-motor habit, or thanks to unconscious visual processing in the dorsal
stream.17

17
Milner and Goodale 1995.
 Affordances and the Contents of Perception   63

Could we be solicited by the environment without perceptual experience?

Let us grant for the sake of argument that we could. Perhaps actions such as slip-
ping a tennis racket back into its case are sometimes completed without much
need for perceptual experience at all. The same might be true of belief: one might
believe that the racket is back in the case, without any conscious experience or
memory of putting it there or seeing it slip in. It is uncontroversial that such
actions proceed without deliberation. Might they proceed without experience as
well? If experienced mandates can extinguish perceptual experiences, then even
if all such experiences have content (as per the Content View), the Content View
would not illuminate the role of perception in action or belief in cases of experi-
enced mandates.
But there is little reason to think that experienced mandates extinguish percep-
tual experiences. It is one thing for motor habits to carry you through the actions
of putting away the racket, or stepping aside so that the oncoming person can pass
you on the narrow sidewalk, so that these aspects of your perceptual experience
become highly inattentive and inaccessible to memory. It is something else for
sensory-motor habits to blip out all surrounding perceptual experience, or for step-
ping aside to prevent you from consciously seeing the oncoming passer-by and the
rest of the scene at all. If such occurrences were normal, our conscious lives would
be interrupted with waking but blank durations, like seizures sprinkled throughout
the day, triggered by habitual actions like putting away a tennis racket, filling up
one’s tea kettle, or opening the mailbox. The habit-discontinuity thesis (HD) is that
such discontinuities occur on a regular basis with habitual actions.
The HD thesis predicts that we could never take in novel stimuli at the level
of experience or notice anything unusual while completing habitual actions—
actions that presumably don’t use up much attention. Since habitual actions
would seem to free up attention rather than expending it, this prediction is
likely to be false, if the usual moral drawn from inattentional blindness experi-
ments are correct. The usual moral is that our capacity for experience, or our ca-
pacity to remember it, is reduced when we undertake visual tasks that demand
a lot of attentional resources, not by tasks that are relatively undemanding, such
as habitual actions.
The HD thesis also predicts that most of the time when we reflect afterward on
whether anything was visible to us while we were completing such actions, we would
find that our memories were blank. It seems plain that this prediction is wrong. It’s a fa-
miliar occurrence that we complete a habitual action, realize afterward we were paying
little attention to what we were doing, and yet can still remember how other parts of
the scene looked as we were completing it. You might not realize that you were sliding
the tennis racket back into its case, or adjusting your position on a path to let other pass
more easily, yet plainly these inattentive actions are sometimes accompanied by your
noticing the sunset or hearing that the passer-bys were speaking German.
So far, I’ve argued that experienced mandates provide no special reason to
back away from the standard reasons to think that perceptual experiences have
 Content Views
64   

contents. Against the assumption that perceptual experiences have contents, do

any of these contents pertain especially to the felt mandate? This question is sharp-
ened in the form of Q2.
Q2. Do experienced mandates have any representational contents relevant to
explaining why it makes sense to the subject to perform the action she experi-
ences as mandated?
Let’s say that a content of an experienced mandate is explanatorily relevant, if it is rel-
evant to explaining why it makes sense to the subject to perform the action that she
experiences as mandated. For instance, in the hair case, if the subject moves the hair
out of the way, then the contents “hair is in front of eyes” and “hair is to-be-moved”
would be explanatorily relevant. This notion of explanatory relevance is considerably
loose, but we won’t need more precision in order to address Q2.
We can distinguish between two kinds of properties that the component per-
ceptual experiences might represent, and then consider whether contents involv-
ing each kind of property are explanatorily relevant.
First, the content could involve properties that rationalize the mandated action.
This notion is best introduced through examples. When the interlocutor’s hair
looks to be covering one eye, it mandates adjustment. When the bed looks fluffy,
it mandates plopping down on it. When the forest looks canopied, it mandates en-
tering the forest. The ways the hair, the bed, and the forest look in these examples
(covering one eye, fluffy, canopied) are rationalizing properties, in the sense that
the fact that those things have those properties makes the action involved in the
mandate straightforwardly intelligible. Adjusting the hair is intelligible if the hair
covers an eye, plopping down on the bed is intelligible if it is fluffy, seeking protec-
tion in the forest is intelligible if the forest is canopied. If rationalizing properties
figured in the accuracy conditions of these examples, then experiences would be
accurate only if: the hair covers the eye, the bed is fluffy, the forest is canopied.
Second, the content could involve properties that don’t rationalize the man-
dated action. For instance, the ball is green, the hair is blonde, the forest trees are
swaying in the wind. In these examples, the properties green, blonde, and swaying
are non-rationalizing. In an experimental setting, one might operationalize the
notion of non-rationalizing properties as task-irrelevant information. Let us begin
with these non-rationalizing properties and ask whether contents involving them
are explanatorily relevant, in the sense that Q2 asks about.
Consider Dreyfus’ example of a chessmaster playing lightning chess, in which
there is barely time to look at the board before the next move rearranges the pieces
on it. How does the chessboard look to such a player? Dreyfus suggests that at the
very least the master sees patterns of pieces on the board, even if their expertise
leaves them with no need to reason explicitly from those patterns to the next move:
After responding to an estimated million specific chess positions in the process
of becoming a chess master, the master confronted with a new position,
 Affordances and the Contents of Perception   65

spontaneously does something similar to what has previously worked, and lo

and behold, it usually works. In general, instead of relying on rules and stan-
dards to decide on or to justify her actions, the expert immediately responds
to the current concrete situation. . . . When the Grandmaster is playing light-
ning chess, as far as he can tell, he is simply responding to the patterns on the
board. At this speed he must depend entirely on perception and not at all on
analysis and comparison of alternatives.18
The chessboard looks to have its pieces positioned in a certain way. Perhaps only
the relevant pieces appear any way to the chessmaster, and the pieces she knows to
be irrelevant to that stage of the game are attentionally suppressed.19 Different parts
of the board are presumably salient to the chessmaster than would be salient to the
novice in those extremely brief periods between chess moves. The chessmaster’s
expertise might reduce the level of attentiveness to the overall state of the board,
without going so far as to extinguish the experience of the board altogether. If so,
the experienced mandate would structure the perceiver’s attention, but wouldn’t
be systematically related to the contents of component perceptual experience, in
the way that rationalizing properties rationalize mandated action.
By hypothesis, subjects of experienced mandates experience themselves as
being pushed forward from one moment to the next by the situation they’re in.
One might think this makes non-rationalizing contents of experience dispen-
sable in guiding action. But that thought seems mistaken. Even if the mandates
afforded to the chessmasters, for example, have no systematic impact on the
contents of their experience, it is implausible to suppose that a chessmaster
could play lightning chess without experiencing the board at all. By Dreyfus’
own description, the chessmaster “depends entirely on perception” in playing
the game. Their experiences of the board seem indispensable in guiding their
action, or in explaining why the action makes sense to them, even if they are
not rationalized in any way that we or they could reconstruct from the nonra-
tionalizing contents.
Turning to contents that involve rationalizing properties, it might seem ob-
vious that these properties play an explanatory role in making the action seem ap-
propriate to the subject. But the role of such contents in an experienced mandate
could differ, depending on the direction of explanatory priority. In the content-first
direction, you experience the mandate (move the hair, hit the ball back, enter the
forest, etc.), because the component perceptual experience represents the rational-
izing property. Here the forest mandates entering it, at least partly because it looks
canopied. Alternatively, in the action-first direction, the component perceptual ex-
perience represents a rationalizing property, because you experience the mandate.
In this case, the forest would look canopied, at least partly because the experience

18
Dreyfus 2005, p. 8 of web version.
19
On attentional suppression, see van Rullen and Koch 2003.
 Content Views
66   

mandates entering it. A third option is neither factor is explanatorily prior to the
other (perhaps they are connected by a feedback loop).
In the content-first and feedback loop options, the contents of the percep-
tual experience could clearly help explain why the subject performs the man-
dated action. First, they contribute to making the affordance salient, and this
helps explain why it is experienced as a mandate. Second, since on these options
the component perceptual experience is explanatorily upstream of the experi-
enced mandate, they help explain the role of the experienced mandate in guiding
action.
What if the rationalizing contents are related to the mandate in the action-first
direction (e.g., the bed looks fluffy, because the subject experiences plopping down on
it as mandated)? The perceiver collapses onto the bed because she is exhausted, not
because of the way the bed looks to her. Here the contentful sub-experience contrib-
utes to the perceiver’s intellectual coherence and integrity. Compare a case of psycho-
logical (as opposed to normative) rationalization. When people with excessive fear
of heights stand on high balconies, their acrophobia ends up exaggerating how high
they believe the balcony to be, compared to height estimates by non-acrophobes.20 Let
us suppose for the sake of argument that they don’t fear the height because the bal-
cony seems so high off the ground, but rather that the balcony seems so high off the
ground in part because they are afraid of heights. On the assumption that with all else
equal, it is more reasonable to be nervous about standing on a higher balcony than
a lower one, the acrophobes’ mistaken belief about how high the balcony is brings
their fear into harmony with beliefs—even if the beliefs themselves are unreasonable,
caused as they are by an excessive fear, rather than by an accurate assessment of the
situation. Even the craziest, most irrational subjects sometimes display this type of
internal cognitive harmony, such as the schizophrenic patient who is highly anxious
because he thinks that the world is about to end and finds the arrangement of chess
pieces on the chessboard to be ominous.
In these cases, the beliefs that the chess pieces are ominous and that the bal-
cony is very high rationalize the background anxiety or fear in something like
the way that the fluffy-bed experience and the contentful sub-experiences in our
other examples rationalize experienced mandates. The contentful states in all of
these cases, whether they are beliefs or experiences, give us a way to describe this
phenomenon in which subjects (or their subpersonal processes) bring their psy-
chological states into a type of cognitive harmony.21 To the extent that contentful
states figure in these processes, they are not explanatorily idle.

20
Stefanucci and Proffitt (2009) provide some evidence that something like this phenomenon
actually occurs. Using a variety of measures, both acrophobes and non- acrophobes tend to overestimate
the height of balconies they are standing on, but acrophobes exaggerate the height substantially more
than non-acrophobes.
21
If perceptual experiences stand in rational relations to one another, then they belong to a
domain in which rational assessment apply. For discussion of larger implications for this idea for
epistemology, see Siegel 2013 and Siegel, forthcoming.
 Affordances and the Contents of Perception   67

Let us turn to the third question about the role of representation in experi-
enced mandates.
Q3. Is there any proposition that could reflect the mandate that the subject
experiences, if that proposition were the content of the experience?
If all we knew about a subject’s perceptual experience is that it had non-rationalizing
properties, then we could not read off from her perceptual experience the fact that
a subject feels solicited to do something by her environment, let alone that she is
motivated to meet the situation with the solicited action. Neither could we read
this off from the fact that the experience represents properties that would ration-
alize plopping down on a bed or moving a tuft of hair out of the way. Some other
aspect of the experience would have to account for the soliciting and the motivat-
ing aspects of experienced mandates.
Can we identify a proposition that could reflect both the soliciting and moti-
vating aspects of experienced mandates? Let us start with the soliciting aspect of
experienced mandates. How could contents locate the solicitation in something
perceptually experienced? Contents of the form “X is to-be-phi’d,” such as ‘the
hair is to-be-moved’, ‘the passerby is to-be-made-way-for’ are a straightforward
way to characterize the mandated action, while preserving the idea that the man-
date is experienced as issued by something external that we perceive. Consider
the difference between the predicates “to-be-phi’d” and “to-be-done.” Whereas
“to-be-moved-out-of-the-way” is applied to the interlocutor’s tuft of hair or to
some other obstacle, “to-be-done” is applied to actions, such as moving hair out
of the way, or making room on the sidewalk. Given the assumption that solicita-
tions are experienced as being issued by things like cake and forests, a natural
hypothesis is that experienced mandates represent specific to-be-phi’d properties,
attributing them to things perceptually experienced.
What about the motivating aspect of experienced mandates? To bring this
aspect of experienced mandates back into focus, recall the two music cases. The
two experiences of music both solicit dancing, but they differ in whether they
motivate the perceiver to dance. What, if anything, can differentiate between
pairs of cases that differ only in this way, consistently with experiencing the
mandate as issuing from the thing perceived? Could the motivational part of the
mandate be reflected in contents that characterize how the music sounds, or how
the forest, hair, sidewalk, tennis ball, etc. look?
One might think this aspect of experienced mandates is not representable in
experience, on the grounds that it is a conative state, and experiences only take a
stand on how the world actually is, not on how one wants it to be or is motivated
to shape it. If Hume, Searle, and others who endorse this sharp division are right,
then at most, the soliciting aspect of experienced mandates could be represented
in experience.
A first attempt to find a place for the motivating aspect of experienced man-
dates in accuracy conditions adds the issuing of a mandate by X (cake, forest, hair,
 Content Views
68   

etc.) to the ‘to-be-phi’d property. The result would be that experience has conjunc-
tion of contents. Here are some candidates for the conjunction:
Exp: X is to-be-phi’d and X . . .
. . . wants me to phi
. . . is telling me to phi
. . . commands me to phi
. . . intends for me to phi.
But this option does not identify any difference between the two music cases,
for the same reason that ‘X is to-be-phi’d’ does not identify any such difference.
One need not feel moved by what X wants, tells you, or intends. As evidenced by
long-standing, unfulfilled to-do lists, representing that something is to be done,
for instance by writing it down on a list of things to do, does not suffice to moti-
vate you to do it.
A different strategy is to complicate the second component of content further,
by adding a causal relation that links the soliciting aspects to a desire or an action:
• Exp: X is to-be-phi’d, and the fact that X is to-be-phi’d
. . . makes me want to phi.
. . . is making me phi.
• Exp: X is to-be-phi’d, and because X is to-be-phi’d, I am going to phi.
These proposals posit ascriptions of one’s own desires or intentions as part of the
content of experience. One might worry that this fails to respect the way in which
perceptual experience is directed outward, characterizing things external to the
subject’s mind. The phenomenal integration of solicitation and motivation is re-
flected in the unreflective nature of habitual and specialized skilled action. Sartre
describes the integration when he uses a locution of the form “to-be-phi’d” to
describe being “plunged into the world of objects”:
When I run after a streetcar, when I look at the time, when I am absorbed in
contemplating a portrait, there is no I. There is consciousness of the streetcar-
having-to-be-overtaken . . . I am then plunged into the world of objects; it is
they which constitute the unity of my consciousness; it is they which present
themselves with values, with attractive and repellant qualities—but I have dis-
appeared; I have annihilated myself. There is no place for me on this level.
And this is not a matter of chance, due to a momentary lapse of attention, but
happens because of the very structure of consciousness.22
Sartre may overstate the ‘disappearance of the subject’. Representations of direc-
tionality and distance have an implicit first-person component, so contents involv-
ing the subject seem indispensable. But he seems right that the subject disappears
as a subject of desire. We need not be aware of our motivation to hit the tennis ball,

22
Sartre 1957, pp. 48‒49. I put in bold the locution of interest.
 Affordances and the Contents of Perception   69

or flick the light switch, or move aside to let the oncoming pedestrian pass. We’re
just aware of the ball, the switch, the sidewalk, etc.
A more promising proposal is that the motivating aspect of experienced man-
date casts its shadow on the contents of experience, in the form of contents that
include a property related to answerability. What’s needed is a hypothesis that
identifies contents that are correlated with the feeling of answerability to a solicit-
ing affordance.23 The effort to find such contents may seem futile, for any of three
reasons:

• Challenge 1: The feeling of answerability has no internal structure to it,

because it is a simple ‘buzz’ akin to a valence that attaches to thing you
feel answerable to.
• Challenge 2: The feeling of answerability belongs on the side of the
force-content distinction belonging to force.
• Challenge 3: The only contents that are correlated with the feeling of
answerability in experienced mandates are a version of “A is to-be-
done,” and these have the same problem considered earlier. From the
fact that you represent something as to-be-done, it does not follow that
you are motivated to do A.

Regarding the first challenge, assuming the notion of a felt valence has psycholog-
ical validity, it seems reasonable to suppose that it figures in feelings of answer-
ability in experienced mandates.24 But it is doubtful that the felt valences exhaust
those feelings of answerability. If a pack of chewing gum looks appealing, it need
not cost you any dissonance to ignore it. Here the perception of the gum does not
involve any persisting inner mobilization of the sort that characterizes the feeling
of answerability.
The other two challenges are not easily rejected. On the face of it, feeling
answerable to a solicitation seems to be a mode of experiencing the solicitation.
One experiences it in a mobilizing way, as opposed to experiencing it indiffer-
ently. This observation suggests that it must belong on the side of the force/
content distinction belonging to force. And while a natural candidate for an ac-
curacy condition associated with the feeling of answerability would seem to be
“A is to-be-answered,” this suggestion invites the charge that like other ‘to-be-
phi’d’ contents, it too can be entertained indifferently.
The strongest answer to these substantial challenges combines structural con-
siderations with closer attention to what is experienced in the feelings of answer-
ability that are at issue. The structural consideration is that contents could have a
nested structure that reflects the main idea driving Challenge 2. The nested struc-
ture seems to respect the fact that there are both motivated and indifferent ways to

23
I leave open whether the contents are constitutively linked to the phenomenal character, rather
than being merely correlated with them.
24
For discussion of microvalences, see Lebrecht et al. 2012.
 Content Views
70   

experience soliciting affordances. Leaving a crucial element partly blank for now,
the structure could be this:
Experience: [It is < . . . answerability . . . > that: X is to-be-phi’d].
What fills in the ellipsis to create an accuracy condition? “. . . to-be-answered . . . ”
sounds prospective, and suggests that answering to the soliciting affordance is
something the subject may or may not go on to do, whereas what we’re looking
for is a way to reflect the fact that the subject is already answering to the soliciting
affordance. A better proposal is thus:
Experience: [It is answered that: X is to-be-phi’d.]
I’ll call the contents on the right side of the word “Experience” answerability contents.
Of course these contents are not anything one would find natural to say in
describing the experience. But the same is true of many other accuracy conditions.
The proposal here respects the integration of soliciting and motivating aspects
of experienced mandates, and the fact that the soliciting affordance generates a
feeling of answerability.
In cases where an experience also represents rationalizing properties, the
contents embedded in “it is answered that” may be more complex, integrating the
rationalizing properties with the to-be-phi’d property. For instance, “It is answered
that: the bed is to-be-plopped down upon and is fluffy.” I leave it open whether the
rationalizing relation itself might be represented in experience, as would be reflected
in contents such as “ . . . to bed is to-be-plopped down upon because it is fluffy.”25
What are some cases in which these accuracy conditions are met, and what
are some cases in which they are not met? It seems plausible that the feeling of
answerability suffices for the subject of the feeling to be answering to something
in a minimal way—a way that does not consist in taking the thing that they are
answering to (such as a piece of cake) to be a source of normative constraint. If
so, then answerability contents are always correct. Compatibly with this result, a
subject could in principle make an introspective error about whether she is or isn’t
feeling answerable to something.
In addition to probing the conditions under which answerability contents are
true, we can also ask about the conditions under which the ‘X is to-be-phi’d’ contents
they embed are true. What would it take for it to be the case that the hair really is
to-be-moved, or that the forest really is to-be-entered, the oncoming pedestrian is
to-be-made-way-for, or that the cake really is to-be-eaten? These questions have no
general answer, because the contents do not specify what kind of ‘ought’ underlies
the mandate. For instance, if the morally correct thing to do on the path is move
aside by giving the passerby lots of room to pass, then relative to moral ‘ought’, the
pedestrian is to-be-made-way-for. In the hair case, relative to the ‘ought’ of social
mores, it is not the case that the hair is to-be-moved, but perhaps relative to the

25
For discussion of seeing reasons, see Church 2010.
 Affordances and the Contents of Perception   71

‘ought’ of communicative efficacy or aesthetic rightness, it is the case that the hair
is to-be-moved (assuming that the interlocutor really does have an eye underneath
the hair).
Even though answerability contents do not specify a norm relative to which
they are accurate, which mandates a person experiences can indicate which norms
she is sensitive to. In talking to a person you might feel a mandate to stop lis-
tening, or deeply discount what they say, or (on the other side) to put a lot of stock
in what they say. Such experiences might manifest and perpetuate background
attitudes of deference or disrespect. They illustrate the potential use of the frame-
work of answerability contents in analyzing the interpersonal interactions. Such
modes of aversion and approach can be the social instrument by which social
patterns are maintained, such as patterns of exclusion and inclusion, or trust and
dismissiveness.
These considerations can illuminate the conditions under which to-be-phi’d
contents would be accurate. Relative to the social norms that such experienced
mandates manifest, the answerability contents would be accurate, where relative
to epistemic or moral norms, they might be inaccurate. For instance, if one experi-
ences one’s interlocutors’ comments as to-be-discounted when such discounting
would be epistemically inappropriate, then the ‘to-be-discounted’ contents could
be accurate relative to a social norm that mandates discounting, but inaccurate
relative to epistemic norms.
The ‘to-be-phi’d’ contents that answerability contents embed are thus not
complete accuracy conditions, because they leave unspecified a parameter that
needs to be fixed in order to generate an accuracy condition. The reason to think
this parameter is left unspecified is that there don’t seem to be phenomenal dif-
ferences that track different norms relative to which to-be-phi’d contents could be
assessed for accuracy.
I’ve replied positively to question Q3 by arguing that answerability contents
could identify a proposition that reflects the mandate that the subject experiences.
Answerability contents are always true, and their embedded contents can be accurate
or inaccurate only once an additional parameter is fixed. In light of these facts, does
this response to Q3 concede anything to the overall challenge to the primacy of rep-
resentation from experienced mandates? To the extent that the embedded contents
are accurate only relative to a type of norm that the experiences do not themselves
specify, experienced mandates fail to provide full accuracy conditions by themselves.26
But rather than being irrelevant to experienced mandates, answerability contents
let us express important features of this phenomenon. They let us identify rational
relationships between subexperiences, they provide a framework for understanding

26
Klein (2007) observes that experiences with imperatival contents (such as “step gingerly on
your left foot”) would not threaten representationalism about experience. This observation is true but
irrelevant to the discussion of experienced mandates, since replacing the declarative content “the hair
is to-be-moved” with the imperatival content “move the hair” would not suffice to reflect the motivated
aspect of experienced mandates.
 Content Views
72   

mechanisms of social interactions including power relationships exerted through dis-

course, and they give us a way to analyze common phenomenal strand throughout
different forms of normativity. The remaining question is then whether experienced
mandates have answerability contents.

3.2.2. DO EXPERIENCED MANDATES HAVE ANSWERABILITY

CONTENTS?

With answerability contents on the table, we can return to the stronger thesis that
experienced mandates have contents like these. This thesis goes beyond answering
our initial three questions Q1‒Q3. Can we discover whether experienced man-
dates have such contents?
A possible strategy starts from the substantial assumption that affordances
can be represented in experience (and that experiences therefore have accuracy
conditions), rather than trying to defend the stronger thesis from the ground up.
Given this assumption, one might try to argue for two conditionals:
Affordances → Solicitations: If affordances are represented in experience,
then solicitations are too.
Solicitations → Mandates: If solicitations are represented in experience, then
mandates are too.
Favoring the Affordance → Solicitation conditional, one might reason roughly as
follows. From the cases of advertising and dance music, the feeling of being solic-
ited by the environment to do something is familiar. If we are entitled (by our
starting assumption) to use the idea that affordances are represented in experience
to analyze perceptual experience, it is natural to use it for the special class of salient
affordances that we experience as soliciting. There seems to be no principled bar to
extending the analysis of perceptual experience from representing of affordances
to representing the special case of soliciting affordances.
Favoring the Solicitation → Mandate conditional, one could try to use a
method of phenomenal contrast (Siegel 2010) to evaluate hypotheses about what
best explains of the phenomenal contrast between the two music cases, or another
pair of cases in which two subjects seem to have experiences with the same so-
licitation content, but differ in whether they feel motivated to fulfill the solicita-
tion they experience. The method could either be used to evaluate the hypothesis
that the motivational aspect is best analyzed in terms of accuracy conditions of
the form:
It is answered that: X is to-be-phi’d.
(The same cases and method could also be used to evaluate other hypoth-
eses that posit different accuracy conditions, or something other than accu-
racy conditions.) Since the burden of the strategy is to show that this is the
best explanation of the phenomenal contrast, alternative hypotheses need to
 Affordances and the Contents of Perception   73

be considered. We considered some alternatives earlier in the conjunctive pro-

posals. The alternatives also include one suggested by Dreyfus’ remarks about
the dynamic of tension and relief that fuel the intentional arc. This dynamic
might be invoked to analyze both the soliciting and the motivating aspects
of experienced mandates. Relief arises from meeting the situation with the
mandated action, and tension from not meeting it that way, or from not yet
having done so. Given these correlations, the dynamic of tension and relief is
well-suited to guide the action. Its suitability is part of what threatens to make
the contents of perceptual experience explanatorily irrelevant.
But by itself, the dynamic of tension and relief does not account for the
experience of solicitation per se. Nothing in the ebb and flow of tension and
relief reflects the experience of someone’s hair, a passerby, or the forest solicit-
ing one to perform an action that will relieve the tension one feels in that sit-
uation. You might feel worse if you don’t move the hair, and better if you do,
but those facts could obtain even if you don’t feel pulled by the hair to move it
aside. One need not be “plunged into the world of objects” for the facts about
tension and relief to hold.
This observation forms the basis of a dilemma concerning the relationship
between the situation and the feelings of tensions and relief that it gives rise to.
Either the course of tension and relief is merely caused by the situation in one’s
immediate environment, or else those feelings are psychologically more complex
responses to the situation that involves some type of understanding of what the
situation demands. If the relationship is merely causal, then that relationship
by itself does not illuminate how the situation is experienced as soliciting the
action to which tension and relief are sensitive. If instead the tension and relief
are byproducts of understanding what the situation calls for, then that under-
standing, whatever form it takes, has just as much claim to guide the action as
the dynamic of tension and relief has. Either way, the dynamic of tension and re-
lief by itself is inadequate to account for experiences of being solicited by things
in the environment.
I have outlined an argumentative strategy for the thesis that mandates are
reflected in the contents of experienced mandates, and criticized an alternative
explanation. To follow the strategy through, much more would need to be said
about which pairs of experiences contrast phenomenally in the right way, and the
alternative explanations of that contrast. But the discussion of questions Q1‒Q3
suggests the strategy as a starting point.

3.3. Conclusion

Dreyfus and other writers who have described experienced mandates call atten-
tion to an important fact about perception: sometimes our perceptual experiences
are pervasively structured by our role as agents responding to social situations.
 Content Views
74   

In other situations, our dominant mode is not that of an agent, but a spectator—
for instance when we are freed from immediate pressures of spatial negotiation,
simply taking in our surroundings. These writers are right to emphasize that phe-
nomenologically, perception feels quite different depending on whether it is dom-
inantly structured by our roles as agents or not. And that raises a question: to what
extent are our experiences structured by experienced mandates, to what extent
aren’t they? An upshot of the discussion here is that even if the extent is great, the
role of the spectator never disappears completely, even when we’re in the throes
of action.

References

Bengson, J. (n.d.). Practical Perception. Manuscript.

Bortolotti, L. (2013). Delusion. In Edward N. Zalta (Ed.), The Stanford Encyclopedia of
Philosophy. Fall 2013 edn. http://plato.stanford.edu/archives/fall2010/entries/delusion/.
Church, J. (2010). Seeing reasons. Philosophy and Phenomenological Research, 80(3),
638–670.
Cussins, A. [1990] (2003). Content, conceptual content, and nonconceptual content. In Y.
Gunther (Ed.), Essays in Nonconceptual Content (pp. 133–0163). Cambridge, MA: MIT
Press.
Dreyfus, H. (2002). Intelligence without representation: Merleau-Ponty’s critique of mental
representation. Phenomenology and the Cognitive Sciences, 1, 367–383.
Dreyfus, H. (2005). Overcoming the myth of the mental: How philosophers can profit from
the phenomenology of everyday expertise. Proceedings and Addresses of the American
Philosophical Association, 79(2), 47–65. Reprinted as Overcoming the myth of the
mental. Topoi, 25(1–2), 43–49 (2006).
Dreyfus, H. (n.d.). A Phenomenology of Skill Acquisition as the Basis for a Merleau-Pontian
Non-Representationalist Cognitive Science. http://socrates.berkeley.edu/~hdreyfus/
pdf/MerleauPontySkillCogSci.pdf.
Gibson, J. (1977). The theory of affordances. Reprinted in Robert Shaw & John Bransford
(Eds.), Perceiving, Acting, and Knowing (pp. 127–143). Hillsdale, New Jersey: Laurence
Erlbaum, 1986.
Hurley, S. (2006). Consciousness in Action. Cambridge, MA: Harvard University Press.
Kelly, S. (2006). Seeing things in Merleau-Ponty. In T. Carman and M. Hansen (Eds.),
The Cambridge Companion to Merleau-Ponty (pp. 74–110). Cambridge: Cambridge
University Press.
Kelly, S. (2010). The normative nature of perceptual experience. In B. Nanay (Ed.), Perceiving
the World (pp. 146–159). New York: Oxford University Press.
Klein, C. (2007). An imperative theory of pain. Journal of Philosophy, 104(10), 517–532.
Lebrecht, S., Tarr, M., Bar, M., & Feldman Barrett, L. (2012). Microvalences: Perceiving
valences in everyday objects. Frontiers in Psychology, 3(April), Article 107.
Mandelbaum, M. (1969). The Phenomenology of Moral Experience. Baltimore: Johns
Hopkins University Press.
Marr, D. (1982/2010). Vision. W. H. Freeman. Cambridge, MA: MIT Press.
 Affordances and the Contents of Perception   75

Merleau-Ponty, M. (1962). Phenomenology of Perception. London: Routledge and Kegan

Paul; New York: Humanities Press.
Milner, M. A., and Goodale, A. D. (1995). The Visual Brain in Action. Oxford: Oxford
University Press.
Murdoch. I. (1958). The Bell. Penguin Books. New York.
Nanay, B. (2010). Attention and perceptual content. Analysis, 70(2), 263–270.
Nanay, B. (2011). Do we see apples as edible? Pacific Philosophical Quarterly, 92(3), 305–322.
Noë, A. (2006). Action in Perception. Cambridge, MA: MIT Press.
Orlandi, N. (2014). The Innocent Eye. New York: Oxford University Press.
Pautz, A. (2010). Why explain visual experience in terms of content? In B. Nanay (Ed.),
Perceiving the World (pp. 254–310). New York: Oxford University Press.
Sartre, J. P. (1957). The Transcendence of the Ego. United States of America: The Noonday
Press.
Seuss, T. Geisel (1961). The Zax. In The Sneetches and Other Stories. New York: Random
House.
Siegel, S. (2010). The Contents of Visual Experience. New York: Oxford University Press.
Siegel, S. (forthcoming). How is wishful seeing like wishful thinking? Philosophy and
Phenomenological Research.
Stefanucci, J. K., & Proffitt, D. R. (2009). The roles of altitude and fear in the perception of
height. Journal of experimental psychology: Human perception and performance, 35(2),
424–438.
van Rullen, R., & Koch, C. (2003). Competition and selection during visual processing of
natural scenes and objects. Journal of Vision, 3, 75–85.
Witt, C. and Proffitt, D. (2008). Action-specific influences on distance perception: A role
for motor simulation. Journal of experimental psychology: Human perception and per-
formance, 34(6), 1479–1492.
 4

Looks, Reasons, and Experiences

Kathrin Glüer

4.1. Introduction

Whether perceptual experience has content has been a question of lively debate
for the last decade or so. One way of construing the question is whether percep-
tual experiences are mental states much like beliefs and desires—whether they
are mental states, that is, that can be understood on the model of the proposi-
tional attitudes. That is the way in which I shall construe the question in this
chapter. And I shall offer an argument for answering it in the affirmative, an ar-
gument that to my mind at least has not received quite the attention it deserves
in the recent debates. The argument starts from the observation that percep-
tual experience figures in folk-psychological reasons-explanations in a certain
way. And playing certain sorts of folk-psychological roles is arguably one of
the strongest motivations for construing mental states as having content in the
first place—modeling these roles is precisely what contents are for. Whether
experience has content thus clearly is a question of significance not only for the
metaphysics of mind but also for epistemology, the theory of rationality, and the
theory of action.

4.2. Contents and Reasons

Asking whether experience has content immediately raises to two further ques-
tions: What are contents? And: What is it for a psychological state to have a con-
tent? These questions are closely related: Asking what it is for a psychological state
to have content amounts to asking what contents are there for, what their theoret-
ical role or function is. And if contents are whatever best fulfills that role, we can
find out what they are by means of investigating what that role is and what best
fulfills it.
In this section, I shall provide a rough-and-ready answer to these questions
and apply it to the question whether experience has content. The account of per-
76 ceptual experience and its content I shall suggest is highly controversial, but it
 Looks, Reasons, and Experiences   77

will be based on a rather traditional answer to the foundational content-theoretic

questions just outlined—I shall employ a content-theoretic framework inspired
by Donald Davidson and David Lewis.1 That it has a venerable tradition behind it
will, moreover, be the only argument provided for using this framework.2
A simple, old-fashioned answer to the question what contents are is the fol-
lowing: Contents either are, or are essentially such that they determine, truth or
correctness conditions. Truth or correctness conditions are conditions the world has
to satisfy in order for the content in question to be true. Contents thus are whatever
essentially has truth values; they are propositions.3 Until recently, this idea com-
manded broad consensus amongst theorists of content. One seemingly immediate
very significant advantage of identifying contents with propositions that appealed
to many was that it allows us to assign uniform contents to both linguistic expres-
sions (or their utterances) and certain mental states, the propositional attitudes.
For reasons soon to be apparent, I shall follow this tradition. In what follows I shall
thus not make any distinction between contents and propositional contents, and I
shall think of contents and propositions—whatever they more precisely turn out to
be—as that which essentially has truth values.4 This minimal understanding of a
proposition is, I hope, uncontroversial.5
Thinking of contents in terms of propositions is thinking of contents as ab-
stract objects of a certain sort. These objects, by their very nature, are such that
they stand in certain relations to one another. Propositions are essentially such
that they bear logical and inferential or evidential relations to one another. Just like
truth values, at least some of these relations depend on the world. Nevertheless, it
is an entirely objective matter what truth values propositions have, and in which
inferential or evidential relations they stand to one another. Together, these prop-
erties are what uniquely qualifies the propositions for their role as the objects of

1
Cf. Davidson 1973, 1975, 1974, 2005; Lewis 1972, 1979.
2
There is, I think, a discussion to be had about the theoretical role of the notion of content, a
discussion of some urgency even, especially regarding its relation to the notion of semantic value. But
this is not the right place for that discussion.
3
Davidson, of course, thought that we could do without any objects—be they propositions or
whatever—explicitly assigned as meanings to utterances or contents to attitudes. He not only held
such objects to be ultimately redundant in the theory of meaning and content, but also worried that
proposition-talk might (falsely, according to him) suggest that there are unique correct assignments
of meaning and content (cf. Davidson 1974, p. 147). Nevertheless, Davidson himself talks quite freely
of propositional contents, and the minimal notion of propositional content adopted below should be
entirely compatible with his strictures.
4
I shall remain neutral here on the question of whether this content is structured.
5
Tradition thus has it that all psychological states with contents are propositional attitudes, where
‘propositional attitude’ is taken to have an inclusive and hopefully uncontroversial sense, too. In this
sense, entertainings are propositional attitudes, for instance. Tradition also has it that the attitudes have
uniform contents in the sense that all attitudes have whole propositions as contents. This, I take it, rules
out propositional functions as attitude contents. It does not rule out that the contents of propositional
attitudes are best construed as centered worlds propositions, however. Whether or not they are is
another question I remain neutral on here.
 Content Views
78   

those psychological states that we think of as content bearing: The propositional

attitudes.
Their function in this respect is threefold:

(i) to individuate these states in an adequately fine-grained way. Attitudes

with different contents are different;
(ii) to explain certain significant similarities between attitudes of different
attitudinal kinds; thus, a belief and a desire can have the same object;
and
(iii) to explain and predict propositional attitudes and intentional actions on
the basis of other such attitudes and actions.

Contents thus play the following role: They are abstract objects we assign to psy-
chological states to model a specific kind of structure among them, to locate in-
dividual states in such structures, and to predict and explain their formation by
means of other such states present in such structures. The states are those our
folk-psychology recognizes, the explanations are reasons-explanations, and the
individuation of the relevant states is by means of propositional contents and folk-
psychological role. Here is a recent voice, that of Richard Heck, summing up this
way of thinking about content:
Why should we attribute content to mental states at all? A common answer
might be that mental states are representational: Talk of a state’s content is
short for talk of its representational properties. That is certainly true. But why
trouble ourselves with the representational properties of mental states? What
would we lose if we just ignored them? I take it that we would lose the very
idea of psychological explanation. We are in the habit of explaining our own
behavior, and that of other creatures, in terms of what we all believe: We ex-
plain why Joe ran across the room in terms of his believing that his stuffed di-
nosaur was on the other side. . . . The explanations themselves are formulated
not in terms of the neurological features of mental states but in terms of their
contents . . . . And so we might say: The reason we should attribute content to
mental states is because there are things we wish to explain in terms of mental
states, as individuated by their contents. (Heck 2007, pp. 120f.)
If sufficiently fine-grained individuation was our only concern, propositions might
not be the objects of choice for the role of contents; other kinds of abstract objects
might do as well. But psychological explanation requires more than sufficiently
fine-grained individuation; it requires relations beyond sameness and difference.
Propositions by their very nature stand in certain relations to one another that
uniquely qualify them for their job: the logical and inferential or evidential rela-
tions. Heck again:
Why not just take the contents of beliefs to be (possibly transfinite) ordinal
numbers? . . . [T]‌he best answer, it seems to me, is that mental states are not
 Looks, Reasons, and Experiences   79

just distinguished from one another by their contents: They are also related
to one another by their contents. For example, given any two beliefs, there are
several other beliefs that are related to them in familiar ways: Their negations,
their conjunction and disjunction, and so forth. These relations are not just
logical but also psychological: Someone who believes two propositions will,
ceteris paribus, also tend to believe their conjunction, at least when the ques-
tion arises. (Heck 2007, p. 121)
For beliefs, that is, reasons explanations are possible because the beliefs of actual
believers at least to some (minimal) degree tend to instantiate the objective log-
ical and inferential or evidential relations between the propositions that are their
objects. Together with beliefs, the ascription of contents to desires (or pro-attitudes)
then allows for reasons explanation of intentional actions.
From this perspective, content first and foremost is what allows a psycho-
logical state to play a role in reasons-explanations. Consequently, the question of
whether a given kind of psychological state does, or does not, have content is (at
least to a large part) to be decided by the question of whether this kind of state
does, or does not, play a role in reasons-explanations. Reasons-explanations ‘surf ’
on the logical and inferential or evidential relations ‘induced’ between psycholog-
ical states by means of assigning contents to them.
Following this content-theoretic tradition thus suggests the following way
of looking at the question whether a given kind of psychological state has con-
tent: Do the states in question show the kind of inferential integration into their
subject’s system of propositional attitudes that would come with assigning con-
tents to them? Does the state in question provide its subjects with reasons? Is this
an important part of our folk-psychology? If the answers to these questions are
positive, we have an excellent prima facie case for assigning propositional contents
to that kind of state.
There is a complication, however: Assigning propositional content to a
kind of psychological state is necessary, but not sufficient for accounting for its
inferential integration into a system of propositional attitudes. Different kinds
of attitudes play different kinds of reason providing roles. The role of desire as
a (practical) reason provider is significantly different from that of belief. Any
account of the psychological role of a propositional attitude is thus a function
of two variables: content and attitude.
And while there are any number of contents, traditional models of theoretical
and practical reasoning, or reasons-explanation, contain only two kinds of atti-
tudinal ‘slots’: motivational and doxastic slots. So, for any propositional attitude,
it can play one of two reason providing roles: a desire-like motivational role, or a
belief-like doxastic role. To the extent that doxastic role in theoretical reasoning
amounts to a justificatory role, folk-psychology here amounts to folk-epistemology.
With these elements in place, we can now turn to our initial question: What
about perceptual experience? Does experience provide its subjects with reasons?
 Content Views
80   

And if so, does it fit into one of the slots provided by traditional accounts of reason
providing?
In previous writings (esp. my 2009), I have argued that it is, indeed, an in-
tegral part of our pre-theoretic conception of perceptual experience that experi-
ence provides reasons for its subject’s beliefs and actions. Like other proponents
of this claim—such as John McDowell (1994), Bill Brewer (1999), or Richard Heck
(2000)—I have focused on reasons for belief:
(R) Experience provides its subject with reasons for first-order empirical
belief.
Unlike other proponents of (R), I have combined (R) with a doxastic account of
perceptual experience, an account according to which experience is a (peculiar)
kind of belief. One of the main advantages of such an account, I have argued, is
precisely that it allows to us to understand the reason providing role of experi-
ence on the model of the reason providing role of (non-experiential) belief, thus
keeping our overall account of theoretical reasoning unified and traditional.
Availing ourselves of this advantage comes at a price, however. We only get
a plausible account of the reason-providing role of experience from a doxastic
account, if we are willing to construe experience contents in a certain non-
standard way. Standardly, experience contents are construed as being of the
‘naïve’ form x is F, where x ranges over ordinary material objects, and F over
sensible properties.6 Together with a doxastic account, this has a number of
unpalatable consequences.7 Not the least of them is that any inference from an
experience content to the most basic kind of experience based belief would be
of the ‘stuttering’ kind: It would be an inference from p to p. If nothing else, that
makes a hash of the idea that perceptual reasons are defeasible.8 Such conse-
quences can be avoided, or so I have argued, if we construe the contents of ex-
perience as ‘phenomenal contents’. Phenomenal contents ascribe ‘phenomenal
properties’ to ordinary material objects, properties such as looking F.9 In what
follows, I shall also call these contents ‘looks-contents’ or ‘Lp-contents’.

6
Strictly speaking, and on the assumption that there are sensible relations, it should be said that
experiences have contents of the form F(x1, . . ., xn), where xi (1 ≤ i ≤ n) ranges over material objects and
F over sensible properties and relations. Given the richness of experience, full experiential contents
moreover should probably be construed as (long) conjunctions of such predications.
7
For instance, it would result in outrightly contradictory beliefs in cases of known (or believed)
illusion. One might say that this is a kind of compartmentalization, a kind of compartmentalization,
moreover, that is to be expected given the ‘modularity’ of perceptual experience. Alex Byrne, who at
least seems to be quite tempted to construe experiences as beliefs, has taken this line in conversation.
8
I discuss what I call the ‘stuttering inference argument’ at some length in my (2009). The
argument derives from McDowell 1998, pp. 405f.
9
As customary, I shall focus exclusively on visual experience here. It is not completely clear how
to generalize the account to olfactory and auditory experiences as these, prima facie at least, do not
seem to take ordinary material objects.
 Looks, Reasons, and Experiences   81

From the Davidsonian-Lewisian perspective on content, accounting for its

intuitive reason providing role and its consequent inferential integration into
its subject’s system of propositional attitudes is the best possible substantiation
of the claim that perceptual experience has propositional content. In this chap-
ter, I shall defend the claim that experience indeed has propositional content
by further developing and defending the doxastic account of experience I have
suggested.10
I shall proceed as follows: In section 4.3, I shall briefly recapitulate the main
elements of the account and explain in what sense it allows experience to provide
its subject with reasons. Then, I shall take up two challenges to the account, one
concerning the epistemic role of experience and one concerning its contents.
In section 4.4, I shall ask when experiential reasons are good reasons. Can the
doxastic account on offer provide a plausible epistemology for perception-based
belief? In section 4.5, I shall take up a challenge to construing the contents of
experience as looks-contents: ‘Looks’, it has been argued, is itself a propositional
attitude operator. It thus has no place in the content of any first-order proposi-
tional attitude.

4.3. Experiences, Beliefs, and Reasons

The doxastic account of experience I have suggested combines two elements: The
claim that experience is a kind of belief, and the claim that (visual) experience has
looks-contents:
(PB) Perceptual experience
(i) is a kind of belief
(ii) with phenomenal contents.
Of course, there are many important differences between those beliefs that are
experiences and other, non-experiential beliefs. Most importantly, perceptual
experiences have a distinctive sensory nature: Phenomenally, having an experi-
ence is very different from having a non-experiential belief.11 Experience thus is a
kind of belief, a kind that any satisfactory account of experience ultimately should
be able to specify.

10
This defense leaves any number of questions open: In what follows I do not take a stand on, for
instance, the issue of the kind of proposition best assigned as content to experience, nor the issue of
how experiential contents are represented. For all I care here, experiential representation might well
be “iconic” (Fodor 2007) or “analogue” (Dretske 1981, p. 135ff.). On my use of ‘propositional’, that
does not prevent them from having propositional content. McDowell and Crane, by contrast, recently
have come to tie having propositional content to being represented by sentence-like structures of a
language-like medium. Cf. McDowell 2008; Crane 2009.
11
This holds regardless of whether we think that there is such a thing as cognitive phenomenology.
 Content Views
82   

One immediate question for (PB) thus is whether (i) and (ii) are neces-
sary and sufficient for a state’s being a perceptual experience, or only necessary.
That, I think, depends on how precisely looks-contents are construed. Are there
non-experiential beliefs with looks-contents? I tend to think not. I think that the
sense in which ‘looks’ figures in the content of experiences is the so-called phe-
nomenal sense.12 In this sense, an object x cannot look F simpliciter. Rather, it
always looks F to a subject S at a time t. Moreover, x looks F to S precisely if S, at
that very time, has an experience as of x’s being F. Working along these lines, (PB)
might be able to distinguish the experiences from the rest of our beliefs by their
type of content. If that ultimately does not work, (PB) would have to fall back on
experience’s distinctive phenomenology to single out the relevant kind of belief.
Both having a distinctive type of content and having a distinctive kind of phe-
nomenology, however, are compatible with sharing the attitude component with
ordinary common and garden beliefs. Not only is this compatible, the attitude
component in perceptual experience is moreover best construed as that of holding
true or belief—or so I suggest.13 One of the most important and immediate advan-
tages of thus construing the attitude component is that it allows us to subscribe
to (R): If experiences are beliefs, they provide reasons for (further) belief in the
relatively well understood, traditional sense in which beliefs provide reasons for
(further) beliefs.14
Let me spell out in a bit more detail the sense of reason, and reason providing,
in which I take (R) to be an important and integral part of folk-psychology.15 The

12
For more on that, see below, section 4.5.
13
I am not here going to argue against any kind of sui generis account of experience that allows
experiences to be holdings true, without thereby subsuming them under the beliefs. See my (2009),
however, for an argument to the effect that the availability of doxastic accounts that preserve the special
functional role of experience undermines the very motivation for sui generis accounts.
14
The account has other advantages: It accounts for the ‘modularity’ or ‘belief-independence’
of experiences by means of their contents, not by compartmentalization, thus preserving the full
rationality of the subject of a known illusion. It allows for uniformity of contents across veridical and
non-veridical experiences. Most (virtually all) of the experience-beliefs will be true, of course, but
(PB) can account for non-veridicality in terms of misleadingness: non-veridical experiences are those
that provide their subjects with prima facie reasons for false beliefs. (PB) thus is compatible with the
intuition that non-veridicality somehow is ‘downstream’ of experience, a matter of (non-experiential)
belief (cf. Brewer 2006; Travis 2004). It also accommodates various phenomenological observations
regarding experience, for instance what is sometimes called its ‘immediacy’, its ‘presentational’
or ‘committal’ character. (PB) captures this by construing experience as belief. (PB) also captures
experience’s particularity, i.e., the claim that the veridicality of an experience depends on the intuitive
object of the experience, not on some other object that happens to make its content true. Looks-
contents naturally construe experiences as about those very objects they intuitively are about: Those
objects causally responsible for them (in the right way, of course). Moreover, (PB) accommodates
what is reasonable about transparency: Phenomenal properties, whatever their ultimate analysis, are
properties of ordinary material objects, not properties of experiences. Finally, and on the assumption
that there is such a thing as a phenomenal notion of looks, (PB) satisfies the desideratum—if it is one—
that experience content be “looks-indexed,” i.e., that experience content is determined by the looks of
things (Travis 2004).
15
All I need for my argument is that reason providing in this sense is indeed a deeply entrenched
 Looks, Reasons, and Experiences   83

relation of reason providing we are interested in is a relation between two propo-

sitional attitudes, more precisely between two first-order empirical beliefs. One of
them, the belief that p, provides its subject S with a reason for another such belief,
the belief that q. Strictly speaking, the reason this belief provides S with is a propo-
sition: p. S ‘has’ this reason in the sense that it is the content of one of S’s beliefs. In
this respect, the notion of a reason we are interested in is subjective.
Such reasons rationalize further beliefs. If p is a reason for believing that q,
then p must be such that it—from the subject’s own perspective—speaks in favor
of believing that q. If S forms the belief that q on the basis of believing that p, cit-
ing p as a reason must confer some degree of rationality on believing that q. This
rationality, too, is subjective in the sense that the explanation provided by citing
the reason is such that it shows that something spoke for believing q from S’s own
perspective.
At the same time, however, S’s perspective needs to be recognizable as a per-
spective by others: Believing that p needs to be such that it would provide any
subject with a reason to believe q. This objective, or at least intersubjective, aspect
of reason providing can only be secured by an underlying, objective relation of
inferential or evidential support between p and q. In order for the belief that p
to provide its subject with a reason for believing that q, there needs to be a valid
inference (of some sort) from p to q.
At this point, it is important to note that we are not concerned with good
reasons—yet. A subject can have reasons for and against forming a certain be-
lief. These reasons need to be weighed against one another to determine what the
subject has good reasons to believe—if anything. But so far, we are concerned
with the more basic notion of having, or providing, a reason—of having, or pro-
viding something that even qualifies for such weighing. We are, in other words,
concerned with prima facie reasons.
Two more (negative) characteristics are crucial: Having reasons for one’s
first-order empirical beliefs does not require forming them by means of conscious
inference. Having a reason for believing that q does not require anything regarding
how that belief is formed, not even that it actually be formed at all. And anal-
ogously for reasons explanations: We can explain S’s believing that q by means
of her believing that p without implying anything about any conscious thought
processes or deliberations on S’s part. To be sure, for one belief to reasons-explain
another, the latter must somehow be based on the former, but the relation need not
be one of conscious inference.16
Connected with this is the observation that having or providing reasons does
not require the possession of second order states. What we are concerned with

part of folk-psychology. I do not need to deny that there are other, more objective notions of reason,
notions that might be useful in epistemology (or elsewhere). I do not even need to deny that such
notions also are part of folk-psychology.
16
Ever since Davidson’s (1963) paper “Actions, Reasons, and Causes,” there has been widespread
consensus that reasons explanation is a species of causal explanation.
 Content Views
84   

here are relations between first order propositional attitudes and their contents.
A creature has reasons in the required sense as soon as its beliefs and actions can
be explained by means of its further beliefs and desires. The capacity to think
about these beliefs and desires is not required, and even less the capacity to think
about these beliefs and desires as providing the reasons in question.17 What is
required, however, is a certain minimal, subjective rationality. There are no rea-
sons explanations, be it for beliefs or actions, unless a creature’s beliefs actually to
some minimal but significant degree instantiate the basic inferential or evidential
relations objectively obtaining between the propositions that are the contents of
their beliefs.18
Now, if we think of perceptual experiences as beliefs, we can accommodate
their reason-providing role by simply extending our account of how beliefs, in
general, provide reasons for further beliefs to them. This requires some care, how-
ever, as we at the same time want to preserve certain peculiarities of the inferential
or evidential relations between experiences and other beliefs.
The two maybe most important characteristics of experiential reason pro-
viding are the following: First, the evidential relation between the contents of expe-
riences and even the most basic further beliefs based on them must be defeasible.
Experience can, and does on occasion, mislead. And when we know (or believe)
that we are in circumstances where experience is apt to be misleading, the reasons
it provides us with can be, and often are, overridden or defeated. If I know that
the light is iffy, or that there is something wrong with my eyes, I will at least be
hesitant to form beliefs about the color of nearby objects on the basis of my ex-
perience of their color. At the same time experience is such that it can, and often
does, remain completely impervious to such background beliefs about its reliability.
That’s the second characteristic feature of experiential reason providing important

17
Contrary to an anonymous referee’s objection, I do not think there is any tension between this
claim and the idea that reasons rationalize belief or action in the sense of being something that ‘speaks
for’ forming a certain belief q or performing a certain action a from the subject’s own perspective. What
we are concerned with here is first and foremost a perspective on the world, not a perspective on one’s
own mental life. What makes a reason p part of the subject S’s perspective on the world is simply the
fact that S has that reason, for instance, believes p. Then, if p is a reason for q, and S believes p, there is
something that from S’s perspective ‘speaks for’ believing q. A perspective on the world can, but does
not have to include a perspective on one’s own mental life.
That reason providing can be a purely first-order affair is not only born out by our practice of
ascribing reasons to some animals and to small children. It is also supported by the vast majority of
the reasons-explanations we give every day for the behavior of the grown-up people around us. If you
ask me why I am opening the fridge and I tell you that I am thirsty and believe there to be beer in the
fridge, my explanation is not elliptical or deficient in the sense that for it to be complete I would need to
add that I also believe that that belief and desire provide me with a reason for opening the fridge. Note,
too, that in the most basic cases an accompanying second-order belief simply cannot be required: If
believing p never rationalized believing q without also believing that believing p is a reason for believing
q, we would be off on a regress.
18
This much rationality I take to be implicit in the very content-theoretic framework employed here.
Without such rationality, reasons explanations would have no explanatory force whatsoever: Anything
could be a ‘reason’ for anything else.
 Looks, Reasons, and Experiences   85

here. Take well-known illusions such as the Müller-Lyer. Those lines look as if they
were of unequal length no matter what your background beliefs tell you about their
length. And there are many very robust and stable illusions like that. Phenomena
such as these have led philosophers to think of perception as modular, or at least as
‘belief-independent’ (Evans 1982, p. 123).
None of this can be easily accounted for by a doxastic account that construes
the contents of experiences and the contents of basic experience based beliefs as
the same. I have argued that a doxastic account of experience can both preserve
defeasibility and account for the modularity phenomena if it adopts a phenom-
enal semantics for experience. Construing experiences as having looks-contents
allows these beliefs their independence from background beliefs. By the same
token, it explains why the reasons provided by such experiences can be over-
ridden or defeated precisely by considerations concerning a situation’s illusion
inducing potential. Nevertheless, or so I have suggested, experiences do provide
strong prima facie reasons for basic perceptual beliefs: That it looks as if there was
something red in front of you is a strong, but defeasible reason for believing that
there is something red in front of you.
This is where I previously left matters. In the next section, I shall consider
whether a plausible epistemology for perception-based belief can be developed on
the basis of these ideas about the reason providing role of experience.

4.4. Experience and Justification

Experiences with looks-contents, I have claimed, provide strong prima facie

reasons for beliefs. What we want to know now is when the reasons provided
by experience are good reasons. More precisely, what we want to know is when
experience provides its subject with doxastic justification.19
In order to understand how experiences with looks-contents can not only
provide reasons for, but justify beliefs, we first need to have a closer look at the
underlying inferential or evidential relation, however. To provide a reason for
believing that p a belief needs to have a content such that there is a valid inference
(of some kind) from that content to p. For experiences with looks-contents, we
get the following schema for such inferences:

Lp
(S)
p

19
What we want to know, that is, is when a subject S has (doxastic) justification for believing that
p on the basis of her experience—as opposed to when S is (personally) justified in forming the belief
that p on the basis of her experience.
 Content Views
86   

Inferences of this kind obviously will not be deductively valid. Rather, if they are
valid, they are valid in some ‘material’ sense. We can think of inferences like these
in terms of evidential support or probabilification. Such inferences are valid if the
conditional probability of p, given that Lp, is sufficiently high. They are valid, that
is, in the sense of being reliably truth-preserving.
It is at this point that the most basic worry about experiential reason providing
kicks in. It can be put in terms of a dilemma. The first horn consists of making the
inferential connection between experience and belief content ‘too tight’: If experi-
ences and basic perceptual beliefs have identical contents, the relevant inferences
‘stutter’. Defeasibility and modularity go by the board. We avoided this by assign-
ing different contents to experiences and basic perceptual beliefs. But even though
we did this by the seemingly minimal application of the looks-operator to the
content of basic perceptual beliefs, the worry now is that the gap nevertheless is
‘too wide’. The other horn of the dilemma thus consists of losing the connection
between the validity of the inference and the rationalizing power of its premises: If
nothing but brute probabilification is required, experiential premises might no
longer ‘speak for’ forming the relevant basic beliefs from the subject’s perspective
at all. Instead of stuttering, the senses now are in danger of becoming mute.20 Let’s
call this the dilemma of stuttering and silence.
We probably ought to agree that mere probabilification never is sufficient for
justification. If p and q are logically unrelated empirical propositions, believing p
just by itself never provides a subject with a good reason for believing q. But even
if this is right, it might be possible to save experiences with looks-contents from
silence.
A first, unfortunately hopeless idea would be to ‘bridge’ the gap between Lp
and p by means of further beliefs. In order to have good reason for believing the
conclusion of a ‘material’ inference, the more general reasoning here might go,
a subject not only needs to (have good reason to) believe its premise(s), but also
needs to (have good reason to) believe some principle connecting premise(s)
and conclusion. In the case of inferences following schema (S) above, the subject
could for instance believe that inferences of that form are generally reliable, or

20
Michael Pace (2008) offers an argument like this against what he calls ‘subjectivist’ accounts of
experience, i.e., sense datum and adverbial accounts. We should, he says,
capture [the] idea that experiences serve as reasons for belief. However, it is not at all clear how
subjectivist [views] can do so. The properties of which the subjectivist says one is directly aware
perceptually, whether properties of sense-data or of mental states or of ways of being appeared
to, are not the same as the properties one believes objects to have. How then can awareness of
such properties give one a reason to believe that there is something in the world instantiating the
external properties one believes to be present? (Pace 2008, p. 656)
According to Pace, the advantage of both disjunctivist and intentionalist accounts over subjectivist
ones precisely consists in construing experience as ascribing or relating the subject to the very same
properties the relevant beliefs ascribe to the objects in question (cf. Pace 2008, p. 657). His argument
might thus generalize to a ‘phenomenal intentionalism’ such as mine.
 Looks, Reasons, and Experiences   87

that particular such inferences, for instance the inference from it looks as if there
is something red in front of me to there is something red in front of me, usually get
things right.21
This is hopeless, for at least two reasons. For one, it seems psychologically implau-
sible to ascribe belief in bridge principles to all subjects with good experiential reasons
for empirical beliefs. And having good reasons for believing bridge principles does not
seem any less cognitively demanding than actually believing them.22
More importantly, however, further belief clearly cannot bridge the gap in
any case. Rather, requiring belief (or having good reason for believing) in bridge
principles leads into the kind of infinite regress familiar from Lewis Carroll (1895).
Here is one way of illustrating this: As already observed, there is a (logical) gap
between the premise—Lp—and the conclusion—p—of the inference we are con-
cerned with: The inference is not necessarily truth preserving. We are trying to
close that gap by means of belief in a bridge principle. Now, assume that the bridge
principle is an inference schema like (S). Using such a schema to guide our infer-
ences can take two forms: either we treat the schema as admitting of exceptions,
or we treat it as to be followed in every case. If we treat it as admitting of excep-
tions, there will be a gap in the application of the schema to any particular instance
i: Why is i an instance where the schema will lead to a true conclusion? And if we
treat the schema as not admitting of exceptions, there is a gap in the justification of
the use of the schema itself: since it is not necessarily truth preserving why should
it be followed as if it were? In either case, there is a new gap—a gap of the very
same nature as the original gap—in need of closing by means of a further bridge
principle. And so on, ad infinitum.
We should therefore look for some other model on which to construe the
justificatory power of the ‘material’ inferences we are interested in. In what fol-
lows, I shall be solely concerned with perceptual justification. When it comes to
perceptual justification, it is very plausible to think of experiences as providing
reasons of a particular kind, it seems to me—“prima facie reasons” in the sense

21
Another idea would be that the subject has (to have good reason) to believe that experiences
in general justify believing that things are as they appear, or that particular kinds of experiences, for
instance experiences as of red things in front of one, do so. Justification derived from such bridge
principles would only be available to subjects capable of second order thought, however. This violates
the requirement that reasons, even good ones, be available to creatures without such capacity.
22
Ascribing looks-contents might seem cognitively too demanding, too, especially with respect to
the experiences of creatures of limited conceptual repertoire. Since some animals and small children are
supposed to have experiences just like ours, it is thus often argued that the contents of our experiences
must be simple enough to be plausibly ascribed to animals and small children. This argument, however,
might as well be turned around: If animals and small children indeed do have experiences just like ours
then we are fully justified to ascribe to these experiences contents precisely as complicated as required
for performing their characteristic role in us. That we thereby might ascribe some ‘powers’ to animal
experience that the animal itself does not make full cognitive use of does not seem objectionable. By
contrast, it would seem eminently objectionable to argue that animals and children do possess beliefs
in bridge principles. After all, it is quite clear that most of us do not have such beliefs, either.
 Content Views
88   

originally suggested by Pollock (1974). To secure the justificatory power of expe-

riential premises we should not ask for additional reasons or beliefs—rather, what
we should require is the absence of certain other (reasons for) beliefs.
Defeasibility is only a necessary condition for being a prima facie reason in
Pollock’s sense. The basic idea is the following:
[A]‌prima facie reason is a reason that by itself would be a good reason for
believing something, and would ensure justification, but may cease to be a
good reason when taken together with some additional beliefs. (Pollock 1974,
p. 40)
More specifically, a prima facie reason is a defeasible reason that is a good reason
if, and only if, there are no defeaters. Pollock himself thought that two kinds of jus-
tification are to be construed as involving prima facie reasons in this sense: justifi-
cation provided by inductive reasoning, and justification provided by perception.
Both suggestions seem plausible to me. Applying the idea to perceptual experi-
ence—as construed by the phenomenal belief account—we get:
(J) A perceptual experience with the content Lp provides its subject S with
justification for believing p iff
(i) Lp evidentially supports p to a sufficiently high degree, and
(ii) S does not have good reason to believe any defeaters.
In the relevant epistemological tradition, it is common to distinguish between two
kinds of defeaters: rebutting defeaters (Pollock called them ‘type I defeaters’) and
undercutting defeaters (‘type II’). Rebutting defeaters ‘attack’ the conclusion of the
relevant inference directly: They provide independent reasons against believing p.
Undercutting defeaters ‘attack’ the connection between premise and conclusion,
for instance by providing S with reasons for believing that circumstances are such
that inferences following schema (S) are unreliable.
When it comes to perception, examples for both kinds of defeaters are not
hard to come by. Pollock himself provides this example for a rebutting defeater:
‘Jones told me that x is not red, and Jones is generally reliable’ would be a type
I defeater for ‘x looks red to me’ as a prima facie reason for me to believe that
x is red. (Pollock 1974, p. 42)
And in the following quote he illustrates how ‘x looks red’ in the absence of any
undercutting defeaters provides good reason for believing that x is red:
Ordinarily, when I can see an object clearly, and have no reason for supposing
that there is something wrong with my eyes, or that there are strange lights
playing on the object, or anything of that sort, I unhesitatingly judge that the
object is red if it looks red to me. (Pollock 1974, p. 41)
My suggestion, then, is to construe the reasons experience provides for basic per-
ceptual belief as prima facie reasons in the Pollockian sense: They are reasons
 Looks, Reasons, and Experiences   89

that are good unless defeated. Merely having these beliefs is not sufficient for
justification, but the ‘more’ that is required is not more belief, or more reason for
belief, but rather the absence of (good) reasons strong enough to defeat them. If
experiences with Lp-contents do provide Pollockian prima facie reasons, we can
avoid the stuttering of the senses without having to mute them.
But why should we think perception is special? Why should we think percep-
tual reasons are prima facie in the Pollockian sense—when so many other reasons
are not? In particular, why should we think that the absence of defeaters is suffi-
cient to turn mere probabilification into justification in the perceptual case—when
the same combination does not seem to do the trick in other cases? That this is
indeed very plausible is, I think, best brought out by considering some examples.
While many other (defeasible) reasons a subject may have for empirical belief are
not at all plausible as candidates for being prima facie reasons in the Pollockian
sense, experiential reasons are different.23
Consider Larry: Larry has never met Paul, but believes that he is left-handed.
The only ‘reason’ Larry has for this belief is his (justified) belief that Paul is red
haired. This belief clearly is not a plausible candidate for providing him with a
Pollockian prima facie reason: Intuitively, the absence of any defeaters does
nothing to make the ‘inference’ justified. Moreover, this does not change if we
assume that there in fact, but completely unbeknownst to Larry, is a strong corre-
lation between left-handedness and red hair.
Now, let’s turn the example around: Consider Laura. Laura is just like Larry
in that she does not (have any (good) reasons to) believe anything defeating the
‘hair-hand connection’, but she nevertheless never draws any conclusions about
people’s handedness from their hair color. Laura does not in any way strike us as
odd or irrational. Quite the contrary.
But the situation is rather different when it comes to experience and basic
perceptual belief. Take John. In bright daylight, he looks at a book right in front
of him. The book is red, and nothing obstructs John’s line of sight. Nor does John
believe that there is anything wrong with his eyes, or that the surrounding condi-
tions are in any way other than they seem. He does not have any reason to believe
any of this, either. Nevertheless, John does not believe that the book is red. Talking
to him about it reveals that the book does, indeed, look red to him. Asked about
defeaters, he denies believing any of them. Nor does he have any good reason
to. Yet, he assures us ardently that he does not believe the book to be red. This is
immensely odd, and quite clearly irrational.
There is a stark contrast between John and Laura: Intuitively, Laura is per-
fectly justified in not drawing conclusions about people’s being left handed from
their being red haired, while John’s refusal to draw conclusions about the book’s

23
The following discussion was prompted by conversation with Alex Byrne.
 Content Views
90   

redness appears utterly unjustified—in fact, it appears so unjustified that we

might start wondering whether John knows what ‘red’ even means.
But even though the contrast between John and Laura indicates that percep-
tual reasons are special, one might still worry that probabilification is not what
accounts for their being good in the absence of defeaters. Rather, examples such as
John’s might seem to suggest that the validity of schema (S) inferences somehow
is a matter of conceptual or ‘linguistic’ necessity. For John’s example does make
us wonder whether John knows what ‘red’ means, and this might suggest that
schema (S) inferences are ‘analytic’, ‘meaning-constitutive’, or a priori in some
sense. Pollock himself certainly went down this road. According to him, percep-
tual reasons are defeasible and “logical” at the same time, and it is this combination
that makes them prima facie (cf. Pollock 1974, p. 40). He explains what he means
by “logical reason” as follows:

Whenever the justified belief-that-P is a good reason for one to believe

that Q, simply by virtue of the meanings of the statements that P and that
Q, we will say that the statement-that-P is a logical reason for believing the
statement-that-Q. (Pollock 1974, p. 34, emphasis added)24
While I agree that schema (S) inferences play an important and central role in
the determination of meaning and content, I think it is equally important to hold
on to the idea that these inferences are ‘contingent’. These are, after all, the ‘first’
and most basic steps in the justification of the whole of our empirical knowledge.
We must not fudge this fundamental insight by trying to cook up some unholy
empirico-conceptual bridging mix here.25
This does not mean that there must not be more to an inference from prima
facie reason than the absence of defeaters, however. It only means that there is
no alternative to using probabilification as underwriting experiential evidential
support and thus licensing schema (S) inferences in the absence of defeaters. We
do not need more for experiential prima facie reasons than what is already built
into (J):

(J) A perceptual experience with the content Lp provides its subject S with
justification for believing p iff
(i) Lp evidentially supports p to a sufficiently high degree, and
(ii) S does not have good reason to believe any defeaters.

24
Jim Pryor, another contemporary fan of experience as provider of Pollockian prima facie
reason, holds that it is a priori that experiences as of p in the absence of defeaters justify believing that
p (cf. Pryor 2000).
25
No experiential prima facie reason is such that it is a good reason for believing something
simply by virtue of the contents believed. It is good only in the absence of defeaters. Defeaters of an
eminently empirical nature, moreover.
 Looks, Reasons, and Experiences   91

The presence of a relation of evidential support (of sufficiently high degree), how-
ever, depends on the way the world actually is. If the world ‘cooperates’, schema
(S) inferences will be reliable. It is in this sense, and in this sense only, that schema
(S) reasoning is justified or warranted: In the absence of defeaters, such reasoning
is warranted by its reliability. This much ‘externalism’, it seems to me, is ultimately
unavoidable in the theory of empirical justification or warrant. Combining the
phenomenal belief account with this account of experiential justification, we can
avoid the dilemma of stuttering and silence. If the senses provide us with defea-
sible prima facie reasons, they neither stutter, nor are they mute. Whether what
they ‘say’ can be trusted ultimately remains hostage to the world. But that is how it
should be, it seems to me—it’s the human predicament.26

4.5. Superman’s Looks

In this final section, I want to take up another challenge to the phenomenal belief
account. So far, my overall claim has been that to the extent that we can provide a
plausible account of the intuitive inferential integration of experience into systems
of propositional attitudes, we thereby provide the strongest motivation available
for construing experience as having content in the first place. I have argued that
the phenomenal belief account does precisely that: By construing experiences as
beliefs with phenomenal contents it accounts for their intuitive inferential inte-
gration without either jeopardizing modularity or falling prey to the dilemma of
stuttering and silence. Silence is avoided by understanding experiential reasons as

26
It is fairly obvious that there is no (direct) anti-skeptical mileage to be gotten from the idea that
experience provides reliable prima facie reasons for (further) belief. This becomes drastically clear once
we spell things out in terms of probabilities. Plausibly, reason (or evidence) providing is governed by
the following principle (cf. Carnap 1950, pp. 382ff.; Spectre 2009, pp. 91ff.):
(EP) r is a reason for s only if Pr(s/r) ≥ Pr(s).
But now consider the following example:
(Lp) It looks as if there is something red in front of you.
(p) There is something red in front of you.
(q) There is something white in front of you that is illuminated with red light.
It clearly holds that Pr(p/Lp) ≥ Pr(p). But it holds equally clearly that Pr(q/Lp) ≥ Pr(q). Moreover,
Pr(q/Lp) clearly is greater than Pr(q), which means that it is not the case that Pr(¬q/Lp) ≥ Pr(¬q).
Consequently, that it looks as if there is something red in front of you if anything provides you with a
reason for, not against believing that there is something white in front of you that is illuminated with
red light. Of course, this reason will (normally) be much weaker than that simultaneously provided
for believing that there is something red in front of you, but nevertheless: Experiences do not provide
reasons against phenomenally compatible skeptical hypotheses.
Davidsonians might think that—despite the ‘contingent’ nature of schema (S) inferences—
general considerations of content determination might help. While I have my doubts about that, the
phenomenal belief account of experience is perfectly compatible with both a Davidsonian account of
content determination and Davidsonian epistemology in general. For more on this, see my (2012) and
Stroud (2002).
 Content Views
92   

prima facie reasons in the Pollockian sense. A phenomenal semantics for experi-
ence thus does not only not obstruct a plausible epistemology of experience-based
belief—it makes providing one quite easy.
The challenge I want to consider now is the following: Even if phenom-
enal contents allow us to get the epistemology of experience-based belief right,
experience cannot plausibly be construed as having phenomenal contents. The
looks-operator used by the phenomenal belief account is modeled on the so-called
“phenomenal” use of looks-locutions in natural language, specifically English. And
in English, phenomenal ‘looks’ is itself a propositional attitude operator. An oper-
ator modeled on English phenomenal ‘looks’ therefore cannot be used to specify
the content of any first order propositional attitude.27 Let’s call this the ‘attitude
operator argument’.
I shall argue that phenomenal ‘looks’—‘looksp’—is not a propositional atti-
tude operator. The argument will be simple in structure: Propositional attitude
operators create hyperintensional contexts. But ‘looksp’ does not. Therefore, it is
not a propositional attitude operator.
I take it that the assumption that any propositional attitude operator worth its
name creates hyperintensional contexts is uncontroversial. What is controversial
is the claim that ‘looksp’ does not. In fact, observations to the effect that perception
verbs do create such contexts have been used in the literature to defend the very
claim that perceptual experiences have contents at all, and also that they have con-
tents of a particular kind. In Searle, for instance, we find the following argument:
An additional clue that the ‘sees that’ form expresses the Intentional content
of the visual experience is that this form is intensional-with-an-s . . . . The most
obvious explanation of this . . . is that the ‘sees that’ form reports the Intentional
content of the perception. (Searle 1983, pp. 41f.)
This, however, is not the best of arguments. For one thing, intensionality is not
sufficient; (alethic) modal operators such as ‘it is necessary that’ create intensional
contexts, and in their case, the most obvious explanation is not that the ‘it is nec-
essary that’ form reports the content of any mental state. In order to have a better
argument, we should require the creation of hyperintensional contexts. But even if
‘sees that’ creates such contexts, we still would not have a good argument for the
claim that this form reports the content of perception. Arguably, ‘sees that’ implies
belief, and if it does, ‘sees that’ reports probably report not the content of experi-
ence but the content of beliefs formed on the basis of experience.
Nevertheless, a better argument might seem to be very close by: All we need to
do is replace ‘sees that’ by ‘it looksp as if ’. For those doubting the hyperintensionality

27
Arguments like this have been suggested to me by several people in conversation, first by
Susanna Siegel.
 Looks, Reasons, and Experiences   93

created, examples of substitution failure of the relevant kind seem readily avail-
able. For instance:
(1) It looks to Lana Lang as if Superman is flying by.
(2) It looks to Lana Lang as if Clark Kent is flying by.

That (1) is true in no way guarantees that (2) is. Assuming that ‘it looks as if ’ in
sentences such as (1) and (2) is used phenomenally and as a propositional attitude
operator, Brogaard has argued that the content of visual experience, just like the
content of belief, is Fregean (cf. Brogaard 2011a). While I agree with a slightly more
careful claim—that experience content, just like (all other) belief content, cannot
be modeled by possible worlds propositions alone—I do not think that ‘looksp’
creates propositional attitude contexts. Examples like (1) and (2) are misleading.
Before we look into that, however, a few words characterizing the phenomenal use
of ‘looks’ are in order.
Following Chisholm (1957) and Jackson (1977), it has become customary to
distinguish between at least the following three uses of phrases like ‘it looks as if
p’, ‘x looks F’, or ‘x looks like an F’: The ‘epistemic’ use, the ‘comparative’, and the
‘non-comparative’ or ‘phenomenal’ use. While none of these uses seems to have
any dedicated grammatical form (cf. Brogaard 2011b), they can be illustrated by
means of example. For instance,
(3) It looks as if the neighbors are away,
is most naturally used epistemically. Using (3) epistemically, the speaker says
something like that she has good reasons to believe that the neighbors are away.
These reasons can, but do not have to be visual, or even perceptual—as illus-
trated by saying ‘it looks as if Obama won the election’ after listening to the
news on the radio. On the other hand, (4) is most naturally used comparatively:
(4) The neighbor’s car looks like a tank.
Comparative uses arguably are best analyzed as existentially quantifying over
ways of looking, or over looks: Roughly, there is a way of looking such that both
the neighbor’s car and tanks have it (Byrne 2009; Brogaard 2011b). That is, the
car and the tank are compared with respect to their looks, and found to be alike.
What exactly the similarity consists in is, of course, a matter of context; it is often
suggested that the object of comparison (normally) is the way a certain kind of
object, here tanks, normally or under standard conditions look (to normal sub-
jects). But we can easily think of contexts in which the object of comparison is the
way these things look under non-standard conditions. If you are looking for red
apples on a dark summer evening, for instance, you might well use ‘that looks like
a red one’ to say that the demonstrated apple has the look that red apples on dark
summer evenings have (cf. Jackson 1977, p. 32; Chisholm 1957, p. 46).
Analyzing the comparative use of ‘looks’ along these lines, we make use of
‘looks’ in the analysans, however. To complete the analysis, we need an explanation
 Content Views
94   

of what these ways of looking are, an explanation that does not itself make (com-
parative) use of ‘looks’. It is therefore often argued that the comparative use of
‘looks’ presupposes another, third use of ‘looks’. This third use is then identified
as the non-comparative, or phenomenal use (cf. Maund 1986, p. 171; Byrne 2009,
p. 441; Brogaard 2011b). This amounts to analyzing comparative looking in terms
of comparisons between non-comparative or phenomenal looks.
The crucial question for any completion of the analysis of comparative
‘looks’ is the following: What is it that comparative uses of ‘looks’ quantify
over? What are these ways of looking that get compared when we compare the
looks of things? Intuitively, the look of an object o is a property of o that varies
with the conditions under which o is viewed. Thus, a white object can look red
if viewed in red light. And a red apple can look a certain shade of grey when
viewed on a dark summer evening. Moreover, such looks intuitively also vary
with the viewer; things look different with my glasses on or off, and the way a
red apple looks to my color blind father might well be the way it looks to me on
dark summer evenings. A very natural idea therefore is to ‘phenomenalize’ looks
along the following lines: Ways of looking are relational properties of material
objects, properties somehow involving the experiences of subjects looking at
these objects. They might, for instance, be dispositions to cause experiences of
certain phenomenal kinds under certain conditions. Or they might be properties
objects have precisely when causing experiences of certain phenomenal kinds.28
Whatever looks precisely are, the basic idea behind the claim that the compar-
ative use presupposes the phenomenal use of ‘looks’ is that what gets compared by
comparative uses are precisely such phenomenalized looks. But when using ‘looks’
comparatively, the way the object looks remains (literally) unspecified: What is

28
Martin (2010) also suggests that comparative ‘looks’ be analyzed as existentially quantifying over
looks. But according to him, looks are intrinsic properties of objects, more precisely, they are identical
to the basic visible properties such as redness and squareness (Martin 2010, pp. 161, 207ff.). Moreover,
Martin argues, most ‘looks’-statements—even those not obviously comparative on the surface such as
‘o looks red’—are to be analyzed as comparative, i.e., as comparing an object’s look to that characteristic
of a contextually relevant class of objects. According to Martin, there is no semantic reason to prefer
an account according to which looks are relational properties involving the phenomenal character
of our visual experiences of objects to what he calls “parsimony,” i.e., to identifying looks with basic
visible properties (cf. Martin 2010, p. 222). And indeed, the semantics he suggests can account for
our intuition that ‘o looks bent’, said of a straight stick halfway immersed in water, is true. According
to Martin, we here do compare the stick’s straightness to the characteristic look of bent things, i.e.,
bentness, but we do so with respect to a contextually fixed similarity measure: “The stick is similar to
bent things simply with respect to how it strikes me, or the subjective bearing it has on me” (Martin
2010, p. 215). This requires the semantics to be doubly context-dependent, however: Martin needs
both a contextually fixed comparison class, and a contextually fixed similarity measure to get the
intuitive truth values of ‘looks’-sentences right. In my (2013) I argue that there therefore is semantic
reason to prefer construing looks as relational properties involving the phenomenal character of our
experiences—because it allows for a significantly simpler semantics for ‘looks’-sentences. I also argue
that, contrary to appearances, “parsimony” has no advantage when it comes to explaining the intuition
(if it is one) that things have looks even if no-one is around to see them (cf. Martin 2010, pp. 209, 220).
 Looks, Reasons, and Experiences   95

‘said’ is that there is a way of looking that the object shares with (an)other object(s)
with a certain property (cf. Maund 1986; Byrne 2009, p. 440). In phenomenal uses
of ‘looks’, by contrast, the way an object looks gets specified, or referred to, in a
(more) direct way.29
It seems very plausible to me that we make phenomenal use of ‘looks’ in nat-
ural language. There are some difficult questions regarding our use of ‘looks’, ques-
tions like whether the phenomenon we are observing is one of genuine ambiguity
or polysemy, and how to secure a compositional semantics for phenomenal ‘looks
F’.30 For the purposes of this chapter, we can work with the understanding of phe-
nomenal ‘looks’ developed so far. I’ll assume that there is such a use in natural
language and I shall indicate it by means of ‘looksp’.
I should point out, however, that for me, nothing really hangs on this latter
claim. Should it turn out that there is no phenomenal use in natural language, I’ll
just define a phenomenal looks operator (and predicate modifier) that works in the
way indicated and use it in my semantics for experience.31 This might (or might not)
pre-empt the attitude-operator argument, but if it does, so much the better for me.
Jackson thought that the phenomenal use of ‘looks’ is restricted to, and maybe
even induced by, combining ‘looks’ with predicates of color, shape, and distance
(1977, p. 33). However, it seems plausible that sentences of the form
(5) x looks red,
can be used comparatively (and probably even epistemically). It also is very plau-
sible to think that sentences of the form
(6) x looks old,
can be used non-comparatively or phenomenally. Consider (7):
(7) x looks red and old.

29
For those of us impressed by the possibility of inverted spectra (or inverted phenomenal
qualities more generally), there is a complication here: There is no ‘direct’, intersubjectively accessible
way of specifying phenomenal kinds of experiences. We can still refer to any phenomenal kind,
however, for instance by specifying it functionally as that phenomenal kind that in the subject plays
a certain epistemic or reason-providing role. For a construal of sensation terms that could be used
as a model here, see Pagin 2000. In recent conference talks, both Maund and Pagin have suggested
analyzing phenomenal ‘looks’ along such lines.
30
Byrne (2009, p. 444) suggests in passing that ‘looksp F’ is “idiomatic in the interesting way ‘red
hair’ is.” Given the great variability of ‘F’ here, that would be bad news for the compositionality of
natural language. As hinted in the previous note, it might be possible, however, to analyze phenomenal
looks in terms of prima facie reason providing. Such an analysis might preserve the compositionality
of ‘looksp F’. My own preferred solution acknowledges that ‘looksp F’ is not compositional (in the
traditional sense). But it is not idiomatic, either. Instead, it is general compositional—which arguably
is just as good (see below, n. 39 for a tiny bit more on this. See Pagin and Westerståhl 2010a, b for the
notion of general compositionality). This is a topic for another paper, however.
31
Of course, the precise semantics for this operator needs to be worked out more precisely. But
that holds whether or not this operator exists in natural language.
 Content Views
96   

(7) is clearly well-formed, and that means that there is at least one uniform inter-
pretation of (7). To me, it seems more difficult to read (7) as uniformly compara-
tive than to read it as uniformly phenomenal. If read phenomenally, what we say
by means of (7) is that x both has the ‘red-look’ and the ‘old-look’ (cf. Byrne 2009,
p. 442; example originally from Thau 2002, p. 230).
Byrne (2009) uses observations like these to make trouble for the idea
that phenomenal looks “index” the content of experience.32 We cannot read off
experience contents from phenomenal uses of ‘looks’, Byrne argues, because
objects can lookp F without being represented as F by the subject’s experience.
Take Byrne’s own example: naked mole rats. Naked mole rats are bald, pink,
and wrinkled; they lookp old no matter how old they are. All their life, that is,
naked mole rats have the ‘old-look’. But, Byrne submits, that does not mean
that experience represents them as old. Rather, experience represents naked
mole rats as bald, pink, and wrinkled. Using “exing” as his term for the expe-
riential propositional attitude, Byrne concludes: “If a naked mole rat looksnc
old to S, then S exes, of the rat, that is wrinkled, pink, etc.—not that it is old”
(p. 443).33
Even though Byrne thinks that the possibility of using ‘looks old’ phenom-
enally thus spells trouble for the idea that ‘looksp’ “indexes” the content of visual
experience, he probably does not think that these observations ultimately pre-
vent us from construing ‘looksp’ as a propositional attitude operator: Whenever
something looksp F to a subject S, there is a p such that S exes that p. It’s just
that p does not have to be x is F. Nevertheless, I think Byrne’s observations
do provide us with some important clues here. They point us towards features
of ‘looksp’ that ultimately undermine its construal as a propositional attitude
operator.
Take some naked mole rat, for example, and call it Mora. By means of (8)
(8) Mora looksp old,
I can ascribe the old-look to Mora. But I could ascribe the very same look to Mora
by means of (9):
(9) Mora looksp bald, pink, and wrinkled.

32
Travis (2004) argues that there is no notion of looking that “indexes experience content,”
i.e., determines a unique content for a given experience. He does not consider phenomenal looking,
however, and several authors have claimed that phenomenal looks do “index” experience content (cf.
Brogaard 2011b; Schellenberg 2011). Byrne takes issue with this claim; according to him, not even
phenomenal looks “index” experience contents.
33
As already noted above (n. 12), the phenomenal belief account of experience gets around the
problem of indexing (if it is one). Phenomenal contents are indexed by phenomenal ‘looks’—rather
trivially so. Phenomenally speaking, ‘looks old’ and ‘looks bald, pink, and wrinkled’ are equivalent;
both are ways of specifying the very same phenomenal look, and, thus, the very same experience
content—according to the phenomenal belief account. This is precisely the feature of ‘looksp’ that we
shall use in what follows.
 Looks, Reasons, and Experiences   97

The ‘old-look’ and the ‘bald-pink-and-wrinkled-look’ are, in this context at least,

one and the same.34 Phenomenally speaking, that is, the predicates ‘old’ and ‘bald,
pink, and wrinkled’ here are equivalent: When modified by ‘looksp’, they ascribe
the same property to Mora. I shall say that such predicates are ‘co-phenomenal’
or ‘phenomenally equivalent’. I shall also say that the properties associated with
these predicates themselves are ‘phenomenally equivalent’. Semantically speaking,
phenomenal equivalence, or co-phenomenality, is a very strange animal indeed.
As we shall see directly, it is intriguingly independent of co-extensionality. Most
intriguing, however, is the observation that looksp-contexts are such that phenom-
enally equivalent (or co-phenomenal) expressions can be substituted salva veritate.
To better understand phenomenal equivalence, let’s consider some examples.
What the examples are supposed to illustrate is the very idea of phenomenal equiv-
alence. When, precisely, are two expressions phenomenally equivalent? For two
predicates F and G, that is the question of when the complex predicates ‘looksp F’
and ‘looksp G’ express the same property, i.e., ascribe the very same way of looking
to the object they are predicated of.35
Our first example is that of Gretria. Gretria is green and has the shape of a
triangle. We can truly describe Gretria by means of both (10) and (11):
(10) Gretria looksp triangular.
(11) Gretria looksp green.
But ‘green’ and ‘triangular’ are not co-phenomenal. Not even in worlds where eve-
rything green is triangular, and vice versa. They do not ascribe the same property,
the same lookp, when modified by ‘looksp’.
The same holds of ‘is a duck’ and ‘is a rabbit’. The duck-lookp is different from
the rabbit-lookp. Interestingly, that holds even of Dura, the duck-rabbit. Even
though Dura in a certain sense ‘has’ both looksp, Dura never simultaneously looksp
ducky and rabbity to anyone. Looksp that require a Gestalt-switch to be instan-
tiated by the same object are not the same, and the relevant predicates are not
co-phenomenal.
Necessarily co-extensional predicates, however, seem to be phenomenally
equivalent—at least, if they are “phenomenal” at all.36 Take Gretria again. In (10),
we can replace ‘triangular’ by ‘trilateral’ without changing the lookp ascribed

34
There is, of course, more than one way of lookingp old. Houses, for instance, lookp old in a
different way than mole rats do. Which way of lookingp old is the relevant, or salient, one depends on
the context. I shall abstract from this complication here.
35
On the assumption that a subject S needs to know what lookingp F is like (to S) to fully
understand the predicate ‘looksp F’, it seems reasonable to assume that if F and G are indeed
phenomenally equivalent, no rational subject understanding both ‘looksp F’ and ‘looksp G’ will assign
different truth values to ‘o looksp F’ and ‘o looksp G’.
36
It is an interesting question which predicates are what we might call “phenomenal,” i.e., are such
that there are, or can be, objects satisfying the complex predicate formed by means of combining them
with the predicate modifier ‘looksp’.
 Content Views
98   

to Gretria. But as shown by the example we started with, the example of Mora,
predicates do not have to be necessarily co-extensional for there to be contexts in
which they are phenomenally equivalent. Moreover, they do not even have to be
co-extensional. But if two predicates are phenomenally equivalent, they can be
substituted salva veritate in looksp-contexts.
Phenomenal equivalence is not restricted to predicates. Proper names can
be co-phenomenal, too. Take two qualitatively identical tomatoes, Tim and Tom.
Their names, like many others, can be used to ascribe looks to objects. Tim and
Tom are interesting, however, because the Tim-lookp clearly is the same as the
Tom-lookp. ‘Tim’ and ‘Tom’ are co-phenomenal proper names. But Tim and Tom
are different tomatoes. ‘Tim’ and ‘Tom’ do not co-refer. Consequently, ‘Tim’ and
‘Tom’ are not co-intensional, either. Nevertheless, (12) is true whenever (13) is. In
looksp-contexts, ‘Tim’ and ‘Tom’ are intersubstitutable salva veritate:
(12) It looksp as if Tim is sitting on the table in front of me.
(13) It looksp as if Tom is sitting on the table in front of me.
This is in stark contrast to (1) and (2), of course. ‘Superman’ and ‘Clark Kent’ do
co-refer, but cannot be substituted salva veritate in (1). And on the assumption that
‘Superman’ and ‘Clark Kent’ have the same intension—an assumption that I do not
share, but will not challenge here—the contrast is even starker. But now, we can explain
why ‘Superman’ and ‘Clark Kent’ cannot be substituted salva veritate in sentences
such as (1)—without having to say anything about (1) being a propositional attitude
context: ‘Superman’ and ‘Clark Kent’ cannot be substituted salva veritate in (1) be-
cause the Superman-lookp is very different from the Clark Kent-lookp: ‘Superman’
and ‘Clark Kent’, even though co-referential and co-intensional, are not only not
co-phenomenal—they are such that nothing can have both looks at the same time.37
Co-phenomenality thus is rather special when compared to co-extensionality.
More precisely, co-phenomenality and co-extensionality appear to be independent
properties. Proper names can be co-extensional and co-phenomenal, but they
can also be co-extensional without being co-phenomenal, and most importantly,
they can be co-phenomenal without even being co-extensional. The same holds
for predicates. For (phenomenal) predicates, it also seems to hold that they are
co-phenomenal if they are necessarily co-extensional. These observations suffice
for present purposes. The claim I have put forward is the following:
(LP) Co-phenomenal expressions can be substituted salva veritate in
looksp-contexts.

37
Even though Superman both can have the Superman-lookp and the Clark Kent-lookp, he cannot
have them at the same time. So, even though it is true that Superman (sometimes) looksp like Clark
Kent (and vice versa), substitution of the names within looksp-contexts not only can result in truth value
change—it always does. If it is true that Superman looksp like Clark Kent at time t, it is not true that he
looksp like Superman at t. And vice versa. Note, however, that there are plenty of non-identical looksp
that an object can have at the same time, for instance the red-lookp and the old-lookp. Substituting ‘red’
for ‘old’ in a looksp-context thus does not necessarily change truth value.
 Looks, Reasons, and Experiences   99

Together with the observations about the relation between co-phenomenality

and co-extensionality, it follows from this that looksp-contexts are not hyperin-
tensional. On the hopefully uncontroversial assumption that every propositional
attitude operator worth its name creates hyperintensional contexts, this means
that looksp is not a propositional attitude operator.38
The attitude operator argument was that the phenomenal notion of looks
cannot be used to specify the very content of visual experience because ‘looksp’
is a propositional attitude operator and thus cannot occur in the content of any
first order propositional attitude. I have argued that ‘looksp’ is not a proposi-
tional attitude operator. Nothing we have encountered in the course of these
considerations therefore prevents it from ‘going into’ the content of visual
experience.

38
In her Introduction to this volume, Berit Brogaard points out that there is an understanding of
hyperintensionality on which an operator is hyperintensional “just in case substituting an expression
for a logically (or metaphysically) equivalent expression under the operator changes the truth-value
of the whole” (p. 11). Let’s call any understanding of hyperintensionality according to which a
context is hyperintensional iff substitution of “co-intensional” expressions (in the relevant sense of
‘co-intensional’) can result in truth value change “weak hyperintensionality.” On the assumption
that ‘Superman’ and ‘Clark Kent’ are indeed co-intensional (in the relevant sense), ‘looksp’ is weakly
hyperintensional simply because substituting the one for the other in sentence (1) does result in truth
value change.
So, what did I have in mind when arguing that ‘looksp’ is not hyperintensional? Well, first of all,
I was thinking of extensional, intensional, and hyperintensional contexts along the following lines.
In extensional contexts, co-extensionality suffices for salva veritate substitutability. In intensional
contexts, it does not. What suffices instead, is co-intensionality (the relevant form of which, of course,
can be understood in different ways—e.g., as sameness of classical possible worlds intension or, on
two-dimensionalism, as sameness of secondary intension or, on evaluation switcher semantics, as
sameness of actualist intension). But co-intensionality entails co-extensionality. And analogously
for hyperintensional contexts: Here, co-hyperintensionality (whatever that amounts to) suffices for
substitutability. But co-hyperintensionality entails co-intensionality and co-extensionality.
Thought of along these lines, hyperintensionality requires more than weak hyperintensionality: it
also requires that what suffices for substitutability entails co-intensionality and co-extensionality.
Let’s call this “strong hyperintensionality.” My point then is that, while weakly hyperintensional,
looksp-contexts do not seem to be strongly hyperintensional: Looksp-contexts are such that what suffices
for substitutability is co-phenomenality. And co-phenomenality does not entail either co-intensionality
or co-extensionality. In looksp-contexts, ‘Tim’ and ‘Tom’ are substitutable, and so are ‘old’ and ‘bald,
pink, and wrinkled’ (in the context of describing Mora’s looks).
Secondly, it is precisely with respect to strong hyperintensionality that looksp-contexts
differ from paradigmatic propositional attitude contexts such as belief contexts. These are strongly
hyperintensional. Moreover, strong hyperintensionality would seem to be required for being a
propositional attitude operator. Looksp-contexts thus share a certain more superficial characteristic—
weak hyperintensionality—with propositional attitude contexts, but deeper down, it seems to
me, they are a very different kind of animal. So, whatever we ultimately think ought to be called
‘hyperintensionality’ (and Brogaard may well be right that her understanding is more standard than
what I had in mind), an operator that does not create strongly hyperintensional contexts is not a
propositional attitude operator.
Why would being a propositional attitude operator require strong hyperintensionality?
Roughly, because the use of expressions within the scope of propositional attitude operators can be
sensitive to the precise content of the attitudes ascribed by means of them. It is a consequence of such
 Content Views
100   

In this chapter, I have further developed and defended the phenomenal be-
lief account of perceptual experience I have suggested earlier. In particular, I
have defended the account against two objections against phenomenal contents:
The objection that phenomenal contents prevent us from developing a plausible
epistemology for perception based belief, and the objection that phenomenal
‘looks’ cannot go into the content of experience because it is a propositional
attitude operator. Construing experiences as beliefs with phenomenal contents
allows us to account for the intuitive inferential integration of perceptual ex-
perience into our systems of beliefs and other propositional attitudes. This in-
ferential integration in turn provides us with one of the best motivations for
construing experience as having propositional contents in the first place. By of-
fering an account of this inferential integration I have thus ipso facto defended
the claim that experience indeed has content.

References

Brewer, Bill (1999). Perception and Reason. Oxford: Clarendon Press.

Brewer, Bill (2006). Perception and content. European Journal of Philosophy, 14, 165–81.
Brogaard, Berit (2011a). Perceptual Reports and the Content of Perceptual Experience.
Manuscript.
Brogaard, Berit (2011b). Do ‘Looks’ Reports Reflect the Contents of Perception? Manuscript.
Byrne, Alex (2009). Experience and content. The Philosophical Quarterly, 59, 429–451.
Carroll, Lewis (1895). What the tortoise said to Achilles. Reprinted in Mind, 104, 691–693.
Chisholm, Roderick (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University
Press.

“content-sensitivity” that substitutions within the scope of an attitude operator that do not preserve
ascribed content can change the truth value of the ascription. And ascribed content is preserved across
substitution only if substituted and original expressions are both co-intensional and co-extensional.
It is precisely this kind of content-sensitivity that expressions within the scope of ‘looksp’ do not
exhibit. As pointed out, in looksp-contexts, co-phenomenality suffices for salva veritate substitutability,
but co-phenomenality does not require either co-intensionality or co-extensionality. It is this rather
peculiar behavior of the looksp-operator that I wanted to draw attention to. This behavior, I think,
precludes the looksp-operator from being a propositional attitude operator.
But why would ‘looksp’ behave in this peculiar way? My hypothesis right now is that ‘looksp’
is such that co-phenomenal expressions within its scope have the same meaning. For instance,
if F and G are co-phenomenal predicates, the expressions ‘looksp F’ and ‘looksp G’ mean the same.
Precisely how that works is a subject for another paper. Pace Byrne (cf. above, n. 31), I think we
should prefer the meaning of the complex looksp-expressions not to be a matter of idiomatic usage. To
achieve the desired identities of meaning, however, we cannot construe the meaning of the complex
looksp-expressions compositionally. Instead, we need to think of ‘looksp’ as what Peter Pagin and I have
called an “evaluation switcher”: an operator that switches the function used to semantically evaluate
what is in its scope. The resulting semantics won’t be compositional (in the traditional sense), but it
will be what Pagin and Westerståhl call “general compositional”—which arguably is just as good (for
more on switcher semantics and general compositionality, see Glüer and Pagin 2006, 2012; and Pagin
and Westerståhl 2010a, b, c).
 Looks, Reasons, and Experiences   101

Crane, Tim (2009). Is perception a propositional attitude? Philosophical Quarterly, 59,

452–469.
Davidson, Donald (1973). Radical interpretation. In id., Inquiries into Truth and
Interpretation (pp. 125–139). Oxford: Clarendon Press 1984.
Davidson, Donald (1974). Belief and the basis of meaning. In id., Inquiries into Truth and
Interpretation (pp. 141–154). Oxford: Clarendon Press 1984.
Davidson, Donald (1975). Thought and talk. In id., Inquiries into Truth and Interpretation
(pp. 155–170). Oxford: Clarendon Press 1984.
Davidson, Donald (2005). Truth and Predication. Cambridge, MA: Belknap Press of
Harvard University Press.
Dretske, Fred (1981). Knowledge and the Flow of Information. Cambridge, MA: MIT Press.
Dretske, Fred (2006). Perception without awareness. In Tamar Szabó Gendler & John
Hawthorne (Eds.), Perceptual Experience (pp. 147–180). Oxford: Oxford University Press.
Fodor, Jerry (2007). The revenge of the given. In Jonathan Cohen & Brian McLaughlin
(Eds.), Contemporary Debates in the Philosophy of Mind (pp. 105–116).
Oxford: Blackwell.
Glüer, Kathrin (2009). In defence of a doxastic account of experience. Mind and Language,
24, 297–373.
Glüer, Kathrin (2012). Perception and intermediaries. In Gerhard Preyer (Ed.), Davidson’s
Philosophy: A Re-appraisal (pp. 192–213). Oxford: Oxford University Press.
Glüer, Kathrin (2013). Martin on the semantics of ‘looks’. Thought, 1, 292–300.
Glüer, Kathrin, & Pagin, Peter (2006). Proper names and relational modality. Linguistics
and Philosophy, 29, 507–535.
Glüer, Kathrin, & Pagin, Peter (2012). General terms and relational modality. Nous, 46,
159–199.
Heck, Richard G. (2000). Non-conceptual content and the ‘space of reasons’. Philosophical
Review, 109, 483–523.
Heck, Richard G. (2007). Are there different kinds of content? In Jonathan Cohen & Brian
McLaughlin (Eds.), Contemporary Debates in the Philosophy of Mind (pp. 117–138).
Oxford: Blackwell.
Jackson, Frank (1977). Perception. Cambridge: Cambridge University Press.
Lewis, David (1972). Psychophysical and theoretical identifications. Australasian Journal of
Philosophy, 50, 249–258.
Lewis, David (1979). Attitudes de dicto and de se. Philosophical Review, 88, 513–543.
Martin, Michael G. F. (2010). What’s in a look? In Bence Nanay (Ed.), Perceiving the World
(pp. 160–225). Oxford: Oxford University Press.
Maund, J. Barry (1986). The phenomenal and other uses of ‘looks’. Australasian Journal of
Philosophy, 64, 170–180.
McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
McDowell, John (1998). Reply to commentators. Philosophy and Phenomenological Research,
58, 403–431.
McDowell, John (2008). Avoiding the myth of the given. In Jakob Lindgaard (Ed.), John
McDowell: Experience, Norm, and Nature (pp. 1–14). Oxford: Wiley-Blackwell.
Pagin, Peter (2000). Sensation terms. Dialectica, 54, 177–199.
Pagin, Peter, & Westerståhl, Dag (2010a). Compositionality I: Definitions and variants.
Philosophy Compass, 5, 265–282.
 Content Views
102   

Pagin, Peter, & Westerståhl, Dag (2010b). Compositionality II: Arguments and problems.
Philosophy Compass, 5, 250–264.
Pagin, Peter, & Westerståhl, Dag (2010c). Pure quotation and general compositionality.
Linguistics and Philosophy, 33, 381–415.
Pollock, John (1974). Knowledge and Justification. Princeton: Princeton University Press.
Pryor, J. (2000). The skeptic and the dogmatist. Nous, 34, 517–549.
Schellenberg, Susanna (2011). Perceptual content defended. Nous, 45, 714–750.
Spectre, Levi (2009). Knowledge Closure and Knowledge Openness: A Study of Epistemic
Closure Principles. PhD thesis. Stockholm University.
Stroud, Barry (2002). Radical interpretation and philosophical scepticism. In id.,
Understanding Human Knowledge: Philosophical Essays (pp. 177–202). Oxford: Oxford
University Press.
Thau, Michael (2002). Consciousness and Cognition. New York: Oxford University Press.
Travis, Charles (2004). The silence of the senses. Mind, 113, 57–94.
PART TWO

Against Strong Content

5

The Problem with the Content View

Mark Johnston

As philosophers of perception we need to understand the experiential condition

of the perceiving subject as it stands prior to the subject going on to judge, “on
the basis of ” experience, this or that about how things stand in the subject’s per-
ceived environment. This prior condition is increasingly well-characterized in
perceptual psychology in terms of theoretical useful distinctions between those
information-bearing representations that are genuinely perceptual—in the sense
of being constituted by operations within the sensory areas of the brain—and
those that are at least partly cognitive. So here, as elsewhere in philosophy, there is
no way around a serious study of the relevant empirical details.
However, it also seems increasingly clear that the information-bearing rep-
resentational states postulated by perceptual psychology are not the states or con-
ditions which we subjects of experience take ourselves to be in, and which we
go on to attribute to others. For one thing, the sensory states we report seem to
be individuated “broadly,” i.e., in terms of the items in the environment, which
in the good case they present or disclose, while the information-bearing rep-
resentations are by their natures the sorts of things we could enjoy even if we
were recently envatted brains, and so had no relevant environment to which such
information-bearing representations were connected. For another, our ordinary
notion of perceiving something as an F (as a dog, a boy, a policewoman . . . ) seems
to straddle the perception/cognition divide, at least as that is drawn by perceptual
psychology.
Are we just dealing with two different sorts of maps of mostly the same ter-
rain, not really in competition with one another? Or are the narrowly individuated
information-bearing representation states somehow more basic, even to the point
of providing the measuring stick against which we can determine the adequacy

The main argument of this paper was presented in a symposium on whether perception is
predicative, held at NYU in 2010. I thank Adam Pautz for his comments on that occasion, as well as
the other discussants at NYU, particularly Ned Block who pressed me on the issue of unconscious
perception, and Tyler Burge who emphasized the distinction between the individuation conditions of
an experience and its essential conditions of realization. 105
 Against Strong Content
106   

of our whole practice of thinking and talking about experience and its epistemic
significance?
In order to begin to address that pivotal question, surely one of the most
important in the philosophy of perception, we need to examine (i) just how we
frame such pre-judgmental experiential conditions in ordinary language when
we report our sensory episodes and (ii) how we appeal to such conditions, so
framed, in our folk epistemology. Only then can we enquire after the ontological
status of such conditions or states or events in the light of the details of the psy-
chology of perception. Only then will it be worth pursuing questions of emer-
gence, reduction and potential elimination of the items ostensibly recognized in
our ordinary thought and talk about experience, and most importantly in our
ordinary attempts to articulate our entitlement to believe this or that on the basis
of what we have experienced.
When we report the sensory experiences of ourselves and others, we often
resort to object-directed idioms in order to characterize datable sensory episodes.
For example:
Paul gazed at the Pantheon in amazement until the tour guide interrupted
him.
Jane briefly smelled the coffee, and then took a cup.
For two minutes, Uri watched the Rottweiler chewing the meat.
Sam listened to Sutherland’s vocal acrobatics, until he could stand them no
more.
Mary tasted the astringency of the calvados and then spat the drink out.
Suddenly, Fred’s attention was captured by the brightness of the moon’s reflec-
tion in the water.
We also report subjects perceiving things as thus and so.
Paul saw the man as waving for help.
Mary felt the texture of the curtain as rough.
Fred heard the key of the piano trio as E-minor.
Then there are propositional attitude reports of experience, for example:
Paul saw that the man was waving for help.
Mary felt that the texture the curtain was rough.
Fred heard that the key of the piano trio was E-minor.
But these last reports are curiously promiscuous. As has been widely observed
from Roderick Chisholm (1957) on, the very same propositional attitude reports
can be used to describe a variety of non-sensory ways of finding out that things
are thus and so, be it by insight, inference or by testimony. You can hear that the
president was shot, without hearing the shooting. You can see that the economy
 The Problem with the Content View   107

is crumbling without this being a form of knowledge grounded directly in visual

perception. Indeed, Paul could have come to his realization that the man was wav-
ing for help as part of an insightful reinterpretation of a crucial clue in a detective
case. Mary could have inferred that the texture was rough from being told that the
curtain was made of burlap. And Frank may be tone deaf, having simply relied on
what he was told.
Even when the propositional attitude reports in question do concern a gen-
uine sensory transaction—even when the relevant judgment that p or the state of
knowing that p is grounded in sensory disclosure of something which guarantees
that p is true—the truth of the propositional attitude report itself requires, in all
but very special cases, that the subject has already come to judge that p. But then,
whatever states of the subject that these propositional attitude reports actually
characterize, those states come too late to be the pre-judgmental states, or better
events, of sensing or perceiving the sensible features of things.
Furthermore, as the longish history of attempts to analyze the genuinely sen-
sory propositional attitude states in terms of the direct-object and “perceiving-as”
idioms attests, the proposition attitude states seem to be in part grounded in the
corresponding direct-object and “perceiving-as” attitudes (Jackson 1977; Hyslop
1983; French 2013).
To understand the pre-judgmental condition of the experiencing subject, as
we happen to frame it in ordinary language and appeal to it in our folk episte-
mology, we therefore should explore the direct-object and “perceiving-as” idioms
and investigate just what, if any, states or events described in the science of per-
ception could correspond to these reports. Only then can we properly address the
central question of the prospects of reduction or elimination of the perceptual
states and events that we attribute to ourselves and each other. And only then can
we see how far folk epistemology is to be supported or undermined by the deliver-
ances of cognitive science.
As things presently stand this is not the dominant research paradigm either
in the philosophy of perception or in epistemology. Instead, in the philosophy of
perception, a hitherto wholly unknown propositional attitude state—certainly one
which we had never previously described subjects as exhibiting—has come to play
a central theoretical role. Furthermore, it is a state whose epistemological signifi-
cance is quite obscure. The modelling of perception in terms of this state seems to
omit the distinctive epistemic contribution of perceptual experience, or so I shall
argue.
In what follows, I shall try to describe and diagnose this situation in the
philosophy of perception, in order to correct it. The main thesis is that sensory
experience, even multimodal sensory experience, is not “predicative” but rather
“presentational,” and hence cannot be properly modeled by propositional attitudes
to the effect that such and such is the case. Indeed, that model occludes the very
thing that makes sensory experience epistemically distinctive. Once we see just
how the object-directed and “perceiving-as” states make room for the distinctive
 Against Strong Content
108   

epistemic significance of perceptual experience, the newly “discovered” proposi-

tional attitude will be seen to be not only an ill-fitting, but an idle, wheel, both
internally to the theory of perception and more broadly within epistemology.
(The present essay is best read as a companion piece to “On a Neglected Epistemic
Virtue” (Johnston 2011a), which sets out in detail the account of the epistemically
distinctive contribution of experience.)
In saying that experience is presentational, I mean only (i) that objects and
quantities of stuff, their parts, their individual features (or “tropes”), and the
events and arrangements in which they are involved are the intentional objects
of our sensory episodes—i.e. they are what are presented in these sensory experi-
ences—and further (ii) that these intentional objects are presented in a variety of
ways, to which they may or may not conform. Whether or not they do conform
defines one version of the veridical/non-veridical distinction as it applies to expe-
rience. As we shall see, this is not, however, the non-illusory/illusory distinction,
since there is room for veridical hallucinations and veridical illusions, where one
gets something right, not by sensing the relevant external thing and its features,
but by what amounts to a happy accident.
The presentational conception of experience allows us to define for each sen-
sory episode, involving as it will an intentional object presented in a certain way,
a proposition whose truth or falsity correlates with the episode’s being veridical
or non-veridical. That proposition is formed by predicating the episode’s mode of
presentation of the episode’s object. It is, as I will say, a predication of the mode of
the object. There is nothing wrong with such simple predicative propositions. Their
truth and falsity does indeed correlate with the veridicality and non-veridicality of
the sensory episodes, at least on one interpretation of that distinction.
However, two points remain crucial, and will be defended in what follows.
First, experience itself is not predicative: our sensory episodes are not themselves
propositional attitudes directed at such propositions. Second, given an appropriate
understanding of the sensory episodes there seems to be no theoretical need to
postulate a novel propositional attitude directed at such propositional contents.
There is, indeed, a way the world has to be in order for a given sensory episode
to be veridical. That condition of veridicality can be captured by a proposition
that predicates the relevant manner of presentation of the intentional object pre-
sented in the episode. However, there is no evidence that our sensory systems are
monitoring these propositional conditions of veridicality of the very sensory acts
the systems make possible. The propositions certainly exist, but they are not the
content of any distinctively sensory attitude. They come into play as contents of
genuine propositional attitudes only at the level of perceptual judgment. Indeed, if
you press them into service too early in your theory of perception by treating them
as the contents of our first sensory acts, you will miss the distinctive epistemic
virtue had by most of our immediate perceptual judgments, a virtue conferred
upon them by our genuine sensory acts.
 The Problem with the Content View   109

5.1. The Content View

The now-standard content view in the theory of perception, a view that counts
a certain type of propositional attitude as central to experience, is most clearly
set out by Alex Byrne (2009) in his characteristically incisive “Experience and
Content”. One of Byrne’s ambitions there is to separate the standard view from
any naïve commitment to introspectable experiential episodes. Byrne is prepared
to doubt that there are such things, and yet still set out and argue for the standard
view of experience.
We happen to disagree on that point, since I believe that experience is best
understood in terms of attentive sensory episodes, and these are introspectable,
datable events—for example, on the basis of introspection, I can give an account
of roughly how long I visually attended to the butterfly’s sitting on my leg—but this
is not the main issue in what follows.
To explicate the standard content view, Byrne proposes that perception consti-
tutively involves a non-factive propositional attitude rather like the non-factive atti-
tude of believing: he calls it EX-ing, meant to suggest experiencing. This non-factive
propositional attitude is directed at a propositional content, which is true when the
perception in question is veridical and false when it is non-veridical. The content of
my EX-ing at any given time is a proposition which states how the scenarios before
my senses would have to be if the EX-ing in question is to be veridical. EX-ing is thus
an attitude directed at its own success conditions, at least in so far as those conditions
are conditions of veridicality. This is why EX-ing has to be non-factive, since the con-
tent EX-ed can be true or false depending on whether the EX-ing is veridical or not.
Indeed, on this model of perception, the veridicality of experience just is the truth of
the proposition EX-ed. Byrne adds:
One may think of the content of the EX-ing attitude as the output of (largely)
informationally encapsulated perceptual modules. Sometimes one will be in
possession of background information that undermines that q; that will not
affect the output, resulting in the subject EX-ing that q while disbelieving it.
CV, as just explained, is intended as a theoretically fruitful description of
the phenomenon of perception, not a piece of unarticulated folk psychology . . .
Various optional extras can be added as desired: that the relevant con-
tents are “non-conceptual”, that there’s a different attitude for each of the dif-
ferent perceptual modalities, and so on. For present purposes, though, we can
work with CV in skeletal form.
Finally, it should be emphasized that the exposition of CV is here is
entirely unoriginal, and merely repeats with minor amendments a charac-
terization that is often found in the literature. For instance, in On Clear
and Confused Ideas (p. 111), Millikan [2000] introduces “visaging”, “a gen-
eral term for what stands to perceiving as believing stands to knowing”; to
suffer a perceptual illusion is to “visage falsely”. And Johnston [1997], in a
 Against Strong Content
110   

postscript to his paper ‘How to Speak of the Colors’ (pp. 172–173), discusses
the view that visual experience involves “a sui generis propositional atti-
tude—visually entertaining a content concerning the scene before the eyes.”
As it happens, I did take the sui generis character of the putative attitude to raise
some suspicions about its existence, but Byrne allays this concern by insisting
that we should construe EX-ing as a theoretical posit in an account of just what
experience consists in. The question arises: is this a theoretical posit in percep-
tual psychology, in which case what is the detailed empirical evidence for it? Or
is it a theoretical posit that helps us to regiment our folk psychological perceptual
notions, in which case why is it useful as such? I take it to be the latter, and I take
Byrne and many others to be describing just why it is useful as such.
Here is one way of getting a feel for the content view. As I walk down the
street the scene changes, indeed there is continuous alteration in all of the sce-
narios before my senses. Some of these changes I perceptually register and these
determine the character of my whole course of experience. Think of that whole
course of experience as I walk down the street as modeled by a series of EX-ings
on my part directed at different contents in accord with the changes that I am per-
ceptually registering. As Byrne sees it, there may be no discrete or distinguished
introspectable sub-events within a course of experience, events which themselves
count as experiences in any important theoretical sense.
Byrne then goes on to offer an interesting account of the kinds of contents
that can figure as the propositions EX-ed. Following Roderick Chisholm (1957)
he distinguishes:

(i) Epistemically grounded uses of “looks” as in “It looks like our suspect is
Scandinavian, since his name is Sven Engstrom.”
(ii) Comparative uses of “looks” in which the speakers intent is to compare
the visual appearance of something to a typical appearance of a kind of
thing, as in “Oddly, in this light that Irishman looks like a Scandinavian.”
(iii) Non-comparative (and non-epistemic) uses of “looks”, such as “He
looks Scandinavian”; where this involves the speaker focusing on a
distinctive kind of visual gestalt made up of “mid-level perceptual
features” such as shape, size, motion, color, shading, texture,
orientation, timbre, loudness, pitch and the like, the very kinds of
features perceptual illusions prompt us to misattribute to objects.
Many ordinary uses of “looks red”, “looks round”, “feels soft” etc. are
non-comparative uses. But there is also such a use of “looks old” and
“looks expensive”, at least if old things and expensive things have
characteristic “looks” or visual gestalts.

Byrne argues against Susanna Siegel (2006) and others that the content of EX-ing
and hence of perception properly speaking is fairly minimal, in that the content
of EX-ings are confined to predications involving the features that can figure in
 The Problem with the Content View   111

non-comparative looks. These are roughly those “mid-level” perceptual features

manipulated in the psychologist’s repertoire of perceptual illusions, and gestalts
made up of them. So EX-ings can involve contents that predicate a certain shape,
size, color and texture of something, indeed even contents that predicate what
are in fact the characteristic visual gestalts of, say, lemons or Scandinavians. But
we do not EX contents that predicate such properties as being a lemon or being a
Scandinavian. These latter contents appear in perceptual judgment, not in percep-
tual experience per se.
That is an interesting feature of Byrne’s particular version of the content view.
Still, the content view is shared by those like Siegel who suppose that the contents
EX-ed are much richer. The content view is also common ground among those
who dispute whether the contents EX-ed are conceptual or non-conceptual, struc-
tured or unstructured, singular or general, Russellian or Fregean, Searlean (in the
sense of including a description the causal influence of the objects sensed) or not.
Most contemporary philosophers of perception can be located within this matrix
of choices, since most are implicitly or explicitly committed to the content view. To
many of them it seems a useful framework within which they can harmlessly cast
their own more particular views of perception.
That is the position I take to be quite wrong.

5.2. The Source of the Content View

Byrne’s insistence that the propositional attitude of EX-ing is intended as a theo-

retically fruitful description of the phenomenon of perception, and not as a piece
of unarticulated folk psychology, is helpful. For outside of the contemporary phil-
osophical framework, no one has ever attributed EX-ing to anyone. How did the
idea of a hitherto unknown but theoretically useful propositional attitude get
going?
The locus classicus for the EX-ing that p view is John Searle’s Intentionality
(1983); although a version of EX-ing—as possibly inhibited belief—seems to
appear much earlier in David Armstrong’s Materialist Theory of Mind (1968)
and is explicitly taken from Armstrong by Gilbert Harman in Thought (1973).
Searle writes:
The content of the visual experience, like the content of the belief, is always
equivalent to a whole proposition. Visual experience is never simply of an
object but rather it must always be that such and such is the case . . . This is an
immediate (and trivial) consequence of the fact that [visual experiences] have
conditions of satisfaction, for conditions of satisfaction are always that such
and such is the case.
Searle goes on to understand sensory experiences explicitly as propositional attitudes
directed at their success conditions or “conditions of satisfaction.” So, seeing the cat
 Against Strong Content
112   

over there is visually experiencing that the cat is over there and (Searle’s self-referential
addition) that the cat is causally responsible for the very experiencing in question.
Searle’s argument that experience has success conditions, and that we should
therefore model specific experiences as directed at their success conditions is now
widely recognized as resulting in contents too recherché for any genuinely sensory
act to have. Searle himself notes that one way an experience might fail is for it to be
a veridical hallucination: you might hallucinate a cat before you, and by accident
there might be a cat before you. Hence his thought that the content of the relevant
experience should not count this experience as a success, and so the content of the
experience should include a self-referential causal condition to the effect that there
is a cat before one causing the very experience in question. There is some plausi-
bility in the idea that perceptual content might include some idea of dependence
of experience on the external world, even perhaps a kind of causal dependence of
the sort Searle emphasizes.
Yet clearly, the self-referential causal condition is not enough to capture the suc-
cess conditions of experience; in particular it is not enough to count all cases of
hallucination as perceptual failures. There is a quite restrictive range of causal condi-
tions required to rule out hallucination, as is shown by the fact that a hallucination
of a cat before us can be caused by a cat before us. This is the old and quite general
problem of “wayward causal chains”. As will emerge later, what is implausible is that
any condition at odds with every wayward causal route to an experience is part of the
content of that experience. Accordingly, the success conditions of an experience are
not captured by the propositional content associated with that experience.
A second unconvincing argument offered by Searle for the content view is
that experience can be less than fully determinate in various ways. You might see a
speckled hen, and see it as having a lot of speckles without seeing it as having some
number n of speckles. But macroscopic reality is here fully determinate: the hen
itself has some number n of speckles. How does the merely determinable element
get into perception? The friend of the content view says that the things predicated
in the propositions that give the content of experience can be more or less deter-
minate conditions.
As is now widely recognized, this argument from the merely determinable char-
acter of experience is at most an argument for the conclusion that experience is inten-
tional in the sense of presenting items in the external environment in certain ways,
or under certain modes of presentation, where such ways or modes can be more or
less determinate. The argument does not decide between a propositional model of
intentional directness towards a not-fully- determinate content and a presentational
model of directedness to an object—the hen—under a not-fully-determinate mode
of presentation, namely as having a lot of speckles (Crane 2006, 2011).
Another influential but unsuccessful argument for the content view moves
from the fact that experiences can be illusory or non-illusory, to the claim that this
“bipolarity” of experience is best modeled by taking experience to constitutively
involve EX-ings that p.
 The Problem with the Content View   113

There is something of an embarrassment facing this line of argument for the

content view. Experiences can be evaluated along two different dimensions of suc-
cess. There is the question of whether they are veridical or not, and there is the
question of whether they are illusory/hallucinatory or not. So there are two dif-
ferent types of bipolarity exhibited by experiences. Once that is recognized, the
content view faces the embarrassment that it can only properly model one of these
dimensions of “success”.
One way to dramatize the distinction between the two different types of bi-
polarity is to dwell on the fact that many of the psychologists’ standard perceptual
illusions are not bipolar, at least along the illusory/non-illusory dimension. In fact,
there are not non-illusory variants of many standard illusory experiences. This is
because the perceptual mechanisms that generate illusions are often robust and
will iterate in situations in which the perceived environment is shaped to fit the
supposed content of the particular illusory experience in question. Nevertheless,
we can imagine subjects enjoying a presentation typical of a standard illusion in
a case where the experience in question is veridical in that it is waywardly caused

FIGURE 5.1 Clinton and Gore

Reprinted by permission from Macmillan Publishers Ltd. Pawan Sinha & Tomaso Poggio, “I think I know that
face. . .” Nature 384, 404.
 Against Strong Content
114   

by an objective array which does satisfy the manner of presentation typical of the
standard illusion. Here the experience is a veridical illusion, i.e. there is something
in the scene before the eyes which satisfies the manner of presentation character-
istic of the illusion, but that thing or individual feature is not itself perceived. (See
the case, below, of memory “filling in”.) The upshot is that either way, standard
illusions often do not have non-illusory counterparts, at least given our perceptual
systems as they actually are; but they do have veridical counterparts. This should
serve to drive home the distinction between veridicality and non-illusoriness.
To set out the point in some detail, we first need to distinguish genuine per-
ceptual illusions from cases where the subject is merely misled by what he or she is
experiencing. Here is a somewhat dated but potentially misleading presentation,
which is not a genuine perceptual illusion.
Of course, that is not a picture of Clinton and Gore, but a picture of “two
Clintons”, one with a Gore-ish haircut. If you initially saw it as a picture of Clinton
and Gore, your expectations caused you to misidentify the second figure. However,
this misidentification is not a genuinely perceptual illusion. It is an immediate
cognitive mistake, based on taking what is genuinely seen as a depiction of Clinton
and Gore. You took the figure depicted in the back to be Gore, because of mis-
leading clues and some things you genuinely saw, namely his Gore-like location at
the rear of the then President, and his Gore-ish haircut. Contrast this with genuine
perceptual illusions. Consider first:

FIGURE 5.2 The Larger Ball Illusion

Reprinted by permission from Macmillan Publishers Ltd: Scott O. Murray, Huseyin Boyaci, & Daniel Kersten,
The representation of perceived angular size in human primary visual cortex, Nature Neuroscience 9, 429–434
(2006).
 The Problem with the Content View   115

This is not a case where you simply come to wrongly believe that the ball in
front is smaller than the ball in back. It is rather that the perceptual experience
itself presents relative sizes of the balls under a visual manner of presentation that
presents the ball in back as bigger. Indeed, if you are very familiar with the tricks
of perspective drawing you might have no tendency to believe the ball in the back
is bigger; you might make no mis-identification of the ball’s relative size, but you
still see it as bigger, and you also see the visible basis for this. Similarly with the
apparent curvature of the lines in the Hering Illusion; it is not just something we
are readily led to believe. There is a visible basis for that belief.

FIGURE 5.3 The Hering Illusion

So also with this illusion:

FIGURE 5.4 The Smaller Center Illusion

 Against Strong Content
116   

So also with the sine wave illusion; where in fact the lines are all of the same
length.

FIGURE 5.5 The Sine Wave Illusion

Day RH, Stecher EJ (1991) Sine of an illusion. Perception 20;49–55

Here is perhaps the most famous genuinely perceptual illusion of all.

FIGURE 5.6 The Müller-Lyer Illusion

Now, consider the event of you attending to that array, an event which
involves the relative lengths of two lines being presented in such a way as to look
unequal in length. One question is: could an experience of that same type have
been veridical? If we individuate types of experience in terms of contents EX-ed
then it seems that an experience of that same type could have been caused, in a
non-standard way which by-passed some of the normal mechanisms of visual per-
ception, by an array which is just the way the array above appears to be. But even
so any such veridical version of the experience type in question would be illusory
in the sense that the subject would not be taking in the relevant visible features
of the array; she would not be aware of them as a result of the causal mechanisms
which typically operate in ordinary perception. For example, the visual system’s
 The Problem with the Content View   117

standard reliance on shape cues would have to be overridden or “tricked” in some

way.
If this kind of overriding or jiggery-pokery does not occur, the original
illusion will persist. Suppose that the actual measure of the proportions of the
apparent lengths given in your experience has the left line being 150 percent
longer than the right line. Suppose we now alter the display to make that true.
Oops. Now the actual measure of the lengths given in your experience will
be that the left line is, say, 225 percent longer than the right line. That is, you
would see the altered display as involving an even greater difference between
the lengths of the lines. This is a quite general problem. Many illusions iterate,
so that they have no non-illusory counterpart.
For example, what is the non-illusory version of your experience in the Ponzo
Illusion?

FIGURE 5.7

Imagine that you call out the details of the display to your artist friend, who
follows your excellent directions exactly. You then look at the array he has pro-
duced and say “No. No. That is not it; the right vertical line is too long. Let us try
again.” No matter how many times you try, the right vertical line in the artist’s
model looks too long to match your original experience.
Your artist friend will be frustrated in just the same way in the case of the
Larger Ball Illusion, the Hering Illusion, the Smaller Center Illusion and the
Müller-Lyer Illusion. The perceptual mechanisms which generate these illusions
are such that their operations are constitutive of the particular acts of seeing the
scenes in question. That is why there are no non-illusory versions of such illusions,
at least absent serious rewiring of the human visual system.
 Against Strong Content
118   

However, as with the Müller-Lyer illusion, there could be veridical versions

of such illusions, at least as veridicality is modeled on the content view. This is
easiest to recognize in those versions of the content view that treat perception as
merely imposing general conditions on the scene before the eyes. So in the case
of the Müller-Lyer illusion, the relevant part of the general content might be that
that there is an array before one in which the left line is considerably longer than
the right line, and that the lines are “feathered” and “arrowheaded” in the relevant
ways. Now suppose I am not in fact seeing the lines in question, but hallucinating
a Müller-Lyer array. It could also be the case that in the very scene before me there
is, unbeknownst to me, an array in which the left line is appropriately longer than
the right line, and in which the lines are “feathered” and “arrowheaded” in the
relevant ways.
This is a case of veridical hallucination. There could also be veridical illu-
sions of a related sort. These would be cases in which we manage to see the lines
in question, but not their differences in length; instead the apparent differences in
length, though they by accident turn out to be veridical appearances, are merely
“internally generated” by the visual system in a way that is unresponsive to the
scene before the eyes.
What does this show? Minimally, it establishes that even if the content
view does successfully model the veridical/non-veridical distinction in terms
of the truth or falsehood of the proposition EX-ed, the content view does not
thereby correctly map the non-illusory/illusory distinction. Indeed, as al-
ready noted in the discussion of Searle’s version of the content view, the model
cannot map this second distinction without implausibly importing into the
proposition EX-ed something that is in effect a characterization of that very
distinction.
The veridical/non-veridical distinction—when it is not just some version of
the non-illusory/illusory distinction—does not get to the heart of what is at issue
in perceptual experience. The issue is not just whether a perceptual experience
puts you in a position to believe true propositions, what is also at issue is whether
the perceptual experience presents, or puts you in touch with, the worldly items—
in the Müller-Lyer case, the two lines, and their respective individual lengths—in
a way which entitles you to believe the propositions in question, the propositions
supposedly EX-ed.
By collapsing the two distinctions, or by in effect ignoring the second
distinction, the content theorist fails to model one of perception’s most dis-
tinctive features, one which potentially bears on the distinctive epistemolog-
ical contribution of perceptual experience itself, and thus helps to explain why
we are better placed epistemically than a perfectly reliable blind-“sighter” or
deaf-“hearer”, i.e., subjects who lack the relevant perceptual experience without
having any resultant defect in the reliability of their immediate perceptual
judgments.
 The Problem with the Content View   119

5.3. Singular Propositions to the Rescue?

One response on behalf of the content theorist is that resorting to genuinely singular
propositional contents can capture the still somewhat metaphorical idea of percep-
tual experience presenting, or putting us in touch with, the worldly items that entitle
us to believe the propositions we do about the scenarios before the senses.
So consider a phenomenally seamless transition from your hallucinating three
lights on in a ceiling, say as a result of the direct stimulation of your visual cortex,
to your then seeing each of the three lights on in the ceiling, say as a result of the slow
diminution of the direct stimulation to appropriate parts of your visual cortex,
while at the same time the lights in the ceiling are being turned up. At the begin-
ning of the seamless transition you were not “perceptually in touch with” the lights
on in the ceiling, at the end of it you were. One attempt to capture this difference
and thereby discharge the metaphor of presentation or perceptual contact is to
say that only at the end did you genuinely come to see those three lights on in the
ceiling. But this is not a report of a propositional attitude. How then does the con-
tent theorist model this transition in terms of a difference in the contents EX-ed at
the beginning and the end of the seamless transition?
One kind of content theorist, who supposes that the content of perceptual ex-
perience is always general (but for, perhaps, indexical reference to the subject and
the time of the experience), will not even attempt this, but instead will emphasize a
certain regular causal connection between the three lights on in the ceiling and the
EX-ing of the content that there are now before me three lights on in the ceiling.
If things are left there then the theorist is making nothing of the idea that our per-
ceptual experience itself puts us in touch with worldly items in a way that confers
a distinctive entitlement on what we typically go on to immediately believe on the
basis of that experience. For then, the seamless transition is not a transition in the
content EX-ed, and hence it is not a transition in perceptual experience itself, as
opposed to its external causes. It will then remain unclear why we are better placed
epistemically than a very reliable veridical hallucinator, who is never in experien-
tial contact with the relevant items in the scene before him.
Fortunately the content view is not, as such, committed to this kind of defla-
tion of the significance of non-hallucinatory experience. The content view can take
a different form. It can allow that the content EX-ed at the beginning of the seamless
transition is merely a general existential content to the effect that there are three lights
presently illuminated in a ceiling before one, while the content EX-ed at the end of the
transition is the singular content that those lights, that one, that one, and that one are
presently illuminated in the ceiling before one. In modelling that second content, the
lights themselves and the ceiling itself will be constituents of the proposition EX-ed.1

1
Perhaps they figure in the proposition under modes of presentation, in which case the singular
proposition will be in one way Fregean—in that it involves “de re senses”—rather than purely Russellian;
but this distinction does not matter for present purposes.
 Against Strong Content
120   

The resultant state of EX-ing may thus deserve the name of a distinctive form of con-
tact with the worldly items in question; for they enter into the individuation condi-
tions of the proposition that is EX-ed. The condition on the scene before the eyes
imposed by the experience so modeled is that those very lights be illuminated in that
very ceiling. Now the seamless transition from hallucination to genuinely successful
perception is modeled by a transition in the contents EX-ed. It can thus be claimed
to be a transition in what is experienced, and not just in how a given experience was
caused. What is experienced in successful perception thus goes beyond what is ex-
perienced in the corresponding hallucination precisely in respect of including the
worldly items actually seen or sensed.
The thing to notice is that even if this is an advance on the part of the content
theorist, one which may deserve the name of modelling how perceptual experi-
ence discloses, presents or “puts us in touch” with worldly items (the three lights,
the ceiling), the model does not explain how this kind of contact could distinc-
tively entitle one to go on to judge that there are three lights illuminated in the
ceiling. Of course, the belief this judgment lays down may be reliably formed, and
there may be nothing else one believes, or should believe, that counts against it.
But we should be holding out for a view on which the perceptual experience itself
plays a distinctive epistemic role in relation to immediate perceptual judgment.
Otherwise, sensory experience will end up being modeled as just a sensuous “light
show” merely accompanying the real epistemic transaction, as that transaction
is understood by the reliablist or, alternatively, by the theorist of prima facie jus-
tification. (For more on what is wrong about those pictures of the epistemic role
of experience see “On a Distinctive Epistemic Virtue” and the discussion of the
Wallpaper View in Johnston (2006) “Better than Mere Knowledge”.)
At the end of the seamless transition in question, the crucial thing is that you
are perceptually aware of the individual arrangement of those three lights in that
ceiling; your perceptual experience presents or discloses that individual arrange-
ment or “trope”-like entity to you. If you go on to judge that those lights are so
arranged in the ceiling then what you judged has what you experienced as its truth-
maker. The predicative structure of your judgment matches the complex exemplifi-
cation of features you experienced. Your experiencing the complex exemplifications
was not your enjoying some propositional attitude; to represent it as such simply
leeches out the distinctive epistemic contribution of experience.
There is a general lesson emerging here, one which will reappear in what fol-
lows. The content theorist who favors Russellian content has still failed to model
perceptual experience as putting us in touch with the worldly items in a way that
entitles us to immediately believe the propositions we do about the scenarios before
the senses. To the extent that the content theorist incorporates the worldly items
sensed into the propositional content EX-ed, he or she may capture the idea that
those items genuinely figure in the experience, but the epistemic significance of
the items figuring in the experience is then lost. This is because experience of the
 The Problem with the Content View   121

items in question is not correctly modeled by those items simply being subjects of
predication.
In successful perceptual experience we are aware of exemplifications of per-
ceptual features, and not simply of predicative relations among items and features.
The exemplifications can play an epistemicially significant role that no mere pred-
ication of perceptual features of perceived items can play. For the exemplifications
are the truthmakers of the corresponding predications, and hence of what we go
on to immediately judge. Once we recognize this, there will be no need to theo-
retically posit a propositional attitude like EX-ing that p. Perceptual experience is
directed at exemplifications (under certain modes of presentation), or in the case
of illusion and hallucination, merely ostensible exemplifications (under certain
modes of presentation).2
In contrast, perceptual judgment and the beliefs laid down by perceptual
judgment do involve predications of perceptible features of perceptible objects.
Perceptual experience is epistemically relevant not because it is a “prehearsal” in
the sensory system of such predications, but because it discloses truthmakers for
what we immediately judge on the basis of perception.3
(It should be noted that this claim does not involve a commitment to the so
called truthmaker axiom to the effect that there are no truths without truthmak-
ers, any more than the mention of tropes commits one to trope nominalism as a
solution to the problem of universals.4)

5.4. Disclosure of Truthmakers

Another example may serve to further illustrate the idea of successful perception
as the disclosure of truthmakers. Consider a seamless transition from a case where
memory fills in a feature that is not in fact sensed, to a case where the feature is
genuinely sensed. Perhaps memory fills in low-light vision, so that in the gloaming
or twilight we have a more vivid appearance of the colors of just those objects that
are familiar to us. So imagine you are looking into the backs of cars when dusk has
settled. (Don’t ask!) With unfamiliar cars you see the leathers in shades of grey.
But with your own car, memory fills vision in, and you see your leather not in the
grayscale but in the red scale. You don’t however see the redness of the leather.
Memory’s function here is, we may suppose, not part of the normal functioning
of the visual system itself, and so it is not a way of seeing the individual colors
of things. So when things work like this we have a case of veridical illusion; you

2
In “The Obscure Object of Hallucination” (2004) I called such ostensible exemplifications
“sensible profiles”. There I failed to emphasize that even such ostensible exemplifications are given
under modes of presentation.
3
For a similar view, derived in part from Edmund Husserl, see Kevin Mulligan 1995.
4
For a general theory of tropes and of trope nominalism as a solution to the problem of universals,
see Johnston (1983) Particulars and Persistence (Princeton University, Ph.D.)
 Against Strong Content
122   

don’t see the redness of the leather, you are only aware of an expanse of redness,
which is not the individual redness of the leather. But then there is a transition,
not detectible by you; either the discriminative capacity of your rods is suddenly
augmented, or a discrete unnoticed light goes on in the car, so that you then do see
the redness of the leather.
The thought is that this transition, just like the seamless transition from hal-
lucinating lights to seeing lights, is potentially epistemically significant in that, at
the end, you have one sort of exploitable entitlement to judge that the leather is red,
one that you did not have at the beginning. At the beginning, you may well have
been prima facie justified in believing that the leather is red, in that it seemed to
you to be that way, and no other belief you had or should have had was at odds
with this. Moreover, memory’s filling in perception during the gloaming may even
be a reliable process, so that even at the beginning, your immediate belief that the
leather is red was reliably formed. Still, at the end, but not at the beginning, you
were visually aware of the redness of the leather, and this itself provides an exploit-
able entitlement to go on to judge that the leather is red. (As to just how we exploit
the entitlement, see “On a Neglected Epistemic Virtue”.)
The source of that entitlement is entirely occluded if we model your percep-
tual experience at the end—that is, your awareness of the redness of the leather—
as a propositional attitude with the content that the leather is red. For once again
this would trade in your awareness of an exemplification for the having of a prop-
ositional attitude directed upon the corresponding predication. Thereby, we lose
precisely the distinctive sort of thing which successful perception discloses or
presents.
Notice that that the source of the entitlement is also occluded even if we go for
more arcane propositions as the contents that are EX-ed. There is no good reason
why tropes like the redness of the leather cannot be themselves subjects of predi-
cation. We can predicate existence of them, we can predicate location of them, we
can make comparisons concerning them, and so on and so forth. But the proposi-
tion that the redness of the leather exists is not what makes that proposition true,
it is the redness of the leather that does this, and that is why enjoying a perceptual
experience that discloses the redness of the leather is epistemically relevant to the
belief that the redness of the leather exists.

5.5. Searlean Content to the Rescue?

It is at this juncture that I seem to hear John Searle in the wings, saying “I told you
so!” (Sadly, it is, as far as I can tell, only a hallucination on my part.) For Searle’s
original version of the content view was precisely designed to close the gap be-
tween an experience’s being veridical in the sense of saying something true about
the subject’s environment and its being non-hallucinatory and even non-illusory.
Once we left the ambition to close that gap behind, we opened up space for the
 The Problem with the Content View   123

kind of cases I have been discussing, namely veridical hallucination and veridical
illusion, where the non-Searlean content theorist seems unable to properly locate
the relevant defects at the level of the content EX-ed, and so cannot explain why
the transitions in question to the “fully successful” perceptual experiences might
be improvements from the epistemic point of view.
(Remember, we are not trying to answer skepticism here, so we are not
making the mistake, often execrated by Searle, of trying to condition our account
of perception by the need to answer skepticism. We are instead trying to explain
how, skepticism aside, our being fully successful experiencers, in particular our
not being merely reliable victims of hallucination or illusion, could confer upon
us some distinctive epistemic advantage. It is a bit of a scandal if we have nothing
interesting to say about that.)
So we should look again at Searle’s own version of the content view, as it
applies to the case of perceptual illusion in which memory fills in colors. In that
case, there is a transition from seeing the leather but not the redness of the leather,
to seeing the redness of the leather, thanks to augmented rods or better lighting.
Now Searle has a general argument that a perceptual experience involves more
than characterizing its target in terms of how it is intrinsically and in terms of its
relations to the objects around it. Perceptual experiences also characterize their
targets as causing the very experiences in question. Searle motivates this by way of
a contrast between the phenomenology of genuine perception and what he plau-
sibly supposes might be the phenomenology of a perfect eidetic imager. Searle
holds that such a person would not have the sense of being passively under the
influence of the target, as we are in standard cases of perception. The passivity of
perception is part of the experience of perception itself, and Searle proposes to
capture this fact by supposing that the content of perception mirrors this. If one
is seeing a yellow station wagon over there then the content of the perception is
not just that there is a yellow station wagon over there but also that it is causing
the very experience in question. In this way, Searle hopes to close the gap between
perceptually entertaining a veridical content and having a fully successful percep-
tual experience.
So in the case where memory is filling in, the content to the effect that the
redness of the leather is (in part) causing the very experience in question might be
taken to be false at the beginning, while at the end of the transition the content to
the effect that the redness of the leather is (in part) causing the very experience then
in question is true. Is that why one is better placed epistemically at the end of the
transition?
Several remarks are in order here. In the case at hand, memory “fills in red-
ness” on the basis of remembering the redness of the leather. On any reasonable
account of memory, the state of remembering the redness of the leather is (in part)
caused by the redness of the leather, and so given that, in this kind of case, a cause
of a cause of a thing is also going to be a cause of the thing it follows that, con-
trary to the suggestion above, the content that the redness of the leather is causing
 Against Strong Content
124   

this very experience will be true during the experience that involves memory
filling in. This is in effect the observation that, as stated, Searle’s account of the
self-referential causal content of experience does not in fact close the gap between
veridical experience and fully successful experience. Precisely because of wayward
causal chain cases like memory’s filling in, we have to add that the causation be
of the right sort, namely of the sort that is typical of fully successful perception.
But this addition to the content of each perception is implausible; the need for
“causation of the right sort” was discovered by sophisticated thinkers late in the
philosophical day—I suppose in the early 1960s—in response to proposed causal
analyses of action and perception, and not by attending to how the world seems to
be in perceptual experience. In any case, the required patch-up for Searlean content
goes far beyond what Searle motivates by way of noting the sense of passivity—the
sense of being under the causal influence of the target—in sensory experience.5
However, let us put that somewhat familiar objection aside for now. Suppose
we allow that seeing the over-there-ness of a yellow station wagon and collaterally
sensing its causing the very experience in question reduces to having a perceptual
experience to the effect that there is a yellow station wagon over there and that the
yellow station wagon is causing the very experience in the right way. It still seems
that this version of the content view makes a form of the mistake that I earlier
suggested was characteristic of the content view. It begins with objectual attitudes
directed at “tropes” or individual characteristics and it translates those into prop-
ositional attitudes directed at the corresponding predications, i.e. the predication
of a certain location of the car and the predication of causing experience of the car.
These are not only very different things that are being assimilated by the trans-
lation, but the translation again occludes what is epistemically interesting about
perceptual experience, namely that it involves the disclosure of truthmakers for
what we are inclined to go on to judge. It is one’s seeing the relative location of the
yellow station wagon that provides an exploitable entitlement to judge that the
station wagon is over there. It is one’s experiencing the yellow station wagon as
producing our experience that provides an exploitable entitlement to judge that
the yellow station wagon is producing one’s experience.
So the most revealing objection to Searle’s distinctive version of the content
view is just that is it a version of the content view, and thereby occludes the epi-
stemic significance of perceptual experience; that is, its capacity to disclose truth-
makers for what we typically go on to immediately judge.

5.6. The Same Issue Arises for Friends of Narrow Content

So far, in arguing that the content view occludes the significance of the percep-
tual presentation of truthmakers, we have focused on cases where the truthmakers

5
A similar point is made by Millar (1985).
 The Problem with the Content View   125

are individual characteristics of, or individual arrangements of, external items.

However, the central point about the epistemic significance of the deliverances
of perceptual experience does not depend on the supposition—in my view a cor-
rect supposition—that in perceptual experience we are typically aware of external
items, items to be met with in public space. The same point can be made even in a
case of an experience where we are not aware of an item to be met with in public
space. Accordingly, the point does not depend on the supposition that perceptual
experience is always “broadly individuated”; that is, always partly individuated in
terms of some external item perceived.
It is a notable fact about the visual system—one for which there is no counter-
part in taste or touch, for example—that it causes a “default” presentation of color
when there is no retinal stimulation. If you close your eyes firmly, you become
aware of an expanse of a certain very dark shade. If you now put your hands tightly
over your closed eyes, you become aware of an expanse of an even darker shade.
If you were to succeed in excluding all photons from your eyes then you would
become aware of an expanse of one of the determinate shades which sensory psy-
chologists refer to by the determinable “Brain Grey”. This shade is the default pres-
entation of your visual system in the sense that when that system is unstimulated
you are aware of an expanse of that shade.
Suppose you succeed in excluding all photons from your eyes, then you will
be aware of an expanse of a determinate of the determinable Brain Grey, call it
“Brain Grey*”. Notice that in this case no external item is presented as exempli-
fying the quality Brain Grey*. It is rather that your visual field is filled with an
expanse of Brain Grey*. (Or so it seems.)
The content theorist will model this experience as your EX-ing that you visual
field is filled with an expanse of Brain Grey*. Suppose you now go on to believe that
your visual field is filled with an expanse of Brain Grey*. As the content theorist
models things, you may have a prima facie justification for believing that that your
visual field is filled with an expanse of Brain Grey*, a justification which derives
from (i) the supposed fact that your EX-ing that p is one way in which it can seem
to you that p and (ii) the principle that if it seems to you that p and nothing you
believe or should believe counts against p then you are prima facie justified in
believing that p. Moreover, the step from this case of EX-ing to the corresponding
belief may be a step of a very reliable type. But once again these facts that the con-
tent theorist is in a position to exploit seem to significantly under-describe your
epistemic situation with respect to the judgment or the belief that that your visual
field is filled with an expanse of Brain Grey*.
What more is to be said about your epistemic situation? Your perceptual
experience just is your being visually aware of an expanse of Brain Grey*
filling your visual field. This objectual attitude is directed at a truthmaker—
the expanse of Brain Grey* filling your visual field—for the corresponding
belief or judgment that there is an expanse of Brain Grey* filling your visual
field. Your perceptual experience thus puts you in a position to exploit a
 Against Strong Content
126   

conclusive entitlement to believe or judge that that your visual field is filled
with an expanse of Brain Grey*.
The same considerations go through even for those of us who believe that
there are no visual fields (Johnston 2011b). In that case, what one is aware of in
being aware of Brain Grey* is an expanse of Brain Grey* appearing before one.
The content theorist will model this as an EX-ing of a proposition to the effect that
an expanse of Brain Grey* is appearing before one. But once again this obscures
the contribution of perceptual experience. The perceptual experience in question
was directed at a truthmaker for that proposition. As such it puts the subject of the
experience in a position to exploit a conclusive entitlement to believe or judge that
an expanse of Brain Grey* is appearing before him or her.
(These same points could be made with after-images, and with expanses of
“film color”, a neglected but extremely interesting phenomenon, of considerable
import in the philosophy of perception.)

5.7. Modifying the Content View

So far we have seen that the content view (i) mischaracterizes the success conditions
of experience, and (ii) occludes the distinctive epistemic significance of experience
by modelling it terms of a propositional rather than an objectual attitude.
If we simplify and treat hallucination as a certain kind of extreme illusion,
namely the kind in which there is an illusion of a particular individual, event or
quantity of stuff being presented to us, then we can say that experience is bipolar
in that it does come in two forms or types, namely illusory and non-illusory
types. But this “bipolarity” of experience is not captured by the truth or falsity
of propositions EX-ed. That dimension of contrast merely serves to highlight
the veridicality or non-veridicality of experience, and as we have seen there can
be veridical illusions, i.e. illusions in which we are not presented with the truth-
makers of what experience naturally prompts us to judge, even though those
truthmakers are there to be experienced in the scenarios before the senses, but
are nonetheless missed because of perception failing us, as in the case of mem-
ory filling in. The lesson is that if we are to characterize the illusory/non-illusory
distinction, or more generally the hallucinatory cum illusory/non-illusory cum
non-hallucinatory distinction, we need to recognize that fully successful percep-
tual experience involves objectual attitudes directed at truthmakers.
This is the same conclusion that is made plausible by the consideration that
the content view systematically mischaracterizes the kind of entitlement that per-
ceptual experience puts us in a position to exploit. That kind of entitlement is
typically conclusive, at least for the simple immediate perceptual judgments that
we habitually make.
The two routes to the same conclusion are obviously related. For if you mis-
characterize the illusory/non-illusory distinction as the non-veridical/veridical
 The Problem with the Content View   127

distinction you will as a result fail to appreciate just what the epistemic import of
perceptual experience is.
Given this description of the terrain, those trying to play out the argumentative
chess game on behalf of the content theorist might now make the following move:
Right; we content theorists will now accept there are two ways in which per-
ceptual experience can go wrong. It can be non-veridical and it can be illusory.
The existence of veridical illusions and veridical hallucinations shows that, but
the proper response to this fact can be accommodated within a modified con-
tent view.
First Attempt (Conjunctivism): EX-ing that p, we forgot to tell you, comes
in two forms, namely luding that p and the more demanding form of percep-
tually taking in the fact that p. A case of luding that p is a mere case of EX-ing
that p. It is veridical just when p obtains, but even when luding is veridical it
falls short of what is involved in perceptually taking in the fact that p. Your
perceptually taking in the fact that p involves your EX-ing that p being caused
in the right way by a truthmaker of p there in your perceived environment.
Second attempt (Disjunctivism): “EX-ing that p”, we forgot to tell you,
is an overarching term for two quite different kinds of states, namely lud-
ing that p and perceptually taking in the fact that p. Beware: states of the
second are not special variants on the states that make up ludings; they are
not appropriately caused ludings, they are sui generis states. A case of luding
that p is veridical just when p obtains, but even when it is veridical, luding
falls short of what is involved in perceptually taking in the fact that p. Your
perceptually taking in the fact that p involves the underlying functioning of
your visual system being caused in the right way by a truthmaker of p there
in your perceived environment, but it is more than that. It involves the pres-
entation of the fact that p.
Now we can explain the distinctive epistemic significance of “fully suc-
cessful” veridical experience, i.e. perceptually taking in the fact that p. If
you perceptually take in the fact that p, you are aware of a truthmaker for
what you go on to immediately judge on the basis of perception. You have
an exploitable entitlement to judge that p, one that differs from the reli-
ability of the process leading to that judgment and from any prima facie
justification that attends that judgment.
A number of remarks need to be made here. First, the conjunctivist version of the
revised content view does not in fact model perceptual experience as putting us in
touch with a truthmaker for p; at best it models experience as relating us to true
propositions, and not to their sensed truthmakers. On the conjunctivist version,
those truthmakers come in simply as the causes that distinguish the good case of
fully successful perception from mere true luding that p.
It is the disjunctivist version, which is reminiscent of John McDowell’s (1996,
1998) account of perception, which delivers perceptual awareness of truthmakers
 Against Strong Content
128   

and thereby secures the distinctive epistemic significance of perceptual experi-

ence. But it achieves this within the attitude framework only by working with
the idea of perceptually disclosed facts—things having properties and standing
in relations. That is, instead of working with the very objectual attitudes that we
ordinarily report, such things as
Seeing the redness of the leather
the disjunctivist re-figuring of the content view works with such postulated factive
attitudes as
Visually taking in the fact that the leather is red.

5.8. Do We Perceive Facts?

As a result of recent work in ontology and philosophy of language, we are in a

better position to evaluate what is now the pertinent question: Is the perceived
environment, a totality of facts (where these are conceived of as things having
properties and standing in relations) or a particular arrangement of objects,
quantities of various sorts of stuff, events and exemplifications (tropes)?
For one thing, it is increasingly clear that to do full justice to the relevant lin-
guistic and ontological phenomena we need to distinguish colors like red from the
corresponding properties, such being red or having red as your color. Moreover,
neither red nor the property of being red should be confused with redness, which
is best understood as class of “tropes” or exemplifications of redness (Johnston
2007; Moltmann 2013). The property of being red, as with any property under-
stood as the designata of canonical property designators such as “being F” or “the
property of being F”, admits of a pleonastic treatment, and hence so does the fact of
something’s being red, and the fact that something has the property of being red.
That is to say, we would lose no distinctive descriptive power from our language if
we ceased to talk about such properties and facts. But red and cases of redness do
not admit of a pleonastic treatment; they remain as genuine ingredients of reality,
and not just as semantic values for certain forms of words that we could drop from
our language without any real loss of the language’s descriptive powers. Once these
considerations are set out in detail, I believe it will become obvious that it is not the
fact of the leather’s being red that falls within the ambit of experience, but a quite
different thing, namely the redness of the leather, which is a particular exemplifi-
cation of redness. (Johnston, n.d.)
That, however, is an argument for another time. The more direct point to
be pressed here is that the disjunctivist’s postulation of still another psycholog-
ical attitude hidden from us until now is simply unnecessary rigmarole. (Notice
the claim that there is such an attitude is not intended as a claim in perceptual
psychology.)
 The Problem with the Content View   129

It is unnecessary rigmarole for two reasons. First, we already do report objec-

tual perceptual attitudes directed at the exemplifications that are the truthmak-
ers. Indeed, this is one dominant form of perceptual reports. And once we see
that, then even if we postulate such further states as the disjunctivist’s perceptually
taking in the fact that p those postulated states will appear to be at least partly
grounded in the familiar objectual perceptual attitudes directed at truthmakers.
A second worry attends the idea that perception is a taking in of facts. Facts
are too etiolated to admit of many of the comparisons which perception immedi-
ately puts us in a position to make. (It seems to me likely that this has to do with
the pleonastic character of facts.)

5.9. Attentive Sensory Episodes

Recall these reports:

Paul gazed at the Pantheon in amazement, until the tour guide interrupted
him.
Jane briefly smelled the coffee, and then took a cup.
For two minutes Uri watched the Rottweiler chewing the meat.
Sam listened to Sutherland’s vocal acrobatics, until he could stand them
no more.
Mary tasted the astringency of the calvados and then spat the drink out.
Suddenly, Fred’s attention was captured by the brightness of the moon’s
reflection in the water.
These reports describe episodes, things that take time. They are naturally con-
strued as reports of episodes in which the subject of the experience is attending
to a certain item in his or her environment. The items attended to are respectively
an object, some stuff, an event, an ensemble of events, an individual characteristic
or trope, and another individual characteristic or trope. And it is natural to un-
derstand such episodes not only as presenting certain items but also as presenting
them in a certain specific ways, ways which allow for experience (i) being less than
fully determinate, as in the case of an experience of a speckled hen, or (ii) illusory,
as in the case of the pencil that appears bent when half submerged in water.
Such attentive sensory episodes—hereafter ASEs—are individuated by their
time of occurrence, their subjects, their intentional objects and their manners of
presentation of their intentional objects. Token ASEs always present what they
present; if they are not hallucinatory or illusory then they are never so, and could
not become so. Token ASEs are non-veridical if and only if their manners of pres-
entation mischaracterize their objects; they are veridical if and only if their man-
ners of presentation correctly characterize their objects. But then each such token
ASE is unipolar: if it is veridical it is by its nature so, i.e. it is guaranteed to be so in
virtue of the relevant relation of matching holding among two of those things—the
 Against Strong Content
130   

episode’s intentional object and the manner of presentation of that object—which

enter into its individuation conditions. Likewise, if a token ASE is non-veridical
it is by its nature so, i.e. guaranteed to be so in virtue of the relevant relation of
mismatching holding among two of those things—the episode’s intentional object
and the manner of presentation of that object—which enter into its individuation
conditions.
As a result, attempts at propositional attitude paraphrases of reports of token
ASEs are unpromising candidates to re-describe the episodes in question. At best,
such paraphrases would relate subjects to contingent propositions concerning
some aspect of the environment to which the subject is attending. But since such
propositions can be true or be false, the propositional attitude paraphrases would
imply that there are false versions of the veridical token episodes in question, and
true versions of the non-veridical token episodes in question. Instead of capturing
the unipolar token ASEs, the paraphrases would characterize bipolar token states,
which may correspond to nothing that actually makes up courses of experience.
Similarly, paraphrases in terms of factive attitudes would also misrepresent the un-
derlying episodes. What facts are Paul, Jane, Uri and Sam reported as perceptually
taking in? None whatsoever!
The friend of facts perceptually taken in might initially find more comfort
in the reports concerning Mary and Fred. Mary, it might be said, is perceptually
taking in the fact that the calvados is astringent; and Fred is perceptually taking in
the fact that the moon’s reflection in the water is bright. But these kinds of para-
phrases are a manifestation of ontological insensitivity, as many considerations
could be marshalled to show.
Just to cite one: suppose Mary next tastes the peatiness of a single malt, and
spits that liquid out as well. On the basis of the two episodes she could be in a
position to judge that the astringency of the calvados is more remarkable than the
peatiness of the single malt.
What Mary is reported as experiencing by the two objectual reports:
Mary tasted the astringency of the calvados and then spat the drink out.
Next, Mary tasted the peatiness of the single malt and then spat the drink
out.
namely, the astringency of the calvados and the peatiness of the single malt does
provide her with the basis to make that comparative judgment. Notice, however,
what the factive paraphrases report Mary as “taking in” perceptually, namely the
fact that the calvados is astringent, and the fact that the single malt is peaty, do not
provide her with the basis to make the comparative judgment. They only provide
her with the basis for a quite different comparative judgment, namely that the fact
that the calvados is astringent is more remarkable than the fact that the single malt
is astringent.
To see the difference, suppose that Mary is spitting the drinks out because she
is a spirit judge, and thus the fact that the calvados is astringent and the fact that
 The Problem with the Content View   131

the single malt is peaty are not at all remarkable to her, so neither is more remark-
able than the other. Still, she could find the astringency of the calvados and the
peatiness of the single malt both remarkable, and the astringency of the calvados
even more remarkable that the peatiness of the single malt. She could make such
comparisons on the basis of what she experienced (and her background knowl-
edge.) So the astringency of the calvados and the peatiness of the single malt
should be counted among the things she experienced.
The calvados’ particular astringency which Mary tasted could not continue to
exist without maintaining its utterly determinate character, and the single malt’s
particular peatiness which Mary tasted could not continue to exist without main-
taining its utterly determinate character. Yet, the fact that that the calvados is astrin-
gent could continue to obtain while the calvados’ particular determinate degrees
of astringency varied, and the fact that the single malt is peaty could continue to
obtain while the single malt’s particular determinate degrees of peatiness varied.
Facts register the obtaining of certain conditions, without thereby registering the
more determinate details of how they obtain. But experience takes in the tropes or
individual features of things, and those features are individuated by the determinate
way they are. It is by taking in tropes or individual characteristics that experience
provides for comparisons of the sort Mary is reported as making when we say
Mary found the astringency of the calvados more remarkable than the peati-
ness of the single malt.
The friend of facts may see a way forward here, by building in a specific determi-
nateness into the facts perceptually taken in by Mary. If these factive attitude reports
were true
Mary perceptually took in the fact that the calvados was astringent-to-degree n.
Mary perceptually took in the fact that the calvados was astringent-to-
degree m.
Mary would have been put in a position by her experience, as it is here modeled,
to find the degree of astringency of the calvados more remarkable than the degree
of peatiness of the single malt. At least she would be, if as we might suppose, she
is aware of the common degrees of astringency and peatiness of the respective
spirits. (The suggestion would have to be elaborated for any dimension of com-
parison along which Mary might find one of the respective tastes more remark-
able than another. For example, Mary may in fact be finding the astringency more
remarkable because of its purity, not its intensity.)
But this suggestion runs afoul of the fact that our perception of magnitudes,
as opposed to our subsequent theorizing about them, is unit free. “Degree n”
and “degree m” are only given sense in a theory of measurement or comparison,
and that is a late achievement of cognition, not part of the everyday deliverances
of perception. Moreover, we can experience determinate individual features
which admit of comparison in respect of their degrees along some dimension,
 Against Strong Content
132   

without thereby experiencing just what their degrees along that dimension are.
But the factive paraphrase, employed to solve the problem of Mary’s compar-
ison, appears to place the degrees themselves in the facts experientially taken in.
Thus, reverting to facts taken in rather than individual characteristics pre-
sented invariably misdescribes what we are given in experience. And this is just
the beginning of a cascade of points against the idea that the right way to parse
our experiences is as the “perceptual taking in” of facts. Indeed we need to resort
to that unfamiliar idiom precisely because facts are not seen, or heard or smelt or
touched or tasted. We might eventually arrive at a relation to a fact on the basis of
what we see, hear, smell, touch or taste—we might see that the flag is blue—but
that is a destination from, and not a starting point in, what we are aware of in
experience.
For these reasons, among others, the objectual reports involving Paul, Jane,
Sam, Mary and Fred cannot be adequately paraphrased either in terms of proposi-
tional or factive attitudes. The ASES that the objectual reports describe need to be
recognized in any account of experience as we ordinarily conceive of it. The ASEs
involve the presentation of objects, quantities of stuff, events and tropes under
certain manners of presentation. The items presented are truthmakers for the
simple existential and predicative judgments we immediately make on the basis
of perception.

5.10. Travis’ Worry About Experiences

Recognizing the ASEs also enables a response to a pressing worry faced by recent
philosophical discussion of experiences, the worry that much talk of experiences
is itself too unspecific to be of real philosophical use (Travis 2004). There is the
quasi-mass noun use of “experience”, as in “experience is mostly veridical”. There is
the dummy sortal use, as in Alex Byrne’s example—“My three most embarrassing
experiences in graduate school”—where we can paraphrase away apparent ref-
erence to three token experiences in favor of three occasions of embarrassment.
Travis and Byrne urge that philosophers need an account of experience that goes
beyond the thin commitments of such usages, at least if the notion of an experi-
ence is to do any real philosophical work.
Hence the relevance of the ASEs; they are paradigmatically experiential, they
are countable and datable episodes, and they are taken for granted by, and studied
in, perceptual psychology and psychophysics (Johnston 2011a). Moreover, we do
in fact report such episodes in describing our sensory interactions with our envi-
ronment. Furthermore, we can use attentive sensory episodes to understand the
notion of a stretch or course of experience, i.e. the total experiential state of a sub-
ject as it evolves over a given period of time.
I open my eyes, and something happens; I enjoy a course of visual experience,
which involves and supports a range of token ASEs. Nothing could be that very
 The Problem with the Content View   133

course of experience unless it involved and supported the same token ASEs. As
the lighting or perspective changes, and as my attention-driven gestalt groupings
change, the course of experience changes its character. A course of experience is
thus partly constituted by a series of ASEs, with one or another evolving out of pre-
vious ASEs in accord with changes in the subject and the environment.
For the reasons already canvassed we can’t fully explain the nature of courses
of experience in terms of successions of propositions EX-ed or facts taken in; for
both sorts of successions omit, each in its own way, the presentational character
of the ASEs that are at least in part necessary to a course of experience being what
it is.
True, not everything given in a course of experience is an object of attention,
and hence a target of an ASE, but if something is experienced it is a potential ob-
ject of an ASE (residual elements aside). So let us individuate token courses of
experience in terms of the ASEs they ground, or actually and potentially support.
Then we can individuate types of experiences by abstracting away from the sub-
ject, particular constitution and time of occurrence of token courses of experi-
ences. We can say:
A type of course of experience E = E* if and only if (subject, token constitu-
tion and time aside), E and E* have the same actual ASEs (same targets, same
manners of presentation) in the same order and the same ground for potential
ASEs, and the same residual elements, all in the same sequence and pattern.
Among the residual elements which enter into a course of experience may be
forms of unconscious perception, which shape the character of what is perceived
both at the level of objects presented and their manner of presentation. There may
also be ASEs that are themselves unconscious in the sense of not being available
in introspection, and these also may shape the conscious character of a course of
experience (Block 2011). The important point is that such residual elements need
not be modeled in terms of propositions EX-ed or facts perceptually taken in.
In any case, all we need to rely upon in what follows is a necessary condition
on being a particular course of experience, namely:
E = E* only if (subject, token constitution and time aside) E and E* have the
same actual ASEs.
Since there is no modeling of courses of experience without modeling the ASEs,
it follows that if ASEs cannot be modeled adequately by propositional attitudes or
facts taken in, then neither can courses of experience.

5.11. On the Redundancy of EX-ings

ASEs involve attention, attentive search and the like. Although in some cases ASEs
may be forced on us—the thud may be so loud that it captures our attention—often
 Against Strong Content
134   

FIGURE 5.8

they are things we do because of explicit beliefs and desires that we have. We can
be ordered or urged to attend. “Try to visually discriminate the depiction of the
young woman in the array!” Just because ASEs are early on in perceptual proc-
essing doesn’t mean they are not directed by what is already there “later on”.
Thirsty people may be better at visually detecting transparency. And there is some
evidence that emotional attachment effects size perception. Moreover, we are often
enjoying sensory episodes as part of an explicit investigation of our environment,
as in a search for clues.
Even so, undergoing an ASE in which a certain object is presented as being a
certain way does not intrinsically involve one in taking the thing to be that way.
Someone very familiar with the following illusion and someone entirely unfa-
miliar with it could each have an ASE of the very same type, one in which the
pencil is presented as bent. The first person will not go on to believe that the pencil
is bent, while the second is likely to do so.
In presenting the pencil as bent, the relevant ASE is not silent as to how the
world might be, but in enjoying that ASE the subject him-or-herself is not com-
mitted to the world being a certain way. Experience, understood in terms of ASEs,
is not neutral as to the way the world is, but undergoing an ASE does not commit
 The Problem with the Content View   135

the subject to the world being a certain way, and certainly not in the fashion in
which the subject’s judgments or beliefs do.
In this way we can make sense of the idiom of “the testimony of the senses”.
Experience involves a commitment to how the world is, but it is, as it were, experi-
ence’s own commitment, which a subject can take on as his or her own by trusting
experience, i.e. coming to adopt the beliefs that experience supports, the beliefs
whose truthmakers are presented in experience, under recognizable modes of
presentation.
For many friends of the content view, the theoretically posited EX-ings
that p were intended as experiential attitudes that were prior to perceptual
judgment and belief, not only in time, but also in the sense that although
EX-ings had their own conditions of veridicality and hence themselves in-
volved “a commitment” to the world being a certain way, the EX-ings could be
enjoyed by subjects without those subjects being committed to the world being
a certain way.
That role—the role of characterizing the kind of commitment character-
istic of experience, in virtue of which it can come in veridical and non-veridical
forms—can also be played by the ASEs. But as our sequence of standard illusions
illustrated, we do better in this regard by relying on the ASE’s; we can explain how
the commitment characteristic of experience allows for a further kind of failure
exhibited by illusions which are nonetheless veridical.
So EX-ings are not only inadequate models for the ASEs; they are redundant
and not fully adequate as models of the kind of commitment incurred by experi-
ence. Moreover, they are theoretical posits, rather than episodes we actually report.
Finally and most importantly, unlike the ASEs, the EX-ings do not model experi-
ence in such a way as allow for the distinctive epistemic virtue that attends our im-
mediate perceptual judgments; those judgments are formed in the sensed presence
of their truthmakers, and this provides us with a distinctive kind of entitlement to
those perceptual judgments, an entitlement which we can produced if challenged.
What gives one the right to believe that the leather is red? Nothing could serve
as a better response than saying “I see the redness of the leather.” If that latter claim
is true, the subject possesses and is thereby exploiting a conclusive entitlement for
his belief. His belief couldn’t but be true given what he experienced. Saying “I am
EX-ing that the leather was red” describes no such entitlement. The proposition
that the leather is red leaves open whether it is true or not, and so leaves open
whether the belief that the leather is red is true or not. Successful perception actu-
ally closes that option off.

5.12. Could EX-ing be Believing?

The standard version of the content view, which has EX-ing that p come before
judging or believing that p, faces the old question of how something that is not a
 Against Strong Content
136   

belief can nonetheless bear on what one should judge or believe. Awareness of a
truthmaker for p can bear on the judgment that p, in that it is a presentation of an
exploitable entitlement to judge that p. But how does some propositional attitude
that is not itself the belief that p, indeed some attitude that does not itself involve
the subject in taking p to be true, bear on whether the subject should judge or be-
lieve that p?
It is noteworthy that Byrne is here led to cut the Gordian knot, and propose
that EX-ing is a kind of believing, the kind of believing where the particular sorts
of contents believed are those characteristic of perceptual experience, and not
those that enter into the corresponding verbal judgments. That is, the believings
that are EX-ings have contents restricted to the predication of visual gestalts made
up of “mid-level perceptual features” such as shape, size, motion, color, shading,
texture, orientation, timbre, loudness, pitch and the like, the very kinds of features
perceptual illusions prompt us to misattribute to objects.
Of course, Byrne is well aware of the fact that in cases of known illusion such
as the illusion of the bent pencil in water, we are not inclined to believe what the il-
lusion seems to show, i.e. that the pencil is bent. Byrne returns to the view of David
Armstrong (1968), suggesting that in such cases the belief in question is prevented
from playing its usual causal role because of the other things we believe, namely
that the presentation of the bent stick is an illusory appearance of the pencil’s shape.
Whatever other difficulties attend this view6, the main point to make is that
this is just another way of giving up on the distinctive epistemic contribution of
fully successful perceptual experience. Our immediate perceptual beliefs are often
reliably caused. But beyond that there is little to be said on a view like Byrne’s about
the distinctive epistemic significance of perceptual experience. Someone who had
reliably caused non-experientially based beliefs about the scenarios before the
senses—say a perfect blind-sighter and deaf-hearer—could be as well-placed epi-
stemically as those of us who can see and hear.
A theory’s yielding that implication is the definitive sign that the theory is
ignoring the distinctive epistemic significance of experience. It now appears that
all varieties of the content view do this in one or another way.
The “content of experience” frame that has shaped the recent discussion of
perception over the last thirty years is far from innocent; instead it is demonstrably
misleading, and it is disabling when it comes to providing an account of the dis-
tinctive epistemic significance of perception.

References

Armstrong, D. M. (1968). A Materialist Theory of the Mind. London: Routledge.

Block, N. (2011). Attention and mental paint. Philosophical Issues, 20(1), 23–63.

6
Does it conflate an experience’s commitment or “testimony” with a commitment of the
experiencer?
 The Problem with the Content View   137

Chisholm, R. (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press.
Crane, T. (2006). Is there a perceptual relation? In T. S. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 126–146). Oxford: Oxford University Press.
Crane, T. (2011). Is perception a propositional attitude? In K. Hawley & F. Macpherson
(Eds.), The Admissible Contents of Experience (pp. 83–100). Oxford: Blackwell.
French, C. (2013). Perceptual experience and seeing that p. Synthese, 190, 1735–1751.
Harman, G. (1973). Thought. Princeton: Princeton University Press.
Hyslop, A. (1983). On ‘seeing-as’. Philosophy and Phenomenological Research, 43(June),
533–540.
Jackson, F. (1977). Perception. Oxford: Oxford University Press.
Johnston, M. (1983). Particulars and Persistence. PhD thesis, Princeton University.
Johnston, M. (1992). How to speak of the colors. Philosophical Studies, 68(3), 221–263.
Johnston, M. (1997). Postscript: visual experience. In Alex Byrne & David Hilbert (Eds.),
Readings on Color I: The Philosophy of Color (pp. 172–176). Cambridge, MA: MIT Press.
Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 120(1–3),
113–183.
Johnston, M. (2006). Better than mere knowledge? The function of sensory awareness.
In T. S. Gendler & John Hawthorne (Eds.), Perceptual Experience (pp. 260–290).
Oxford: Oxford University Press.
Johnston, M. (2007). Objective mind and the objectivity of our minds. Philosophy and
Phenomenological Research, 75(2), 233–268.
Johnston, M. (2011a). On a neglected epistemic virtue. Philosophical Issues, 21(1), 165–218.
Johnston, M. (2011b). There are no visual fields (and no minds either). Analytic Philosophy,
52(4), 231–242.
Johnston, M. (n.d.). Ontology without Properties. Manuscript.
McDowell, J. (1996). Mind and World. Cambridge, MA: Harvard University Press.
McDowell, J. (1998). The Woodbridge Lectures: Having the world in view: Sellars, Kant, and
intentionality. Journal of Philosophy, 95, 431–491.
Millar, A. (1985). Veridicality: More on Searle. Analysis, 45(March), 120–124.
Moltmann, F. (2013). Abstract Objects and the Semantics of Natural Language. Oxford: Oxford
University Press.
Mulligan, K. (1995). Perception. In B. Smith & D. Smith (Eds.), Husserl, Cambridge
Companions to Philosophy (pp. 168–238). Cambridge: Cambridge University Press.
Searle, J. (1983). Intentionality. Oxford: Oxford University Press.
Siegel, S. (2006). Which properties are represented in perception? In T. Gendler & J.
Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Travis, C. (2004). The silence of the senses. Mind, 113(449), 57–94.
 6

The Preserve of Thinkers

Charles Travis

In a common mediæval outlook, what we now see as the subject

matter of natural science was conceived as filled with meaning,
as if all of nature were a book of lessons for us; and it is a mark of
intellectual progress that educated people cannot now take that
idea seriously, except perhaps in some symbolic role.
—john mcdowell, Mind and World (1994, p. 71)

It is as though we had imagined that the essential thing about a living

person was his outer form, and so produced a block of wood in this
form; and were abashed to see the dead block, which had no similarity
to the living being at all.
—wittgenstein, Philosophical Investigations (§430)

Have we made the progress McDowell speaks of? Science has, no doubt. In philos-
ophy, though, that mediæval idea may masquerade as science itself. Masquerade
only: nothing, so not science, suggests messages in nature of the sort found in
books. So I will argue. Nature is full of messages for us. That red sky at night tells
a sailor something. But it is superstition to approach such messages as one would
a text or utterance or speech act—though superstition which still tempts some.
What distinguishes the messages in texts or speech acts? First, they are issued,
produced, conveyed, by some author. (They are also borne by, or contained in, the
text or act itself.) Representing can just be holding a stance or posture towards
things, a condition one is in. Representing something to be so (henceforth
representing-to-be), e.g., can just be taking it to be so. I will call such represent-
ing autorepresenting. Such will be a side issue here. By contrast, the authoring of a

A slightly different version of this essay also appears in my, Perception: Essays After Frege
(Oxford: Oxford University Press, 2013). I thank OUP for permission to print it here. I am grateful to
Mike Martin, Mark Kalderon, Guy Longworth, and Craig French for helping me see where some of
138 the lines here lead.
 The Preserve of Thinkers   139

message is an episode, a happening. (We can, of course, think of a book’s bearing

of an authored message as a condition it is in.)
Second, the episode in question is one of producing something. An author
of the sorts of messages found in texts and speech acts, issues a message, thus
assuming responsibility; liability to praise or blame for achieving, or not, those
successes or failures at which the message is to be taken to be aimed. Where a
book contains messages, there is a door at which blame is to be laid.
Third, the kind of representing involved in texts and speech acts is
representing-as: it is (inter alia) representing things as being some way there is
for things to be. Not all representing-as is representing-to-be. Pia may represent
Sid as a ballerina by sketching him in tutu and third position, without suggest-
ing that he is one. Correspondingly, not all the representing found in books is
representing-to-be. But all representing-to-be is representing-as. To express the
wish that Sid stop snoring, Pia must represent Sid as being one who snores, and
assign this a certain status: what is wished to become not so. Equally, to represent
Sid to be one who snores, she must represent him as a snorer and assign this a
certain status: a way things are. (I do not claim that speech acts have unique pars-
ings). So representing-as is a general case of which representing-to-be is a genre.
What had the capacity to represent-to-be, thus what had the capacity to represent
truly or falsely, would ipso facto, more generally, have the capacity to represent-as.
Fourth, issuing a message, so bearing one, is making it suitably available, its
issuing manifest. So for representing of the kind at stake here, the kind books go in
for, to be is to be suitably recognizable. Making something recognizable requires
suitable means for doing so. Among the means at work in any given case of the
representing I am after here is what I will call a vehicle. A vehicle is, first, some-
thing which is recognizable as what it is—so as occurring, present, or not—inde-
pendent of whether any representing is going on, or of what messages, if any, it
bears. Second, it is such that its production, in the circumstances in which it serves
as vehicle, makes recognizable just that representing-as done by its author (pro-
ducer) in producing it. It might, e.g., be some English words, or some graphic form
they have. If an author may be said to have represented things as being thus and so,
or to have assigned that way for things to be a certain status, then his vehicle may
be said (on a different reading of the verbs ‘represent’ and ‘assign’) to do so too. Pia
said that Sid snores, her words say that he does.
With an eye to the contrast with autorepresenting, I will use the term ‘allorepre-
sent’ (and its derivatives) for representing which is authored by an author who (which)
thus incurs responsibility for its successes and failures, which is representing-as, and
which as such that for it to be is for it to be suitably recognizable. I mean this to be
read so that both the author and his (its) vehicle can be said to allorepresent, each on
his/its proper reading of the verb. My theses are then: allorepresenting and autorep-
resenting are the only forms of representing-as, hence of representing to be so; only
a thinker, or a thinker’s vehicles, can allorepresent. (For a vehicle to represent-as as
it does is for it to be the vehicle it is for the thinker, or thinkers, whose vehicle it is.)
 Against Strong Content
140   

Hence (bracketing for the moment autorepresenting) only a thinker, in a demanding

sense to be spelled out, can engage in representing-as, thus in representing to be so.
Some think that what bore content as an authored vehicle does might, for all that, be
authorless. As noted, I hope to help that idea out of the world.
Allorepresenting contrasts with what I will call effect-representing—a rela-
tion between one historical circumstance and another. Here one bit of history
is what is represented. Another does the representing. That teetering rock rep-
resents æons of wind erosion. Pia’s haggard mien represents years of Sid’s grunt-
ing. Generalizations obtain. Teetering rocks may always represent wind erosion
(except where they do not). Effect-representing is far from reserved for thinkers.
Whatever happens does it. All it takes is an ætiology. Its role here is as what allo-
representing had better not turn out to be.
Allorepresenting is choosier than effect-representing. Those empty seats
in the house may represent (the workings of) poor casting, a hostile press, Sid’s
paunch (he playing the lead), and so on ad inf. No need to choose; a fortiori no
need for the seats to choose. Where there is a case of allorepresenting, there is such
a thing as the way things were thus represented as being. Something must choose
what way this is to be. I mention this now, for elaboration later, because it is a point
that will matter very much.
This essay effect-represents the posing of a question, ‘Does perceptual experi-
ence have content?’ It places that question in a wider context. If having content is
indulging in representing-as, and if my thesis holds, then the answer is no. This is
not to say that someone who enjoys perceptual experience does not, perhaps inev-
itably, in doing so thereby autorepresent.

6.1. Thinkers

The notion of a thinker at work here is Descartes’. Aiming to distinguish res cogi-
tans from dumb brutes and refined machines, he offers two marks,
of which the first is that they [machines, brutes] could never use words or
other signs, composing them as we do to express their thoughts to others.
For one could indeed conceive of a machine being so arranged that it offered
words, and even that it offered certain ones about material actions causing
certain changes in its organs . . . but not of it arranging them diversely so as to
respond to the sense of all that was said in its presence in the way that even
the most mentally deficient men can. . . . And the second is that, while they did
several things as well as, or perhaps better than, any of us, they would infal-
libly fall short in others, by which one would discover that they did not act
through knowledge, but solely by the disposition of their organs. (1637, p. 92)
“Reason,” Hilary Putnam wrote, “can transcend whatever it can survey” (Putnam
1988, p. 119). Such is Descartes’ idea. Take any implementable theory of how to
 The Preserve of Thinkers   141

do such-and-such—a theory with definite predictions as to the thing to do when

faced with such a task. A Cartesian thinker is always prepared to recognize ways
of performing the task other than those the theory dictates; moreover, to recog-
nize whether such a new way, and not the theory’s, would be the thing to do—and
whether the task itself is a thing to do. We, but not swallows, can recognize when
old ways of building mud nests, or times for building them, are not best. Our sen-
sitivity to the world’s bearing on the thing for us to do is, unlike theirs, unbounded
in this sense.
Suppose the task is recognition—e.g., telling pigs at sight. Pigs are recog-
nizable by how they look. No one thinks, though, that to be a pig just is to have
that look. Porcine (or ovine) cosmetic surgery is, so far, pointless but hardly in-
conceivable. Though most of us could not say just what it is that makes a pig,
for any putative porcine feature, we are sensitive to what would bear on whether
what lacked it might, for all that, be a pig (or what had it might not be). Here our
capacities transcend whatever reason can survey, as per Putnam’s idea.
Keeping up one’s end in a conversation is a project, often taxing. Descartes’
first mark of a thinker is, thus, a special case of his second. Pia says, ‘My Porsche
is in the shop.’ For Sid to respond to this—with what intelligibly is a response—
would be for him to say what bears, in some understandable way, and some way
he could understandably aim for it to bear, on the Porsche being in the shop (or
on Pia’s having said so). He might say, e.g., ‘I hope you like Opels’, or ‘I’ll warn
our taxista’, or ‘German over-engineering!’, or ‘I’d better rent some films’, or ‘Have
you been paid this month yet?’, depending on the way this would be understood
to link to what Pia said and the links he aims to forge. Renting films may or may
not be the thing to do when Porsche-less. Any of indefinitely many things might
make it so—because one just would not go out without the Porsche, because the
films will cover the sound of Pia’s weeping, because if you give the mechanics films
to watch, perhaps they will actually fix the Porsche, etc. Moreover, the connec-
tion between the Porsche being in the shop and Pia being Porsche-less for the
weekend is itself contingent. ‘I’d better rent films’ might, or might not, be con-
tinuing the conversation, depending on whether some such connection between
films and Porschelessness is one there might, in the circumstances, intelligibly be,
and one Sid might be understood to be making. Sid need not aim to continue the
conversation. But for him to be an intelligible conversation partner—equipped for
conversing—he must be sensitive to how the world might work in forging such
links, and to their existence or not; to how it might thus bear on the response for
him to make. Descartes’ point: such sensitivity, for a Cartesian thinker, transcends,
in Putnam’s sense, whatever reason can survey.
Not all allorepresenting continues a conversation. A weather bulletin does
not. Pia telling Sid that the Porsche was in the shop started one. So, one might
think, not all allorepresenting requires sensitivity to those same factors on which
the cogency of a response depends. But for a Cartesian thinker, at least, allorep-
resenting is always a project, guided by sensitivity to the world’s bearing on the
 Against Strong Content
142   

thing to do in realizing it—inter alia, on how to represent things—so to those

same considerations which filter responses from mere chatter. What Sid says to
Pia depends not just on what, as he sees things, a reply might be, but also on what
further ends he aims for his allorepresenting to serve—being sympathetic, making
light of things, evincing disinterest, suggesting how Pia can make it through the
weekend. That Cartesian, theory-transcendent, thinking which guides his percep-
tions as to what a response would be (and what response) works here, too, in his
seeing what to do to reach his aims, and, in such matters, what his aims should be.
It is thus at work whether it is a question of continuing a conversation or not. Our
allorepresenting draws, per se, on those capacities which mark a Cartesian thinker
off from an unthinker.
So far, allorepresenting draws on resources reserved for a Cartesian thinker
only insofar as it is in the service of further ends, such as conversing. It has not
yet been shown that all allorepresenting must aim so to serve; nor, more impor-
tantly, that such resources are drawn on anyway in fixing just how things are thus
represented being. A first step in this direction is to note that allorepresenting is
creative. In saying of a Mondriaan, ‘That’s Dutch’, Sid created a new way for things
to be. There was already, thanks to the painting’s creation, such a thing as it being
Dutch; now, thanks to Sid’s performance, there is also such a thing as things being
as he thus represented them. Had Mondriaan not so painted, there would not have
been that first way; had Sid not so performed, there would not have been that
second. The vehicle Sid used might have represented otherwise if used otherwise.
Something in his use of it must identify when things would be as he thus repre-
sented them.
Talk of creativity here may seem mere word play. Sid (somehow) selected a
certain (already existing) way for things to be—for that painting to be Dutch. He
represented things as being that way. There is, to be sure, the question how he
effected that selection. An answer might be interesting. But given this much, when
(in what cases) things would be as he represented them is decided by when that
painting would be Dutch. There is no more to Sid’s created way than this. So one
might think.
But perhaps not. Signs of more emerge when we ask what it would be for a
painting to be Dutch. Mondriaan, born in Amersfoort, with Dutch roots dating
from before the seventeenth century, moved to Paris and spent much of his
working life there. Suppose that he took French citizenship, joined a French col-
lective, and produced the painting, in their signature style, as (an anonymous) part
of their grand entrance into art history. Is the painting, then, perhaps, French? Or,
conversely, suppose that Mondriaan, born of French expatriates in Amersfoort,
had worked there all his life. Is his painting then French or Dutch? Might its style
matter to this? Again, were the van Eyck brothers (of South Netherlands) Dutch?
Such questions have no flat answers. With the van Eycks, for example, it depends
on what you count as being Dutch, or where you so count things. But there may
be unequivocal answers to some parallel questions as to whether things are as Sid
 The Preserve of Thinkers   143

represented them in representing the painting as being Dutch. If Mondriaan had

come from, and worked in, Ghent, for example, it might (depending on the cir-
cumstances of Sid’s allorepresenting) be clear that things were then not as he had
represented them. Given such possibilities, creating a way for things to be—as one
represented them in some episode of representing—might plausibly draw on such
capacities peculiar to a Cartesian thinker as the ability to tailor one’s representing
to the purposes it is to serve.
The ability to converse contains in it a certain freedom in language use, to
which Noam Chomsky points:
A typical example of stimulus control for Skinner would be the response to a
piece of music with the utterance Mozart or to a painting with the response
Dutch. . . . Suppose instead of saying Dutch we had said Clashes with the wall-
paper, I thought you liked abstract work, Never saw it before, Tilted, Hanging
too low, Beautiful, Hideous, Remember our camping trip last summer?, or
whatever else might come into our minds when looking at a picture . . . (1959,
pp. 26–58, 52)
What Pia says as Vic shows her his new Mondriaan might be any of indefinitely
many things. Her ability to allorepresent is one to respond to such provocations,
or any specifiable one, in any of indefinitely many ways. As I hope to make clear, it
would be misunderstanding what freedom is involved here if one took it for any-
thing other than the operation of Cartesian thinking—if, e.g., one thought of it as
merely the ability to produce what was to be one’s representing in the absence of
what it was to represent.
Descartes’ conception of a thinker is not the only one. A simpler one would
be: a thinker is whoever, or whatever, thinks things so. Cats and dogs might do
this, depending on what it is to think something so. I take no stands here. If
cats and dogs are thinkers in this sense, perhaps for all that they fail Descartes’
tests. We would then have a weaker notion. It would remain to decide whether
such weaker thinkers might, not just autorepresent, but also allorepresent—emit
representing-as. But where unthinking representing-as has so far been suggested,
it is not the work of such weaker thinkers. I thus leave this issue unresolved.
There is, though, one reason why allorepresenting might be a more demanding
enterprise than autorepresenting. It is that an allorepresenter is responsible for
which way things are thus represented being—under which generality things are
thus presented as falling—in a way that an autorepresenter is not. One illustra-
tion. Thanks to a happy turn of events in the winter of 1848, there is now a range
of thoughts to think which there might easily not have been. If Pia thinks that
Frege was glabrous, she thinks one of these. If Sid says that Pia thinks that Frege
was glabrous, he assumes liability for there being such a thought. Were there not,
Sid would thus be wrong as to how things are (as to what Pia thinks). By contrast,
Pia assumes no such liability in thinking this. If there really never was a ‘Frege’,
she simply would not have thought what Sid said her to. She would not still have
 Against Strong Content
144   

thought that, but mistakenly. Another possibility. If the wild boar were not in rut,
things would not be as Pia thinks. Do what you like, and but for that fact Pia’s
picture of the world would not jibe. But an overly genteel upbringing has left her
without the notion for a beast to be in rut. Perhaps she might still count as thinking
that the wild boar are in rut; but not thanks to her ability to identify that as a way
she thinks things. Perhaps (for all that matters here) a cat might stand similarly
towards a hole’s presence in a wall. Such, anyway, are reasons for separating two
notions of thinker as I have just done.

6.2. Generality

Effect-representing is a two-place relation, representing-as a three place one. In

effect-representing, one historical circumstance represents another. The presence
of those empty seats represents poor casting. (One type of circumstance might,
as a rule, or invariably, represent another.) By contrast, in representing-as, some-
thing, A, represents something else, B, as something, C.
What fills the A-place in allorepresenting is either its author, or his (its) vehicle—
not circumstances, but that whose being thus and so might be a circumstance. Where
allorepresenting is liable to success or failure (as in representing-to-be), it is the au-
thor at whose door blame, where fitting, is to be lain. He (it) bears the responsibility.
Some suggest that authorless vehicles might bear messages, so represent-as. It would
be obscure where then to lay such blame. It matters, correspondingly, how much it
matters that there should be such a place.
What might fill the B-place? In one case things which may be represented
as being thus and so, on that reading of ‘things’ which bars the question ‘Which
ones?’. ‘Things’, taken straight, means: things being as they are. Modified, it may
refer to things being as they will be, or were, or would be if. . . . Such are cases of
what might be represented as something. So might a thing. Its being as it is is then
what is its being, or not, as represented. So to represent a thing as something is to
represent its being as it is as something.
What matters most here is what occupies the C-place: that as which some-
thing is represented being. What fits in this place is a way for things (or for a thing)
to be. For Sid to snore, or things so being, is a way for things to be, so a way to
allorepresent things being, e.g., in saying so, or asking whether.
Frege identifies a generality inherent in any thought:
A thought always contains something which reaches beyond the particular
case, by means of which it presents this to consciousness as falling under
some given generality. (1882a, Kernsatz 4)
The generality at issue here is not one which distinguishes some thoughts from
others, but one belonging to all thoughts. A thought is, e.g., that Sid snores. It
is thus of things being such that Sid snores. It presents things so being; and with
 The Preserve of Thinkers   145

‘that’ attached, their so being as enjoying a certain status: as part of how things are.
Representing-to-be takes a further step: not merely presenting a given way as being
a way things are—what would be just more representing-as—but as assuming, or
incurring, liability to a particular sort of success or failure, getting it right or wrong.
No thought takes this extra step. It cannot aim at such success or failure (or any-
thing). The thought that pigs swim is not to blame if they do not.
Whence this generality? Following Frege, a thought is what brings truth
into question at all, done only by fixing (or being) a particular question of it; a
particular point on which thinkers might agree or not. One cannot simply aim
at truth tout court. It must be truth in re something. Which is to say: one cannot
aim at everything. So a question of truth cannot turn on everything. Whether that
Mondriaan is Dutch may turn on Mondriaan’s parentage, but not on whether
Pia was at Hédiard yesterday, or Sid is wearing sandals. It follows that a range of
cases—an indefinitely large one—are ones which would, or might, count as things
being such that that painting is Dutch—ones with Sid in socks and sandals, ones
with him pieds nus, and so on ad inf. A thought (and that way for things to be
which it is of), reaches in its own way to particular cases, thus reaching just what
it does. How it reaches is contained in it being the thought it is. Thus a thought’s
inherent generality and that of a way for things to be.
Frege puts two pieces in play. Thoughts, so ways for things to be, are one piece.
The other is what he calls ‘the particular case’—what a thought presents as falling
under some generality. What falls under a generality is intrinsically one-off: nothing
else could be things being as they now are. What makes the particular particular,
though, is rather its lack of reach. Nothing in its being the case it is identifies any
question of truth, or what matters to it. The sun is setting slowly over the Douro’s
mouth. For the sun to be setting slowly is a way for things to be. Things being as they
now are is a case of this. Study that case as closely as you like, and you will not learn
from it what matters, and how, to whether a particular case would be a case of this or
not. For this one must look at just what generality is to be instanced. Generalizing,
no proper part of a generality’s reach determines what further reach it might or
might not have. Generalities and particular cases are thus two fundamentally dif-
ferent sorts of things. I will speak of the first, ways for things to be, as conceptual, the
last, things being as they are, as nonconceptual. That core relation between these two
domains, being a case of, I will call instancing, its converse reaching to.
One can witness, e.g., watch, things being as they are. One does this, e.g., in
seeing the sun, setting over the Douro’s mouth. What is visible—the sun, e.g.,—has
location. What has location is what may interact causally with its surroundings.
Such is part of Frege’s point in insisting that thoughts cannot be objects of sen-
sory awareness. They are the wrong sorts of things for that. They are equally unfit
for causal interaction. It cannot be the causal profile of a way for things to be
which makes it occupy the third place in the relation allorepresenting for given
first and second terms. It has no such profile. Allorepresenting cannot be made of
effect-representing by any such route.
 Against Strong Content
146   

A given item within the conceptual participates in the instancing relation

in a given way. It pairs up in this, in a given way, with the particular cases the
world provides (or allows for). What determines its participation? Not logic.
Logic concerns relations within the conceptual; not those between the conceptual
and something else. Nor do relations within the conceptual, determine this; or at
least not without enough facts already given as to enough other terms of those
relations reach themselves. What makes things being such that Sid snores reach
as it does cannot be some law which dictates when to count a particular case as
instancing that generality, unless it is already given what particular cases that law
reaches.
Nor can it be relations within the conceptual which fix how the conceptual as
a whole relates to the nonconceptual. A question, ‘How, by what, does the concep-
tual reach to the nonconceptual überhaupt?’ can only be misbegotten. For a way
for things to be to be the one it is is (inter alia, perhaps) for it to reach as it does.
There is no identifying it as what it is while leaving it open for something else to
settle where it reaches. There is, accordingly no problem of how something else
could make it reach as it does. So, too, there is no grasping what way for things to
be a given way is without grasping well enough when something would be a case
of it.
Not, though, as though there cannot be reasons for and/or against counting
a particular case as a case of such-and-such. Quite the contrary. A way for things
to be, as Frege argued, is per se a way for our shared environment to be. Its
instancing (if it were instanced) by things being as they are would thus bear in a
particular way on how things would be otherwise. Its instancing would stand at
particular places in webs of factive meaning. There is, then, the question how its
instancing would matter if things being as they are did count as this, and, corre-
spondingly, of how its instancing ought to matter, to how things were otherwise.
Would it be right to count what mattered as its instancing would if this so counted
as instancing this way for things to be? A chrome yellow Porsche would normally
look yellow in daylight. Pia’s Porsche, though painted chrome yellow, would not
so look, e.g., because it is covered with baked-on beige mud. Is its being as it is a
case of a Porsche being yellow? What would follow if we said yes, what if we said
no? Is the way its being yellow would then bear on things consistent with what it
is or might be for a Porsche to be yellow? Such is a topic for rational discussion, in
which one who grasped what it was for a Porsche to be yellow would be equipped
to engage. To count her Porsche as yellow would be to take one view as to how a
yellow Porsche ought to, or might, look in daylight. To refuse so to count it would
be to take a competing view. To grasp what (a Porsche) being yellow is is to be
positioned to way such alternatives properly.
Here Putnam’s words apply again: reason transcends whatever it can survey.
How ought one to expect a Porsche to look if it is yellow? If Pia’s Porsche, while
painted yellow, does not now look yellow in normal daylight, one cannot expect
 The Preserve of Thinkers   147

a theory which generated in advance all the reasons there might be (or might
have been) for this. Nor, correlatively, could one expect to say in advance what
it would mean (factively) for the Porsche to be yellow if its failure here did, and,
again, if it did not, cancel its claim to count as being yellow—as instancing the
generality being a yellow Porsche. So nor could there (plausibly) be a theory
which predicted in advance, when it would be true to what a Porsche’s being
yellow is, where there was such a failure, to rule in the one way or the other.
Which, if right, is to say: there can be no specifiable prosthetic for our sense
of when to say (when it would be true to say), when not, that a Porsche is yel-
low. Which is to say: the ability to see this draws essentially on those capacities
which mark a Cartesian thinker. Thus, too, for the ability to see when things
would be as Sid represented them in representing Pia’s Porsche as yellow, so
the ability to grasp what way his created way—things being as he represented
them—is.
An ability to see what would, or might, count as a case of something being
yellow is very different from a mere ability to detect what are in fact cases of what
does so count, as an ability to see what would count as something being a pig dif-
fers from an ability to tell a pig at sight. An ability to tell a pig at sight is that thanks
to the fact that pigs are recognizable by certain visual features—by how they look.
But we all recognize that to be a pig is not, certainly not just, to have those features.
Not all that grunts is, or need be, porcine; not all that is porcine need grunt. So
an ability to tell pigs at sight is that only in a hospitable environment. Flood the
environment with enough ringers, and it ceases to be an ability at all. An ability to
see what would count as something being a pig transcends such limits. It is, inter
alia, an ability to see when we have ringers to deal with. Such an ability is what is
drawn on in identifying what way Sid represented things being in representing
Pia’s Porsche as yellow. It is for such abilities, I have suggested, that there is no
prosthetic.
An author of allorepresenting is responsible for his creations. Blame for suc-
cess or failure—e.g., for representing things as they are not, or as they ought not
to have been—is to be lain at his door. But he can be blamed only for what is in his
control, for what he/it can be responsive to having done or not. He could, might,
have done otherwise; he is thus blameworthy for not having so done. The point
just made is, in brief, only a Cartesian thinker could be thus responsible for having
represented things in one way rather than another.
To some this will seem wrong. To their eyes, nature, or some of its creations,
though no thinker in any sense, can assume the sort of responsibility for some
of its (or their) productions that one does per se in allorepresenting; notably the
sort of responsibility one does in representing truly or falsely—in making oneself
liable for being right or wrong as to how things are. The rough idea is: those cre-
ations exist to fulfil a purpose; they assume the responsibility something would in
undertaking to fulfil that purpose. What follows, I hope, will demolish that idea.
 Against Strong Content
148   

6.3. Selecting

Allorepresenting must accomplish a certain task. Something in its doing must

select, or identify, some one way for things to be as the way things were thus rep-
resented being. I will call this the selection task.
To identify a way for things to be is to fix its reach—what would be a case of
it. Selecting is thus tracing a path through a cloud. Suppose we think of a space
of ways for things to be. A point in that space is, say, things being such that Pia’s
Porsche is yellow. Now think of a space of particular cases. Take a proper part of it.
Restrict it, say, to all the cases there have been so far. Then that point in the space
of ways traces a class of paths through this space (if you like, fixes a subspace). It
traces those paths which connect all the particular cases to which it reaches. Inter
alia, it traces paths through that proper part, all the cases so far. Now take any par-
ticular case not in that proper part—say, things being as they will tomorrow. Three
classes of ways for things to be trace that same class of paths through the proper
part as our initial way does, but differ in what they do when it comes to this novel
case. One class are instanced by it. Another are not. A third do not settle whether
they are instanced by it or not. In what class is our initial way? The same question
arises for any way for things to be which allorepresenting creates, e.g., for being as
Sid represented things. Where the selection task is accomplished such questions
have answers.
What gives them answers in our own case? I now mention, briefly, two sources
of material. First, we are retrospectively sensitive to what we do. We have, as one
might put it, the capacity to be abashed. Sid can recognize that, as it turned out,
things were as he represented them, would not have been had the Porsche been in
the garage when it burned, but are all the same even though it is now mud-covered.
Answers to the question how he represented things could appeal to what he is thus
prepared to recognize. There is at least that source of material.
Retrospectively we stand detached from what we have done—not catching
ourselves in the act, but now seeing ourselves as other might. The capacity for such
detached stances is also exploited in other ways. We, or some of us, may collec-
tively identify some way for things to be independent of it being a way things were
represented on any given occasion. There is such a thing as calling Pia’s Porsche
yellow. We can then ask what we are prepared to recognize anyway, independent
of any episode of presenting something as falling under that generality, what we
would, or might, count, what not, as a case of things being that way—what we
would be prepared to call Pia’s Porsche being yellow. How much chrome could
you add, e.g.? We are jointly sensitive to such things. We can agree or dispute about
them. Room is thereby made for the objectivity of judgement; room for a given
case’s being one of a Porsche being yellow to be among the ways things are. It is
unclear how else such room might be made.
For Cartesian thinkers allorepresenting is a project, part of, and aimed
at serving, further ones. Such is a second source of material which could effect
 The Preserve of Thinkers   149

selection. A Cartesian thinker guides his projects by his perceptions of how the
world bears on what the thing to do would be—on which projects to execute, and
how. His perceptions reflect an unbounded sensitivity to ways the world does, and
might, bear on this. So it is, in particular, with his perceptions of what to allorepre-
sent, and how. We represent with an agenda. Such agenda may include contribut-
ing to, or furthering, further projects in particular ways—e.g., saying that whose
being so would bear in particular ways on how those further projects are to be
executed. If such is on the agenda, and if the representer succeeds in representing
things accordingly, he will make the contribution. Conversely, if such is (recog-
nizably) on the agenda, and if he can be understood as having represented things
accordingly—saying what, if so, would have that bearing—then such is reason so
to understand him. In what it would be so to understand him there is material
which could, if applied to identifying a way for him to have represented things,
effect selection.
I will elaborate this idea later. For the moment I merely illustrate. Guests are
coming. As Pia opens the wines to breathe, Sid sets the table. He has forks and
spoons in hand, but seems unable to find knives. Noticing this, Pia says, ‘There
are knives in the third drawer.’ Indeed there are. Suppose, though, that the third
drawer had contained Pia’s art supplies. These include a fair collection of matte
knives (roughly, handles mounting razor blades). One can understand there being
knives in a drawer so that the presence of matte knives counts as things so being.
But one might sometimes understand such talk such that such presence, on its
own, would not count as things being as thus represented. When I tell you where
the knives are, matte knives need not count as specimens of what I mean. The
words Pia used might be understood in either way. But she was to be understood
as speaking in aid of what Sid is doing—contributing in the way just scouted to his
project’s execution. Understand her in the first way and there is no such bearing.
Understand her in the second and there is. Such contributes at least to tracing a
path through the cloud of ways for things to be which contains all those which
are there being knives in that drawer on some understanding of there so being.
Whether this does, in fact, achieve selection for things being as Pia represented
them may remain an open question for the moment.

6.4. Agreement

We guide our allorepresenting by what could achieve selection. But, to borrow

Freud’s term, we are not always masters in our own house. When, and how, not?
Frege distinguished the psychology of holding true from the logic of being
true. Similarly one can distinguish the psychology of holding forth as true from
the logic of being as represented. In Adelaide, Sid comes in from the backyard and
announces, ‘The lamb is on the barbie.’ Little Tara screams, ‘Oh, no!’, and runs into
the backyard, where she finds her doll safe and sound on the table, while smoke
 Against Strong Content
150   

rises from the grill. There is something Tara understood Sid to say. That is a psy-
chological fact. There is something Sid meant to say. That is another. There is then
the question what Sid did say, how he is, in fact, to be understood. That ‘to be’, like
the ‘to’ in ‘the thing to do’, removes us from the psychological. Our concern now is
with the logic of being as represented. Such a non-psychological question need not
have a determinate answer: there was an amusing misunderstanding, and there’s
an end on it. But it may. Perhaps Sid said what he meant to, and Tara misunder-
stood. Or perhaps the other way around.
What answers such a non-psychological question? What makes an answer
right? A starting point: for there to be allorepresenting is for it to be (made) recog-
nizable. How are we to understand this ‘recognizable’? Sid’s representing might not
be recognizable to a monolingual Latvian, nor to a Martian, or a cat. Such hardly
matters. His analyst might recognize what he intended. She might recognize this
of still more bizarre performances. Such again does not bear on how he ought to
be—is to be—understood to have represented things. There are, though, those
who ought to be able to understand him; those competent enough, and apprecia-
tive enough of his circumstances to do so. There may then be what they would
have a right to expect if then so addressed; how a competent understander who
knew what he should have of the circumstances would reasonably have taken Sid
to be representing things. So, the idea is, did Sid represent things being.
Who are these people? In the example, most Australians, one would suppose,
and some of the rest of us (most of us, if initiates in Aussie practice and patois). But
what matters is this. You and I (and most Australians, etc.) share a sense of what
to say in cases like Sid and Tara’s. It is a sense which indefinitely many other think-
ers—perhaps not all—either share, or could be brought to share through sufficient
familiarity with our ways of allorepresenting. Given the psychological facts—the
actual facts of our agreement in such matters—there is such a thing as what a com-
petent, appreciative audience for Sid’s words would be; such a way for a thing, or
group, to be as being such an audience. That there is is something we can recognize.
Given what it would be to be this, we can also recognize this to be a way for a thing
to be which is instanced. There are, further, recognizable facts as to how one who
was this way would understand Sid, and his representing, were he so addressed.
Such (non-psychological) facts would of course be recognizable to one who was
the way in question. They are recognizable to us because we are that way.
For any allorepresenting there is its audience—the sort of thing equipped
and placed to recognize it for what it is. To belong to Sid’s audience is to instance
the just-mentioned way for a thing to be. In other cases, it would be to instance
being competent and appreciative in re the representing there occurring. In any
case, the audience is, in principle, indefinitely extendible. Sid, as any Cartesian
thinker, shares with his audience those retrospective abilities I called a capacity
to be abashed. Such can be directed in concert at what Sid has done. Just this is
what allows for effecting the selection task for his allorepresenting. So it is with a
thinker’s allorepresenting. It should be stressed that this way of failing to be master
 The Preserve of Thinkers   151

in one’s own house, so of relying on others for effecting a selection task, is reserved
for Cartesian thinkers. It works where a way the representer ought reasonably to be
taken is as doing what a Cartesian thinker might be doing (in the circumstances).
If the representer is not a Cartesian thinker, then, while mistaking it for one might
be understandable, he cannot have been to be taken as so performing.
Where allorepresenting has an audience (present sense), where there is such
a thing as what it would be to belong to it, to belong to the audience is to have
sufficient insight into how the representing is to be understood, and to have such
insight is to belong to it. When it comes to cases the audience is the measure of
what insight is here. That there is an audience may be manifest in its (extendible)
agreement. Just what is the audience’s role? Following Frege we may take it as in-
trinsic to any given way for things to be to reach just as it does. So fix a way and
nothing extrinsic to it, so no audience can make it reach in one way rather than
another. But what way a given way for things to be is is one question. What way is
such that Sid represented things as that way is another, as is what way one speaks
of where he speaks of lamb being on the barbie. If an audience provides no answer
to the first sort of question, it does provide the answer to the second. For a way for
things to be to be the way Sid represented things being (in speaking of his barbie)
is for it to reach to particular cases just as his audience (in the above sense) would
be prepared to recognize his representing (what he did) as reaching.
Any allorepresenting needs its audience, whether thinkers or not. For a Cartesian
thinker’s representing, the audience is of a certain sort. It shares a capacity: one, as
I put it, to be abashed. This capacity can be directed collectively at any instance of rel-
evant representing. It issues in acknowledgement of particular cases as thus reached
or not (not determined). Just here, in what a thinking audience would expect, the
crucial step is taken from the psychological to the nonpsychological—here from
holding forth to being true. What the audience would do. Where it is this audience,
is no longer a psychological generalization, nor a prediction. It is not like a statement
about what Sid, or Tara, or the average Australian would do. It is about how anyone
would respond to Sid if getting things right.
Our shared sensitivity to the conceptual performs this step for us. We achieve
selection tasks in ways which are the reasonable ones for the sort of representing
we engage in. Our sensitivity to the conceptual, such as it is, cannot be enlisted to
perform this step for an unthinker. If there are (parallels to) psychological gener-
alizations to be made about the unthinker’s doings, such need not be refractory to
us. If there are patterns in its responsiveness to the environment, we need not be
blind to these. But what it would thus do does not yet take us from the psycho-
logical to the logical, as something must if there is to be allorepresenting. Such a
step must be taken by the unthinker on its own, or anyway left on its own by us.
And the unthinker’s mere sensitivity to the presence of yellow, or pigs, in its sur-
roundings, whatever such may be, gives no right to construe any of its responses as
episodes of representing something as being yellow, or a pig, rather than as simply
detecting yellow’s, or porcine, presence. They effect no selection from within the
 Against Strong Content
152   

cloud in which that class of paths, cases of something being yellow are but an ele-
ment. The unthinker’s responses in its (presumably) hospitable environment give
no right to extrapolate from that subregion of particular cases to the space as a
whole. It would be anthropomorphism to construe its responding as it does to pigs
as, e.g., its telling us, or its peers, that a pig is about. If it were to present particular
cases as falling under generalities, the fact that we would be inclined to call what it
is doing detecting pigs, or yellow, gives no right to take those generalities to be at
all like those we can get in mind.

6.5. Deference

An unthinker could not take the step from the psychological (or mechanical) to
the logical in the way just sketched. The unthinker could not be to be recognized as
guiding execution of its representing as a Cartesian thinker would or might. Such
could not be the right thing to suppose of it. The Cartesian thinker’s way of tracing
a path through the space of particular cases could not be the unthinker’s. But per-
haps the unthinker need do no such thing. Perhaps he/it can simply contract the
work out, defer the selection task to some other source. One idea along these lines
would be: there might be a vehicle, identifiable independent of how it represents
things, which as such represents things as being some given way; and which, in
being the unthinker’s vehicle, would make it so that he/it so represented things.
Birds build nests but fail Descartes’ tests. So, Descartes thought, building
nests requires no intelligence. Who would think otherwise? Things are otherwise,
he thought, when it comes to holding conversations. If not all allorepresenting is
holding conversation, perhaps some, like nest building, is achievable by unthink-
ers. Something else would do the work for the unthinker that thinking does for us.
The above is one idea of what that something else might be: vehicles. These, the
idea is, would relieve the unthinker of the burden of selecting on its own. Its inca-
pacity would then not matter.
English sentences might seem a model for such vehicles. An English sentence
as such represents things as being a given way. It speaks as such of that way. The
sentence, ‘Monkeys fly’, speaks of what it does used or not, whether I take it to do
so, speak English, exist, or not. If it speaks of monkeys being flyers, then, where
I speak English, it would do so in my mouth. If it represents things as being a given
way then, the idea is, so do I in speaking it. The idea concludes: where I thus so rep-
resent things, for things to be as I represented them is just for monkeys to be flyers.
If English sentences so work, then, perhaps, so might other things, among which
things which would so work produced by (suitable) unthinkers. That English thus
models deference is an idea I hope now to dispose of. In its place I hope to put
this Fregean idea: the only way for a would-be content-bearer to come by the con-
tent of a thought, or of an element in one—to contribute to representings-as so
as to make them somehow about, say, relevant things being flyers—is for it so to
 The Preserve of Thinkers   153

function, or to be for so functioning, in the expressing of thoughts (by thinkers).

No other life it might lead could confer such content on it. In other words, the only
content-bearers there are (where content is a way of representing-as) are thinkers’
vehicles.
The sentence ‘Monkeys fly’ does say that monkeys fly. It speaks of them as
flyers. But in what aspect of the verb? Shifting aspects may produce illusion. The
aspect in which sentences say, or speak, stands out in other verbs. Robin, showing
his cousins from Peoria the Batcave, comes to the Batmobile. Pointing at levers
and buttons on the dash he says, ‘This one ejects the seats. This one fires the
grappling hooks. This one autodials the commissioner.’ Levers and buttons lack
initiative. It will be a long wait before a lever undertakes a project. Or so we hope.
Such does not reflect on what Robin said. That lever is for ejecting the seats. It is
the thing to pull to eject. If it is in working order, you (new aspect of the verb)
will then eject.
Such it is for levers to eject. This contrasts with the verb’s reading in ‘Don’t let
little Tarquin near the Batmobile. He always ejects the seats.’ It is equally a reading
of ‘fire’ and ‘autodial’ as above. It is one reading, too, of ‘say’. It fits the case where
we speak of the sentence ‘Monkeys fly’ as saying that monkeys fly. That sentence
(used neat) is for saying that monkeys fly, anyway for speaking of their being fly-
ers. If, on an occasion, you wish to say, in speaking English, that monkeys fly, this
sentence is, ceteris paribus, just the thing for you. Use it in speaking English, in
circumstances in which you would say something, and ceteris paribus, such is what
you will say, or at least speak of.
If Sid says, or said last Tuesday, that monkeys fly, such may be reason to think
they do. If the sentence, ‘Monkeys fly’, says that monkeys fly, such cannot be reason
so to think. The sentence, unlike Sid, is the wrong sort of thing to give such reason.
Tarquin always ejects the seats. He ejected them last just as Robin was leaving the
Gotham Diner. The sentence cannot have said last Tuesday that monkeys fly, un-
less this means that it has not, in the interim, changed meaning. To say, where ‘say’
has that past tense, is to incur liability to success or failure—to getting things right
or wrong—of a sort for which a sentence is ineligible.
For an English sentence to say something is for it to have a role in the lives of
(some) thinkers. It is for it to be a means for them to make certain allorepresenting
recognizable to their fellows; thus to execute successfully certain of their proj-
ects of representing. There is no hint of an idea here that sentences might lift the
burden of effecting selection from a being which could not, on its own, find its way
through, or select from, the space of possibilities, of ways for things to be, as we
do in aiming as we do to represent things as some given way, and in recognizing
what we have done as representing with a certain reach. English eases a burden for
those who can perform it, but does not lift it. Sid said that monkeys fly in saying
‘Monkeys fly’ only if he aimed, or ought to have been taken to be aiming, at saying
that; only if ‘Monkeys fly’ was used, or ought to have been taken to be used, for
achieving the success which would thus be aimed at. He would not have said so if,
 Against Strong Content
154   

as Frege puts it, the necessary seriousness were missing, e.g., if he could not prop-
erly be taken so to have aimed.
English has a syntax. It thus generates an indefinitely large set of vehicles, its
sentences, from a smaller set of building blocks by fixed rules. What a sentence
says, or speak of, is then fixed by what its blocks do, plus the rules which structure
them in it. Some ideas for unthinking representers-as require these to have an
indefinitely large set of vehicles they might produce. If perceptual experience rep-
resented things as being given ways, for example, it would need to be able to repre-
sent things as any of indefinitely many different ones. So then it would need a stock
of vehicles built from a smaller set of blocks by some fixed rules. What a vehicle
said would thus be fixed by what its blocks contribute to this. Now, it may seem,
content may accrue to a vehicle merely by virtue of accruing anyway to its blocks.
But if Frege is right, this analogy breaks down. If a building block is to con-
tribute to representing-as, what has accrued to it anyway, independent of this
representing-as, then what has accrued to it anyway must be no less than the fea-
ture of representing-as. In representing-as, truth is made to turn in a particular
way on how things are. An element of such representing makes truth so turn, in
part, in that way. Being what so functions in the context of representing things as a
certain way—making truth turn, full stop, in that way on how things are—is what
a building block would need to be already to function in the imagined way as a
building block at all. Representing-as cannot emerge from mere syntactic struc-
turing. Combine what effect-represents the presence of something puce and what
effect-represents the presence of a Porsche however you like, and all you get so far
is something which effect-represents the presence of something puce and the pres-
ence of a Porsche. What a given vehicle would require for representing some given
thing as puce is, inter alia, a block which, in context, does that. If the block does
that in context by virtue of content it has anyway, then that block must already be
what functions to make truth turn, in part, on how things are. It is difficult to see
how any block could have come by this through interactions with the environment
which are any less than roles in representing it as being thus and so.
In 1882 Frege wrote,
I do not think that the formation of concepts can precede judgements, be-
cause this presupposes an autonomous existence of concepts, but I think con-
cepts arise through the decomposition of a judgeable content. (1882b, p. 118)
Concepts arise through decomposing whole thoughts. A thought is true of
things, where there is no question ‘Which?’ It is true of things, so true, tout court.
A (non-zero-place) concept is true of a thing. Truth-of, Frege notes, can be under-
stood only in terms of truth. For the concept (a thing) being puce to be true, say, of
Ed is for it to be true that Ed is puce. A concept (as here spoken of) is a common
feature in a range of thoughts—e.g., that Ed is puce, that Pia is puce, that that
torus is puce. . . . It is one way each reaches to particular cases. It fixes a generality
under which all such thoughts fall: making truth turn on what is puce. It just is a
 The Preserve of Thinkers   155

common feature of those thoughts. There is no such feature unless there are such
thoughts. Concepts thus cannot precede thoughts.
So, too, for speaking of. There is no speaking of a thing as puce except in
the context of saying something as to what is or is not puce, or, more broadly,
representing things as some way the being which turns somehow or other on
things being or not puce. Speaking of a thing as puce (expressing the concept
of being puce) is not something which can precede speaking of things (catholic
reading) as thus and so. So speaking of a thing as being puce, expressing that
concept, is something a building block could do only in the context of its role
in the expression of whole thoughts. A building block might do that in isola-
tion only in that aspect of ‘speak of ’ in which to do so is to play a role in the
expressing of whole thoughts. Speaking of a certain way for a thing to be thus
cannot precede the expression of whole thoughts. So the accrual of content to
building blocks cannot precede the accrual of content to expressions of them.
What could not select a thought for a whole vehicle to express—a way for things
to be as how it represents things being—could not select a way for a thing to be
as what some building block contributes to such representing.

6.6. Recognition and Responsibility

Where allorepresenting is a project, making recognizable the representing

done may involve making recognizable what project is thereby executed. If it is
one of aiding further projects, or serving further ends, such may be what needs
making recognizable if it is to be made recognizable how things are then repre-
sented being. Those further aims and ends would then play a role in achieving
selection for the representing done. How things were represented being could
vary according to what those further aims were. Such would make allorepre-
senting like conversing and unlike building nests. This section expands that
idea.
In speech acts words are our allies in achieving recognition. We can exploit
their meanings to help make recognizable how we mean to represent things as
being. Such departs from the idea of English as a model for deferred selecting.
Words are aids in achieving recognition. For them to aid as they do need not be
for the way we do represent things being, when there is such success, just to be
the way they do anyway. Their role as aids need not be to be, on their own, the
expression of some given thought. Nor need it be fixed by their meanings alone
just what thought would be expressed in using them. Such is an idea exploited to
great effect in one way by David Kaplan (1989), and in a different way by Cora
Diamond (1991). It has appeared here so far in the idea, e.g., that there need be
no one way one speaks of things being in speaking of there being knives in the
third drawer. I now add: perhaps the ability to stand towards a vehicle as thus
described—to use it to express what need not be just what the vehicle anyway
 Against Strong Content
156   

expresses (two different aspects of ‘express’ here) may be intrinsic to the ability to
author representing-as at all.
Words which aid recognition need not do so by virtue of their meanings; nor
at all. In a restaurant in Abbeville Pia asks for ‘ortalans.’ Of course one cannot order
ortalans, or not in this establishment, or in this salle, or season. But of course,
too, this is not what Pia (an Anglophone) meant. The way she is eyeing the our-
sins shows her to mean them. Habituated to tourists, the waiter simply brings Pia
her oursins. Pia managed to make recognizable what way for things to be she was
representing as wished for. She managed to request oursins. The sentence ‘I’d like
the ortalans’ speaks of a different way for things to be. But not every use of it to
allorepresent speaks of that way. Even where it does, that it does so need not be
enough to identify how it represents things as being.
In the restaurant, the waiter arrives with their plateaux de fruits de mer—
bulots for Sid, oursins for Pia. But he looks perplexed. Clearly he has forgotten his
orders. Seeing this, Sid tells the waiter that Pia ordered the oursins. Suppose Pia
garbled things, or Sid had done the talking. Are things as Sid thus represented
them? If Sid were reporting Pia’s progress with speech therapy, or with her patho-
logical shyness (say, to a worried mother), the answer might be ‘No’. But here Sid’s
words are in the service of a further project, placing orders. In contributing to such
an enterprise, one is hardly to be held responsible for who did the talking. One can
understand ordering oursins so that who did the talking does not matter. So here,
for reasons stated, one is so to understand Sid.
To hold Sid to have represented things in one way rather than another is to
hold him responsible for something—here something as to how things are. There
is a way things had better be if he is to be let off with discharging responsibili-
ties assumed. For what is he reasonably held to account here? Where should his
wishes be acceded to, aims honoured, where not? Is he accountable for who did
the talking; liable to praise or blame accordingly? Is such reasonably reckoned part
of the bargain in the liability he went in for in representing to the waiter as he did?
Above, I suppose, the answer is, ‘No’. He made clear what message he had to offer.
He need be, so is not to be, held responsible for more.
Responsibility gives a reading to that ‘ought to be taken’ in that step, in sec-
tion 6.4, from the psychological to the logical, from holding to being true, the
step in the ‘way one did allorepresent things being is the way one ought to have
been taken to have’. There is what Pia is reasonably held responsible for in then
lending Sid her Porsche (at least what one should have foreseen). There is the re-
sponsibility Sid undertook, signed on for, in saying Pia to have ordered oursins.
Allorepresenting is among a thinker’s means for undertaking responsibility. There
is then what it is fair to hold him to have signed on for in using those means then.
How he represented things as being is fixed thereby.
Sid makes recognizable two things about his allorepresenting. First, he is to be
taken as representing things being as he does in representing them as being a cer-
tain way, namely, such that Pia ordered the oursins. In his execution of his project,
 The Preserve of Thinkers   157

the words he used are assigned the task of making this recognizable. Second, he is
to be taken as representing things as being that way whose instancing would have
a certain bearing on the way to execute a certain further project—the perplexed
waiter’s. He is to be—or asks to be—assigned responsibility accordingly. Perhaps
he could not be doing both these things jointly. Such is one way for it not to be pos-
sible to take him as he asks to be. Perhaps one cannot understand ordering oursins
so that whether Pia spoke does not bear on this, or that whether she did thus bears
on what the waiter is to do. But suppose we can. Sid ought not to be held to be
taking on responsibility he makes recognizable that he is not signing on for. One
ought not so to rely on him. In which case, these two features of what Sid was to
be taken to be doing jointly identify what it would be for things to be that created
way, being as he thus represented things. It is that way which reaches to just those
cases in which the world is such as to bear as it was to be supposed to bear on what
the waiter was to do, where things so being is understandable (might count) as Pia
having ordered oursins. It matters not whether it also has another name.
Sid represents things as he does in speaking of them as a certain way there is
anyway for things to be: such that Pia ordered oursins. He speaks on a particular
understanding of her having done so. One way to picture this would be as filling-
in. That way he spoke of, Pia having ordered oursins, reaches as such in a certain
way. Some range of cases is thus reached. Some other range fails to be. Other cases
remain undetermined. What it is for Pia to have ordered oursins yields as such
no verdict where Sid alone spoke to the waiter. The particular understanding on
which Sid spoke fills in some undetermined cases: on it some of these are reached,
some fail to be reached, by that way he spoke of.
If this is how things are, one might get a further idea. If Sid’s work of repre-
senting fills in understanding of that way he speaks of, so that his representing
things as that way reaches differently than that way on its own, then, perhaps, on
some occasion his representing simply fails to accomplish any such work. Then
things being as he represented them would reach exactly as things being such
that Pia ordered oursins does on its own. Perhaps an unthinking allorepresenter
could represent like that. What Sid thus did contingently would just be, neces-
sarily, its lot.
In what sort of case would Sid have done no such work? One might
think: when Sid spoke to the waiter, his talk of Pia ordering had an agenda. Our
talk often has much less of one. Suppose Sid simply wrote a postcard to Ed back
home: ‘Wonderful dinner last night. I ordered bulots, Pia ordered oursins.’ Not
much there by way of further purpose to be served. But now, must Pia have done
the talking for things to be as Sid wrote? Nothing in his writing this gives one any
reason to suppose so. So if she did, things are as Sid said. If Sid spoke for her, things
still are. But this is a special understanding of Pia having ordered oursins. What it
is for her to have done so does not, on its own, decide whether she needed to do the
talking. What it would be for her to have ordered can be understood in either way.
A case where no filling in was done would be a rather special one. Perhaps we get
 Against Strong Content
158   

closer to it in those rarefied situations where, as philosophers or semanticists, we

ask what one would call ordering oursins, what not—projects of classifying cases
(though we seldom get far with them). To direct one’s ‘Pia ordered oursins’ so as
for it to be understood as contributing to such a project would not be to represent
aimlessly, but, on the contrary, to bend one’s representing in a very special way to
the service of further aims and projects. If such can be done, it is surely available
only to one who, by the same token, has a capacity for filling in.
Sid’s filling in as he does is the exercise of a capacity. Such is a capacity to
direct (orchestrate) his representing so as to achieve representing in some one
way rather than another—to fill in in one way rather than another, if such is to
be the image. He can direct so as to select. For this he must be sensitive to what
there is to aim at—to what would be achieved in directing things in some one
way as opposed to others. Thus that he can assume responsibility, be held to
account. Still retaining the image, he must also be sensitive to the possibilities
for filling in—to when one can understand ordering oursins in various ways,
and how one can. For one so equipped, achieving no filling in is just directing
one’s representing in one way rather than others he might. It is just one case
among others, the null case, drawing on the same capacities—the null case, a
degenerate case if you like.
Sid represents as he does in representing things as a certain uncreated way.
Sensitivity to the possibilities for filling in the reach of that way is just sensitivity to
how that way reaches, an ability to acknowledge it for what it is. Such belongs to a
capacity to represent things as that way punkt, independent of how one represents
things in doing so. Even if Sid deployed a vehicle which, as such, speaks of being
that way (in the only aspect in which a vehicle could), still, it is not automatic that
every use of that vehicle in representing (or attempting to) is a case of represent-
ing things as that way. (Recall Pia and ortalans.) Sid has a capacity to direct his
representing so as to deploy that vehicle for representing, in his representing, that
which it speaks of. It is thus that he can be credited correctly with representing in
representing things as that way which is the very one of which that vehicle speaks
(even, sometimes, when he does not so aim).
The unthinking representer supposed above is saddled with its representing.
It represents blindly. It is not sensitive to possibilities for filling in a way it repre-
sents things being. For, unlike Sid, it is not equipped to acknowledge any way for
things to be as the way it is—what an ability to fill in might add to. As we saw (sec-
tion 6.3), such equipment is proprietary to Cartesian thinkers. If it could represent
things as that way some vehicle it produces represents them, that would not be just
one special case among others of the ways open to it to direct its representing. At
which point the wanted comparison between Sid and this hypothesized unthinker
collapses. The unthinker brings nothing to representing, or nothing which has yet
emerged, to make its use of any vehicle representing things as being any way, no
matter what the vehicle may do as such.
 The Preserve of Thinkers   159

6.7. Force

Those ways we can represent things as being—so those ways we can take things
to have been thus represented (whether by us, or by any representer)—are
such that where we represent things as some such way, it might be any of many
things to be as thereby represented. The last section concerned the capacities
drawn on in such representing. Its idea can also be put in terms of force. For
Sid to have represented Pia as having ordered the oursins in the way he did
is for him to have assigned that way for things to be a certain status: as to
be counted as among the ways things are where its being instanced is under-
stood as it would be for certain purposes. Assigning status cannot be just more
representing-as. This, too, would await a status. To coin a term, it is doing one’s
representing-as with force—here, in assuming responsibility, vouching for the
status thus assigned.
As force is usually conceived it comes in a small range of varieties: assertive,
interrogative, imperative, optative and so on. Things change if, as per above, in
representing things as some given uncreated way there is for things to be, one
can represent them as any of indefinitely many different ways. Throughout one
would present that way in which he so represented things with a certain force. But
it would need to be a different force in each case. He would assign that way for
things to be a status in re being among the ways things are. But that status would
be, not being a way things are full stop, but counting as a way things are when you
understand things so being in a particular way (in the last section’s image, with
a particular permissible filling-in). Force would vary here—even in an assertive,
or an imperative, or etc., case—according to the responsibilities signed on for, as
identified, e.g., as per the last section, in terms of projects to be taken as contrib-
uted to.
There is thus a selection task for force paralleling that for what way things
were represented being. How Sid represented things being in saying Pia to have
ordered oursins is fixed, not just by this being the way he spoke of, but also by with
what force this was presented—how it was presented as counting as a way things
are. An unthinker is as little equipped to effect the task for force on its own account
as it is the selection task for ways for things to be.
An unthinker would be overcome by allorepresenting, as a Tourette’s victim is
overcome with blurtings. The expletives are not the sufferer’s. The allorepresenting,
one might well think, would no more be the unthinker’s. Is there some default force
such unthinking representing might have? Perhaps it is a ‘purely generic’ asser-
tive (or imperative, or optative) force. ‘Purely generic’ here would be abstracting
from all particularities of ways of presenting that way for things to be in which the
unthinker represented things as it did. The status assigned would be: a way things
are no matter how you understand things so being. The usual way of thinking of such
abstraction is in terms of universal quantification: the way things are on all under-
standings of things being it.
 Against Strong Content
160   

The idea here is familiar in philosophy (cf. Clarke 1972). To paraphrase Clarke,
to see whether Pia really ordered the oursins, we stand back from any mundane,
local concerns such as how to tell the waiter what to do, and, purely considering
the concept ordering oursins as such, and the world as such, ask whether Pia’s
doings do, or do not, fall under that concept. Whether there is such a project of
pure inquiry is controversial. As indicated in the last section, if there is, it would
not be one reached by abstracting from all the particular varieties of assertive force
we have now seen there to be (all the ways of counting a way for things to be as a
way things are). It would just be another particular way of so counting things; one
to be applied where very special ends were to be served. And, as we saw in sec-
tion 6.3, classifying things according to the reach of some way for things as such
anyway draws on the full resources of a Cartesian thinker. So such abstraction, if
possible at all, does not relieve the unthinker of a burden. Nor is the burden one
the unthinker would have the capacity to discharge.
Perhaps, then, the task of force-selection is performed for him/it. We already
saw one idea for this: deference to vehicles. We saw already that that idea cannot
work. So perhaps the work is done by whatever thrusts allorepresenting on our
unthinker in saddling him/it with (producing or being) some vehicle. Putting
things in terms of force, though, brings out a point of Frege’s. It is that no vehicle
as such can impose a force on any representing (a version of the point above that
what gives force to representing-as cannot be just more representing-as). To put
the point one way, any way for an instance of ‘Pia ordered oursins’ to fail to be an
assertion is a way for ‘It’s true that Pia ordered oursins’ to fail to be one. When
assertive force is absent, ‘It’s true’ will not restore it, nor will ‘I assert that’, nor
any other form of words. Mutatis mutandis for any other force. In another ver-
sion Frege tells us that there is no assertive force “when the required seriousness
is missing”. Seriousness is not conferred by a vehicle. What is in question is the
seriousness with which it is produced. A thing cannot produce the necessary ser-
iousness by having representing thrust on it. It is the (would-be) representer who
(which) must be serious. This, as we have seen, the unthinker cannot be on its own.
This idea of abstraction, and of resulting generic forces, thus leads nowhere.
Might the unthinker then, perhaps, represent things as ways they are or not
(though not itself thereby representing truly or falsely), while doing so with no
force? When might a vehicle be produced forcelessly? A rhythm poet, or dadaist,
might produce English sentences simply for their sound—the sentence ‘Red balls
roll’, say, simply for the way it rolls off the tongue. Or a graffiti artist might spray
such a sentence on garage doors for its elegant shape. Most red balls probably do
roll, if not made of glue. Such is not what (if anything) makes the poet right. If his
interest in sound is pure enough, then while he wrote a sentence which speaks of
a way for things to be, he did not thereby engage in any representing-as at all. And
if, as we stare at the garage door admiringly, a red ball rolls by, well, what a coinci-
dence! But it is just a coincidence. As some philosophers have it, in (a) perceptual
experience the world is represented to us as a certain way. If we see a pig under an
 The Preserve of Thinkers   161

oak, say, then perhaps as such that a pig is beneath an oak. But if this representing is
conceived as forceless, then it might equally well represent things as any other way,
say, as such that cool waters run deep, or Pia drives a Porsche. Experience’s so rep-
resenting things may mean (factively), effect-represent, or indicate, or make likely,
that a pig is beneath an oak. But if the representing here is forceless, then it is not
through its content that such meaning is effected. It is not as if a reason thus created
for thinking a pig beneath an oak might be that experience, or this representing in
it, might be right. Whether it is right or wrong cannot matter here: without force
there is no way for it to be either. Representing-as thus cancels out. Representing
Porsches as fast would do as well as representing a pig as beneath an oak for nature’s
signal that a pig is beneath an oak. For what meant in this sense of meaning, to
represent things as the way it means they are would just be a curious accident.
Force is part and parcel of the step by which we move from the psychological to the
logical, from mere effect-representing, or its relatives, to that three-place relation,
representing-as.

6.8. Finding and Presenting

There is finding, or marking, instancings, or the instancings, of some given way for
things to be; and there is presenting things as some given such way. The one thing
is not the other. But some might hope for the second to emerge out of the first. This
section explores that idea.
An unthinker lacks capacities which, so far, appear essential for allorepre-
senting; any capacity which might permit that leap from the psychological (or
its counterparts) to the logical, from the psychology of holding forth, to the logic
of being true, which allorepresenting is per se. How, then, might unthinking
representing-as ever be thought a possibility? One prominent idea is that allorep-
resenting might emerge out of (the maintaining of) patterns of effect-representing,
aided, perhaps, by the point of maintaining them. The allorepresenting would be
by, or in, that in which such patterns were maintained. This section explores that
idea.
The simplest patterns are generalizations, the simplest generalizations univer-
sal. Those empty seats in the theatre (or their presence) effect-represent casting Sid
in the lead role. Empty seats in a theatre might always do that, if Sid got around
enough. Or, to complicate things, they might usually, or normally, or (other modi-
fier) do so. So far, it is the presence of those empty seats which does the represent-
ing. If allorepresenting emerged here, what would do it? Frege writes,
No one can be prohibited from adopting any arbitrarily occurring event or
object as a sign for whatever. (1892, p. 26)
Empty seats could be appropriated as a sign that Sid plays the lead. They would
thus be a vehicle of representing-as. What would do the appropriating in the
 Against Strong Content
162   

allorepresenting that emerged here? Whose vehicle would it be? It is sometimes

thought that the question needs no answer; vehicles can get on on their own. We
have seen reasons to think that idea leads nowhere. Bracket them for the nonce.
Such simple patterns are no improvement on the individual case. They put
nothing in play not already there. Two problems. First, if we read nature’s messages
well enough—if we know what empty seats would mean (effect-represent), then
their sight gives us good reason to suppose that Sid is in the lead. If allorepresent-
ing emerged from their effect-representing, then, recognizing this last, we might
also recognize that allorepresenting for what it was. It, too, might give us reason
to suppose that Sid is in the lead. But no different, or other, reason than we had
simply in recognizing the effect-representing; that we would have had if blind to
the allorepresenting thus emerging. This allorepresenting of Sid as in the lead is
no better reason to take him to be in it than that given by the effect-representing
from which it emerged. Whereas allorepresenting is the sort of thing, by nature,
for giving reason of a new and distinctive sort. With this we have Sid’s word for it.
If, as a rule, when Sid says, ‘Roses are red’, this effect-represents a spent compressor
in Pia’s Porsche, the reason this gives to think her Porsche thus in need depends
on the reasons, if any, to think this case of Sid’s speech an exception to the rule.
If we have his word for it, the reason this gives depends on the extent to which
Sid can be counted on to be speaking sincerely, judiciously, knowledgeably, where
the point is not that such things, if they occur, might effect-represent a defective
compressor.
If Sid tells Pia that her Porsche needs a new compressor, there are things one
might recognize his performance to effect-represent without yet recognizing what
allorepresenting thus occurred, or that any did—e.g., successful speech therapy.
That Pia’s Porsche needs a new compressor is not one of these. If nature does hold
such a message in Sid’s performance, we would have to be much better at reading
nature’s messages to get this one than we need to be to recognize Sid’s allorepre-
senting for what it was. But if do now recognize it for what it was, then Sid’s word
for it may give us a reason we did not already have. We need only recognize his
word as good. He represents himself as having settled the issue. We need not our-
selves know how such issues are to be settled (independent of taking someone’s
word for it.) We may nonetheless see in Sid—in his integrity, sagacity, etc.—that
he is doing what he purports to be doing: giving his word for what he has settled.
What we thus see about Sid is not likely to effect-represent Pia’s Porsche needing a
new compressor. But if it is how things are, then the Porsche is thus in need.
(A related point. Where there is allorepresenting, what it is and what it
effect-represents are absolutely independent. Sid’s telling Pia that her Porsche
needs a new compressor may effect-represent a chequered (or chequer-flagged)
past, or a bad hangover—and it may do this regularly, or normally, or usually—
without his thereby allorepresenting things as these ways.)
The second problem in brief. In effect-representing Sid in the lead role,
those empty seats represented a certain instancing of a certain way for things to
 The Preserve of Thinkers   163

be: such that Sid was in the lead. If they did that, they ipso facto effect-represented
the instancings of countless other ways for things to be—countless co-denizens
of clouds within the conceptual that Sid being in the lead inhabits. Such would
remain so if empty seats always effect-represented Sid in the lead—wherever, that
is, nature is so arranged. Effect-representing does not perform the selection task.
Nor would generalizations of the kind just scouted.
But if allorepresenting has not yet emerged, perhaps we are looking at too
simple patterns. Here is another idea. Sometimes we can say: A effect-represents
(or would effect-represent) B if all is/were going right. That needle on the gauge
effect-represents the tank’s being half full if all is going right. Those hands on the
dial (or their present position) effect-represents its being 10 o’clock local time if
all is going right (if Sid remembered to reset his watch). Such patterns, if any,
are those from which someone might think allorepresenting could emerge. I will
expand that idea.
There are designs which, if realized, would make for A effect-representing
B. A device, or mechanism, or system, or (perhaps) phenomenon may realize such
a design. Man, or nature, may provide such devices (or etc.). (Examples above.)
Sometimes by this design, sometimes, perhaps not, the device produces, on occa-
sion, a certain outcome (or type of outcome), A. The output may be a product—a
signal, say, or an effect—or it may be the device’s going into a certain state, A (or
of type A). When such a device is working as per design (as any device is liable, on
occasion, not to do), A, as thus produced, effect-represents B (what it was designed
to). A would occur only if there were B to thank. Such a device may be for real-
izing this design in this sense: but for some need thus to connect A and B, the
device would not have been created. I will call such a device a B-detector: where
the device produces A as outcome, you can bet on B’s occurrence (if all went as per
design). Bs are historical occurrences, e.g., of porcine presence. What the device
detects is thus particular instancings of things being a certain way, e.g., such that
a pig is present.
Depending on what it was detecting, a detector might need to exploit com-
positionality. If, say, it were detecting where animate things were, it would need
indefinitely many dedicated outcomes for each of indefinitely many arrays of loca-
tions relative to it at which there might be such things for it to detect. Composing
outcomes would be called for. For present purpose such changes nothing.
A design might also be for (as I will speak) locating (cases of) B: not just for
making it so that A, if it occurred, would effect-represent B, but also for making it so
that, all working as per design, B, if suitably occurring would be effect-represented
by A. So if B is (suitably) occurring, you can bet on A, again all going as per design.
Now the hope is, either for detectors, or for locators, that, in detecting, or locating,
as the case may be, they will also be representing things as being that way whose
instances are thus detected or located.
On one notion of device, a device for detecting (or locating) would work in the
first instance by responsiveness to proximal provocation. It would effect-represent
 Against Strong Content
164   

such provocation. If allorepresenting were to emerge from its workings, one might
see it as simply representing things as being such that there is such provocation—
in this sense simply representing the proximal. But in most cases of interest, the
device will be for detecting, or locating, the distal, e.g., porcine presence. When it
works as per design, it will effect-represent those distal things too. And the hope
will be that those ways for things to be whose instancings it detects or locates will
be the ways it allorepresents things being. What is thus responsive to the distal
in its responsiveness to the proximal is inherently subject to what would be for it
ringers: there might be the right proximal provocation without the wanted distal
happenings, and vice-versa.
Our samples, so far, are manmade. But such devices can be natural. Pigs chew
straw, it is said in some parts, when it is about to rain. In perceiving, it is sometimes
said, we experience the world being represented (to us) as being thus and so. What
of this case? Pia sees the pig before her. Her doing so is thanks, inter alia, to there
being one. So it effect-represents there being one. Thus far all is in place for her to
be a pig detector, her seeing a pig being the outcome which would be detection.
But it takes no design to realize this connection, a fortiori not one which works via
responsiveness to the proximal. There is no such thing as her seeing a pig failing to
effect-represent presence of that pig. So, though I have tried to remain ecumeni-
cal on the crucial point here (not to anticipate routes by which someone might see
representing-as as emerging), Pia’s case to be a detector by virtue of her capacity to
see pigs sits ill with our present notion of a detector—one designed to fit the intu-
ition that nature might make representing-as emerge.
Design comes into the picture here when it comes to locating cases of porcine
presence. What needs explaining is not that when Pia sees a pig there is a pig, but
rather that she does such a thing at all. Indeed, Pia (if adequately sighted) is such
that, by design, when a pig is before her (and she is looking), all going well, she will
see it. Such is no thanks to her responses to proximal provocation (though it may
depend on some processor’s responses). Nor, correlatively, is this capacity exposed
to ringers: the indicated outcome with no pig before her. Nor to a ringer for no pig
(hence a pig) which she experiences visually without thereby seeing a pig.
If Pia sees a pig, then, though this does effect-represent porcine presence,
such does not bear on the reason she thus gains for taking there to be a pig before
her. Her reason gained is not given by the fact of her seeing a pig, but rather by
what she sees—a pig before her—which she can recognize as a case of a pig being
before her (as being such that a pig is before her). As Frege notes, such recogni-
tion is a function of thought, not vision. What is gained is reason so to suppose—
nothing short of proof that there is a pig—which need not consist in reasons so
to suppose. Such reason gained, considerations of effect-representing could add
nothing, could not so much as be reason so to suppose. If representing-as emerged
from this effect-representing, it would offer exactly that much.
Things would be different if for Pia to see the pig is for her to be in, not just
that state, but some other visual state—one, say, of being appeared to thus—which
 The Preserve of Thinkers   165

she could conceivably be in even were there no pig. Then it needs explaining how
this state in fact effect-represents a pig before her (when it does). A design is
called for; one whose realizing would maintain the right relation (all going well).
A design for given proximal responsiveness seems indicated. With this state as
the indicated outcome Pia fits the model of a pig detector. If allorepresenting
ever does emerge out of effect-representing conditions might now be ripe for
this. Such allorepresenting might even have a point. It might give Pia reason to
suppose of what this state is visual awareness of that this time it effect-represents
what it ought, porcine presence (though it is hard to see how this could ever be
good reason). Reason the more, I suggest, to find this a bad picture of perception.
For present purposes, though, I simply mark this as the picture in which the idea
of representing-as in perceptual experience might look promising.
Back to the general question how representing-as might emerge. What I now
want to stress is the distance by which detecting instancings of some way for things
to be, or locating its instancings, falls short of presenting things as being that way,
or placing things under given generalities—as one does in representing-as. Such
will bear particularly on the second problem above. A good way into the matter
is the following. A detector (or locator) works according to a particular design;
a design for detecting whatever it is that it detects. It works via the proximal,
and it works in a particular way. A pig detector, e.g., may be sensitive in its out-
comes to a pig’s distinctive snout: whether it produces that outcome which is to
effect-represent porcine presence depends on whether it has detected (or done
what, for it, ought to be detecting) such a snout. And it detects such snouts by
marks of such which, by design, would be proximally accessible to it, once again,
all going well. But by what marks, or means, such things as pigs or Porsches, yel-
low or snores, are detectable depends on the environment. Equally for anything
else liable to be detectably present or not in the sublunary world we inhabit. So a
condition for a detector, or locator, so much as being that is that it work in an en-
vironment hospitable to its ways. This point entered the picture already, as we saw,
with the very idea of proximality.
You can tell a pig by its snout. You can rely on this given how things are. But
no one supposes that to be a pig just is to have such a snout. Plastic surgery alone
rules that out. To be a pig is not just, and not per se, to look some particular way.
So in the wrong environment (e.g., too much cosmetic surgery), a pig-detector
which worked by means of snouts simply would not be a pig detector. We can rec-
ognize when it would not be. We can look at its workings, on the one hand, and,
on the other, at the reasons for and against counting what it would be identifying
as pigs (if detecting at all) as pigs. We can exercise our capacities for retrospection,
capacities to be abashed. An unthinking detector has no such capacities, cannot
be abashed.
The unthinking detector lacks a capacity we have: sensitivity, case by case,
to what bears, or might, and how it might, on whether that case is to be counted
as one of instancing any given generality—notably, here, that generality whose
 Against Strong Content
166   

instancings are to be detected. As noted long ago, such capacities are reserved for
Cartesian thinkers. An unthinking detector thus could not be sensitive to what
distinguishes this way for things to be, such that there is a pig beneath the oak,
from countless others, notably, others inhabiting clouds around it. This does not
matter to detection, but certainly does to presenting. For presenting things as
being such-and-such way, a selection task must be achieved. That way things are
presented as being must be distinguished, in the presentation, from its fellows
within conceptual, notably those overlapping with it through some proper part
of the space of particular cases, but diverging from it in other proper parts. Such
selecting is beyond the powers of an unthinking detector (or locator). Mere detec-
tion does not demand it. Assume the detector/locator in a hospitable environment
where he/it always gets things right. If it is representing things as some way it is
detecting in this environment, such fixes something as to how that way reaches. It
reaches to these cases. But such is only a proper part of its reach. The unthinking
detector/locator has no capacity to see when it would have left a hospitable envi-
ronment or how, or, when it has, what detection then would be. It thus cannot do
what any thinking allorepresenter can. Which robs us to our right to suppose that
the notion environment hospitable to its detecting has a determinate sense or appli-
cation, that it is so much as fixed what a hospitable environment for it would be.
A detector might fail either through migration into hostile territory, or thanks
to some one-off ringer. If he/it represented things as those ways whose cases he/it
was thus detecting, such would be cases of representing falsely. But the unthinking
representer cannot represent itself as in a hospitable environment. It is blind to when
this would be so. The Porsche detector which blinks each time a Porsche passes, put
in the world of knockoff Porsches, can do no more than carry on. Its detection work
in no way equips it to approach the question whether it is in a knockoff world. That
it is not cannot be how things are according to it. Nor can it be held responsible,
where ringers are about, for whether what it blinks to is a Porsche. Such thus cannot
really be, in such circumstances, how it represents things being. If it represented-as
at all, it might just as well be that those instancings to which it continues blinking
are just those of the way it represents things being.
Detection buys the unthinker no standing in the realm of allorepresenters,
since it does not equip him/it to effect selection. With which the idea of allo-
representing emerging out of detecting, or locating, collapses. What qualifies an
unthinker as a detector is not what could qualify it as a presenter, or placer, so as
allorepresenting-as (as presenting something as falling under some generality).
Allorepresenting cannot thus emerge.

6.9. Collapse

A pattern which made for detecting (or locating) instancings of some bit of the
conceptual would not thereby make for presenting anything as falling under any
 The Preserve of Thinkers   167

generality. If representing-as emerged from it, so far that representing might as

well be anything. Suppose we decided, though, that such representing must repre-
sent things as some way whose instancings are being detected. We might then try
to say: it must represent things as that way whose instancings are being detected
(or located). If it detects yellow things, the idea is, then it represents things as
such that there is something yellow. But that move is illegitimate. The detector
detects in its actual environment, one hospitable to such detection. By our deci-
sion it thus identifies a proper subpart of the reach of the way it represents things
being. Nothing it does as a detector extrapolates from this sub-reach to the whole
reach of that way it represents things. Otherwise put, nothing in the pattern it
incorporates determines when it would have moved into a world of ringers (or
what a ringer for what it was detecting might be). Thus is the move just tried on
illegitimate.
In its actual environment the detector detects/locates yellow Porsches. It is
endowed by design with its distinctive outcome for suitable encounters with them.
Now it enters a world full of yellow silicon dummy Porsches. There is something
yellow Porsches and such dummies both are. I will call it being a siliporsh. In its
actual home, it detects and locates yellow Porsches, and it detects and locates yel-
low siliporshes. In its new home it detects and locates siliporshes, and detects, but
no longer locates yellow Porsches. For it, the dummies are ringers for Porsches.
Which of the ways whose instancings it was detecting all along (if either) would be
the way it represented things as being (if it were to do this at all)? If it is to repre-
sent things as a way whose instancings it detects, has it now encountered ringers
for the relevant such way? Nothing in its design, or in the patterns whose incor-
poration this maintains, decides this. Patterns of effect-representing thus cannot
perform the selection task for it.
Back to the actual. Our detector produces its assigned outcome on encounter
with Pia’s yellow Porsche this morning. It produces this outcome again for her
Porsche this afternoon. But has it encountered a ringer this time? All those cases
as an object’s being as it was which occurred up to noon today form a subregion
of the space of all particular cases. Pia’s Porsche being as it will be this afternoon
lies outside that subregion. Three classes of ways for things to be trace that same
path through the subregion: ones which go on to reach to her Porsche being as it
will be this afternoon (being a yellow Porsche among these); those which fail to
reach; and those which determine no outcome for this novel case. If you wanted
to detect instancings of some way in the second or third class, you might well rely
on a detector of the present kind. You might well be prepared to count this case
as a one-off ringer, a momentarily inhospitable environment, especially if it were
trouble to guard against it. Call this way being a yellow Porsche*. Then the present
device is a yellow Porsche* detector, on a suitable understanding of inhospitable.
If some natural function is served by equipping us with yellow Porsche detec-
tors (perhaps preservation of the species is furthered by designing females to be
attracted to them), that function is served as well (up to insignificant differences,
 Against Strong Content
168   

unforeseeable at time of design) by a yellow Porsche* detector. Up to noon today,

the detector has been ‘trained up’ on cases of both being a yellow Porsche and
being a yellow Porsche*, just as up to its entrance in the world of silicon dummies
it had been ‘trained up’ on cases of both being a yellow Porsche and being a yellow
siliporsh. Once again, what a detector is detecting could not tell us as what way it
represented things if it went in for that at all.
So far we have been supposing that a detector which allorepresented would
allorepresent things as some way whose instancings it detected. Why should we?
Suppose a weather bureau detects the weather. When the temperature had dropped
to 14˚ it would detect that. Its signal—the outcome reserved for this—might be
some bulletin, ‘Current temperature 14˚.’ But it might also just as well be ‘Wu’s
bird’s nest soup is legendary’, if its first mission is to promote tourist trade and it
sees this as currently best means for that. Or a detector designed to detect temper-
ature without windchill might, in fact, detect temperature with windchill. Might
not such a detector still allorepresent (falsely) temperatures without windchill, if
it did allorepresent at all? A capacity to allorepresent, one would have thought,
makes room for such possibilities. Nothing closes them off in our case except our
capacities, as thinkers, to recognize what projects we engage in, and their respec-
tive successes and failures. What would shut them down in an unthinker’s case
remains obscure. Anyway, not mere blindness to the options.
Nor should we allow ourselves to be impressed by the fact that we think, e.g.,
in terms of Porsches and not siliporshes, or being yellow and not being yellow*. An
unthinker’s selection task is not thus to be foisted off on us. There is no reason to
suppose the unthinker to share our sense for what the thing to do, or, specifically,
the way to represent things, would be. There is every reason not to. The unthinker
has no such sense. Nor, as we are about to see, could any natural function be served
by nature’s arranging for the unthinker’s representing to coincide with what ours
would be.
Where nature incorporates a design for detection (or location) in one of its cre-
ations, the purpose thus served—say, furthering procreation—is always found in
our actual environment (or that at time of incorporation). It is just part of designing
for interaction with an environment that one cannot design for immunity to ringers.
But nature’s purposes are served as well as anything could serve them by designs
which are not so immune. If allorepresenting emerged from such design there could
not but be the problem just scouted. In one form the problem is what to count as a
ringer, or when to count one as having occurred. Something whose instancings are
detected by design is to be the way things are thus represented as being. A ringer
would be a ringer for that. But nothing in the design choses any one such thing. Nor
does the purpose such design might serve. That same problem is now familiar in
another form. Nature designs for the actual environment, thus for a subpart of the
reach of whatever way the instancings of which might thus be detected. For allorep-
resenting to emerge, a move must be made from this subpart to a whole reach. But
such moves are no concern of nature’s at all.
 The Preserve of Thinkers   169

So things stand with the second of our two problems. I return now to the
first. I approach it first through this question. If allorepresenting arose out of
some pattern of effect-representing, with what force would this be done? An
unthinker does no autorepresenting (I here bracket dogs and cats). Where there is
allorepresenting, there is what makes it recognizable, its vehicle. Here this is to be
some occurrence—some production or its product—which instantiates the pat-
tern, is produced by a design for maintaining it. It is to be, in present terms, what
is reserved by design in some detector for signalling detecting of some instancing
of the way as which things are thus represented. The problem I will scout now
arises from the fact that, to speak a bit loosely, all the information carried by the
(supposed) representing is carried already by its vehicle.
The occurrence of this vehicle, like any occurrence, effect-represents all that
to which it owes thanks. If all went well—if it was produced by design, and the
environment was, even locally, hospitable (no ringers)—then this occurrence
effect-represents what the design is a design to detect. On a given understanding of
going well, hospitable, ringer, and so on, this would be the instancing of some way
for things to be. So the occurrence of the vehicle is liable to give reason to think
that there was such an instancing. It gives precisely as much reason to think this
as there is then reason to think that all went well, surroundings were hospitable,
and so on—conclusive reason, plain proof, where such things are not in doubt. The
vehicle itself (properly, its occurrence then) has that much significance. There is,
anyway, this to be recognized as to what to think and do. So much is recognizable
to one blind to the fact that the vehicle is a vehicle of allorepresenting; to what
allorepresenting there thus was, or to there having been any, so long as he recog-
nizes the vehicle as produced, as it was, by what incorporates such a design—by
what would, if working well, etc., thus maintain that pattern of effect-representing
which it is a design for maintaining.
Now suppose allorepresenting to have emerged in this operation of what pro-
duced the vehicle (the detector). What reason does this allorepresenting give for
thinking and doing? How does it, or its occurrence, bear on what the thing to
think, or do, would be? If it emerged from the pattern, as per above, then for it
to have occurred is for it (or its vehicle) to have been produced in the maintain-
ing, by design, of the relevant pattern—one such that, things going well, etc., its
(or its vehicle’s) production would effect-represent the instancing of what is thus
effect-represented. It would have occurred on just that condition on the vehicle’s
effect-representing such instancing. To recognize it as the representing it is is to
recognize it as just such a designed production. So its occurrence, if recognized for
what it is, gives just the reason to think things the way it represents things being as
the vehicle’s occurrence itself gives for thinking this. And if it really thus emerges
from the pattern, its mere occurrence can give no more.
Whereas it is essential to allorepresenting that it (or its occurrence) can bear
on the thing to think (or do), give reason to think things one way or another
(where they are represented to be F, that way) which there would not be anyway,
 Against Strong Content
170   

without supposing it to have occurred, or which might recognizably be present

without recognizing it to have occurred. When it comes to reason-giving, allorep-
resenting cannot be thus inert. If our yellow Porsche detector did what would here
be (emergent) representing there as being a yellow Porsche, what it did, whether
that or not, would anyway give as much reason to think there was a yellow Porsche
as there is to think that all was then going well with it. For its (supposed) allorepre-
senting to give, and be able to give, precisely and only this much reason is for there
to be no representing-as here at all.
Pursuing our question about force is one way to see why. For allorepresenting
to have a given force is for it to be taken as aimed at particular successes (and for
it to represent itself as a success in some of these). If the force is assertive, there is
the success, representing things as they are. If it is imperative, there is the success
obligating so-and-so. And so on. A force is fixed, and is identified by, the successes
thus aimed at. All the more if forces are as multifarious as above suggested. From
this perspective, one isolates the force of Sid’s words to the waiter only in isolating
in what ways what projects would be served if things are as he said. It is by, and
according to, its force that given allorepresenting bears as it does, and not in other
ways, on questions of the thing to think or do. If Sid had said, ‘Pia ordered the
oursins’, and continued, ‘That’s what you think, isn’t it?’ then, though he would still
have represented Pia as having ordered the oursins, his doing so would not have
borne as it did on what the waiter was to do. The force of an allorepresenting is
recognizable in how it bears. Where it (or its occurrence)may be taken to bear in
one way or another on questions of the thing to think, facts of how it is to be taken
may be rich enough to choose some force, from among the panoply of options, as
its. For our yellow Porsche detector, its would be allorepresenting (on the present
scheme) could have no bearing on the thing to think or do. The bearing there
would be on such questions if it occurred would be just that which there would be
without supposing it to have any. In the facts of how its supposed representing is
to be taken, then, there can be nothing to choose between one force and another.
So it can have none.
If such allorepresenting can have no force, then (as we have already seen in
one way) it can have no content either: there can be no such thing as the way
it represents things as being. For, again, what content some allorepresenting has
cannot matter independent of its force. The Porsche detector might signal Porsche
detection in given words, say, ‘Porsche ahead’. But that those words mean Porsche
ahead is irrelevant to their function. The words might as well be, ‘Pigs whistle.’ All
that matters is that those words, whatever they are, and whatever they mean, are
the detector’s response-by-design to the presence of a Porsche, so that if things
are going well there will be a Porsche ahead. The words might happen to speak,
in English, say, of a Porsche being ahead. But for their purpose here they might
as well speak of anything else. It thus cannot be that the detector uses them to
speak of that (to represent things as that way). Whatever the words are, they are
merely recruited by the detector’s design to stand as a synthetic addition to nature’s
 The Preserve of Thinkers   171

messages. From their occurrence one may (sometimes) conclude that there is a
Porsche, as from the pawings in the dust on the trail ahead one may conclude that
the wild boar are in rut.
The difficulty with emergent allorepresenting has been, so far, that what is to
make this representing recognizable as what it is is identical with that in it which
gives reason for supposing things as to how things are, or are to be. Whereas allo-
representing is always a source of new and distinctive sorts of bearing on questions
what to think or do. It is a source of reasons to think things that an unthinker could
not give. In spelling this out a bit, I can reinforce and deepen the points just made.
Suppose that Sid tells us that a Porsche is in the drive. His doing so might,
by any of countless routes, effect-represent the presence of a Porsche in the drive.
(For example, perhaps in his seeing Pia chatting with the countess, by the French
doors leading to the garden.) It could, but need not, effect-represent his ability to
tell a Porsche at sight. Such things may be nice if so. But they are not what make
his representing recognizable. No pattern of effect-representing makes for his rep-
resenting, nor for it being recognizable. What makes for his representing, so for its
recognition, is independent of any patterns of effect-representing in which such
performance may stand (Chomsky’s point). Which makes room for Sid to achieve
recognition through choosing means according to his insight into that audience
(to which he belongs) for which (in the sense of section 6.4) he allorepresents; ac-
cording to that audience’s shared sense for how such means, deployed then, would
be to be taken.
What makes Sid’s representing recognizable is thus also independent of that
which gives it the bearing it does on what to think and do, which makes it the
reason it is for thinking this or that. If Sid represented a Porsche as in the drive,
there would be a Porsche in the drive, provided that he was then executing a proj-
ect of seeing how things were, knowing how to do so and when he would be. Where
this condition is recognizably met, his so representing things gives conclusive
reason to take a Porsche to be in the drive. More generally (in parallel with that
supposed emergent representing-as by a detector) Sid’s saying so gives as much
reason to think so as there is reason to think the condition met. Notoriously we
do not always tell the truth; nor are always able to tell whether we are doing so
or not. But being serious (one’s project being one of saying how things are) and
knowing what one is talking about is something we, sometimes, can manifestly do.
So, sometimes, we can recognize the reason Sid’s saying so gives to take there to
be a Porsche in the drive in recognizing the reason there is to take him to be thus
engaged. Sid’s saying so then gives us such reason.
For Sid’s representing to give the reason it thus does for thinking a Porsche in
the drive, it need participate in no fixed, specifiable, pattern of effect-representing
(a pattern maintained in Sid in some given way). Nor is it in recognizing such a
pattern that such reason is recognizable to us as given. We need know nothing of
such things. Nor could such reason emerge from such participation. So nor could
it be such participation which made it recognizable to us as given. Thus is the
 Against Strong Content
172   

reason of a distinctive sort, a sort which, by nature, (non-auto) representing-as

makes room for.
Sid can give us, in his representing, such reason to think things because,
and only because, he has the capacity to see when a question, or at least the rel-
evant question, has been settled. Such it is to know what one is talking about.
An unthinker, a mere detector/locator, is, a fortiori, denied any such capacity.
For such a capacity transcends any mere design for recognition in just the way
that the reach of a way for things to be transcends those cases occurring in any
given hospitable environment. Sid can tell a pig at sight, as most of the rest of
us can. But you cannot always tell a pig by looking. A capacity to see when it
is settled that it is a pig is a capacity to take such things into account. By the
same failing, an unthinker could not separate what makes its allorepresent-
ing recognizable from what makes it the reason-giving that it is; and (thus)
could not make that reason-giving of the distinctive sort allowed for by allo-
representing in being what it is. The unthinker’s representing-as would thus
be idle. It could not add to the reason-giving there would be without it, just
in nature’s messages. But, for reasons given (above in discussing force), intrin-
sically idle representing-as would be no representing-as at all. In a different
context Wittgenstein said,
Symbols that are dispensable have no meaning. Superfluous symbols signify
nothing. (Waismann 1979, p. 90, emphasis in original)
Such fits the present case exactly. In 1922, telling us that an unneeded sign is mean-
ingless, he offers a reverse side to the coin:
If everything behaved just as though a sign had meaning, it has meaning.
(1922, 3:328)
Perhaps those who propose unthinking allorepresenting (or representing-as)
think they have found what does behave just like the real thing. If so, one might
invoke this reverse side of the coin. But above are reasons why they have not.
Allorepresenting is a complex pattern woven into the fabric of our thinking
worldly doings. For it to be what it is is for it to serve as it does the execution
of our projects. Unthinking allorepresenting would detach this pattern from the
fabric, patching it into some simpler activity, still recognizable there, perhaps
in more primitive form—still allorepresenting. But events, or occurrences, of
representing-as are creative: for each, there is a new way for things to be, things
being as thus represented. Only within that original whole fabric does it makes
sense to speak of creating generalities under which for things to fall. Only there
could doing so be serving the needs thus to be served. It is precisely there that it
makes sense to think of a way of issuing, or bearing, messages which relate the
issuer/bearer to a term, such as the third term in the relation allorepresent, which,
belonging to the conceptual, does not interact with an environment.
Frege notes that thoughts (so ways for things to be)
 The Preserve of Thinkers   173

are not thoroughly without effects, but their effectiveness is of a wholly other
sort than that of things. . . . their effects are triggered by the doings of think-
ers. . . . (cf. 1918, p. 77)
Instantiating, thinkers’ doings are what it takes to place thoughts in allorepre-
senting. Instantiating again, perceptual experience is nothing like being repre-
sented to.

6.10. Afterword

The question this volume poses is whether perception has content. I have under-
stood content as the content of representing-as (representing something as some-
thing), and then that for some representing-as to have the content it does is just for
it to represent things as being as it does. I have then tried to answer the volume’s
question by answering a broader one: Where might there be representing-as at all?
The answer I have argued for is: Only a thinker (and that by which he executes his
projects of representing) can engage in representing-as.
How do I suppose this to answer the volume’s question? First, I suppose
perceiving to be a passion in this traditional sense: it is something we undergo;
something imposed, inflicted on us, something we suffer (like the party next
door at 3 a.m.). Accordingly, I suppose that if there were content in perception,
it, too, would be part of what is inflicted on us, or integral to its infliction. The
point of this is just that it would not be the content of our representing-as by way
of responding to what we see—e.g., responding to the monkey with the tin cup
before one by taking there to be a monkey before him. And, if the perceiver’s own
candidature for the role of representer is thus scratched, no other thinkers appear
in the vicinity to do the needed work. That perceiving (or even, more broadly,
perceptually experiencing) is a passion in this sense has been challenged. I will
return to this. Anyway, such is the strategy.
If I am right, then seeing perception as with content is something like seeing
one’s teacup as (literally) having it in for him. Why, then, should this idea about
perception be as immensely appealing as it seems currently to be? One short answer
would be sociological (or epidemiological): things go around. A short philosophical
answer would be: inattention to details, or, more generously, a congeries of gram-
matical illusions. I except from such answers one variant of representationalism on
which a positive answer is portrayed as the only solution to a certain how-possible
question (in the spirit of Kant). For the moment, though, I bracket this exception.
A first question to ask, then, is: Just where is content supposed to be located in
perception? There are, I think, three main answers: first, in the objects of percep-
tion (or of perceptual, e.g., visual awareness); second, not there, but as otherwise
presented to us in our perceiving (the inflicting on us of the visual awareness thus
enjoyed); third, in certain information-processing mechanisms or their states. On
 Against Strong Content
174   

the first two answers, awareness of the relevant representing, or at least affordance
thereof, is part of the bargain in perceiving. So a perceiver should be prepared to
recognize (perhaps with the help of a bit of elicitation) just how, in his experience,
things were represented as being (or at least when things are, when not, as so rep-
resented). This idea is mandatory insofar as the representing here is supposed to
be a source of knowledge, or, again, deception. On the third answer the perceiver
has no, or no direct, access to the relevant representing; if he has access at all, it is
inferential.
Putting last things first, motivation here is often in part reductionist. The core
idea: for something to look, e.g., yellow to me is for something intracranial to be as
it would be if registering something’s actually looking yellow—thus to be in a cer-
tain representational state. It is anyway implausible that such a state would be one
of representing the world as being thus and so (e.g., something in it as being yel-
low). The comparison here would be with a digital camera. Perhaps a file contains
instructions for recreating the image recorded on it. But the camera merely regis-
ters, does not represent. Nor does either it or the file represent the world around
it as being thus and so. In any case, the answer to my more general question rules
out such locations for representing-as.
As for objects of perception, it was, perhaps, easier to take them as engaged
in representing-as in the days when one could, with a straight face, maintain that
those objects (e.g., of vision) did not cohabit our world with us—not any such
thing as a champagne flute or the bubbles therein. Objects of perception literally
out of this world might invite a certain Lockean ploy. Few would hold, though,
that that penguin on the rock, or teacup on the table, represents anything as being
anything. So, with the demise of the view which invites such a ploy, it becomes im-
plausible that the content of perception is so located. (Though, disclaimers to the
contrary, reports of the demise of this view may be premature.)
The remaining view, then, is that while the representing-as in perception is
not itself an object of, e.g., visual awareness, in being visually aware of what we
are (in seeing what we do) we are also, ineluctably, aware of things (notably, in
the visual case, the scene before our eyes) being represented as being thus and
so (where it is not ourselves who are doing so). Generally speaking (perhaps not
always) the content of the representing here is thought of as, so to speak, indexed
by the way things look to the perceiver. For example, it is not going to be that,
while the sky looks blue to me, my experience represents it as being red.
If representing-as could not diverge from looks in such ways, such would be
a sign of mere idling—unless the point is to answer a certain how-possible ques-
tion: How could perception make the world bear, rationally, for the perceiver on
what to think? How could it make us knowledgeable? How could we ever, e.g.,
know that that monkey has a tin cup because we can see the cup? The obstacle
here is a doctrine. It concerns rational relations (relations by which things can
bear on what is true). The idea is: only representations can stand on either side of
a rational relation. For, it might be said, rational relations come on the scene only
 The Preserve of Thinkers   175

with truth thus properly in the picture. For familiar reasons (and more to come),
it is implausible that representation could really dispel such a how-possible worry
(were there one). In any case, the right way with the worry is to jettison the doc-
trine—for doing which Frege provides compelling reasons. (How could there be
any truths unless some things not fit for truth could bear on whether what could
be true was so?)
A last idea: perhaps seeing a monkey with a bellman’s cap and tin cup is not a
passion in my initial sense. Thinking that one faces a monkey is, per se, representing
things to oneself as such that such is what one faces. Similarly, the idea is, seeing the
monkey standing is, per se, representing things to oneself as some given way. But
there is less than no reason to think this. As Frege notes, ‘see’ in ‘see that’ is not a verb
of perception. The monkey is (now) on the table, too near my glass. That there is a
monkey there is neither on the table, nor in any other location. It is hence, as Frege
puts it, not something which can form images on retinas. Seeing as is sometimes
perceptual, sometimes not. (A locution with many uses.) But in any case seeing a
monkey as a monkey is certainly not something done routinely, or merely in seeing
a monkey. There is no reason to think that any such thing is intrinsic to perception.
It is easy enough sometimes to see things where they are not. It comes easily to
many to see representing where it is not, namely, in perception. The prophylactic
against this is to mind one’s grammar. This, though, is only to be done case by case.
I hoped, in my present contribution, that some general considerations about what
representing-as is might still that urge which calls for prophylactic.
Finally, what is the importance of the question whether perception has (rep-
resentational) content? Not least in that its answer bears on a project Frege laid
out clearly: understanding those two utterly different main forms of awareness we
enjoy of how things are. Perceptual awareness (that awareness which perceiving
is per se) is one. Awareness in thought—e.g., appreciating, realizing, recognizing,
seeing, noticing that such-and-such is so—is another. What one could be aware of
in the one of these is, as Frege stressed, what one could not be aware of in the other.
One can see the meat fall before one onto the white rug. One can watch its fall.
The meat is on the carpet. By contrast, that the meat fell is neither on, nor under,
it. That the meat fell is unlocatable. Correlatively, it can form no images on retinas.
So, though looking at the meat on the rug, one may say, ‘I see that the meat fell
out of the grocery bag onto the rug’, here ‘see’ is a verb, not of perception, but of
cognition. (Thus, that one need not be, and presumably was not (observing the
rug), visually aware of all the events thus mentioned.) It is nonsense to speak of
being visually aware of ‘that the meat fell’. By contrast, there is something the meat
looked like as it fell onto the rug; something its fall looked like. But one cannot
think such bits of history as the meat’s fall. There is no ‘realizing that the meat’s
fall’. For history to have taken the turns it thus did and for its being as it then was
to be a case of a certain general way there is for things to be—such that meat had
fallen—are two quite different things. Undergoing the sight of falling meat is not
bringing anything under a generality.
 Against Strong Content
176   

Such different forms of awareness serve different purposes. Perceptual aware-

ness is, per se, awareness of history unfolding. It is awareness of that which history
provides to be, or not, as we represent things as being in, e.g., representing the
meat to have fallen. The fall of the meat is something we can discuss (dine out on
for weeks). In such discussion we rely on grasp, by ourselves and our audiences,
of, e.g., what way for things to be it is for red meat to have fallen onto a white rug.
Such grasp is, inter alia, a capacity to recognize, of what history provides, or might
have, what would, what would not, be a case of things being that particular way
for them to be. Perception offers us one sort of opportunity for exercise of such
capacities. There would be no making sense of our having such capacities were no
such opportunity in the cards. Such is the first purpose for perceptual awareness
to serve. What is distinctive of awareness in thought is that it allows for action—or
policy—guiding in ways that would be impossible without it.
Each form of awareness is what the other could not be. Some, though, think
perceptual awareness inevitably shackled to awareness (of some sort) of some-
thing else. We see the meat on the rug. We enjoy visual awareness of it. In doing so
(the idea is) we also undergo, or suffer, awareness of it being represented as so—
perhaps to be so—that such-and-such. First remark. Insofar as our visual aware-
ness is of the world around us—that is, insofar as we are perceiving—what we are
thus aware of cannot be a vehicle of such representing, as an uttered sentence may
be the vehicle of someone’s representing where he says something. The sentence
makes the representing recognizable. But the meat, the rug, etc., are not in the
business of making any representing recognizable. They do not bear the content of
the representing supposedly done.
Second remark. If there were such representing, we would not be aware of it
as we are of representing effected in a content-bearer which makes it recognizable.
For there is no such content-bearer. So in some sense, such awareness may not be
perceptual. Which may make experiencing this representing uncanny. Awareness
of it would remain, like perceptual awareness, awareness of an occurrence, just
another bit of history. It would be awareness enjoyed in witnessing. It would not
be awareness of some object of thought: recognition of things as some way there
is for things to be; of some thought. If objects of perceptual awareness could not,
just in such awareness of them, make the world bear rationally for the perceiver
on the thing to think, then such occurrences of representing, were we aware of
them, would not help. They would just be more of the same. It would be their con-
tents, and not they themselves, which were objects of thought. And it would be the
generalities under which such occurrences fell in being as they were, and not the
occurrences themselves, which stood in rational relations (if the underlying doc-
trine here about rational relations is right). It would be these, if anything, which
made the world bear rationally for the perceiver on what to think. But if things
being as we see them to be, the red meat on the white rug, cannot itself bear ration-
ally on what generalities it falls under, how can the generalities these occurrences
of representing fall under get into the picture here at all? Rather than means by
 The Preserve of Thinkers   177

which the objects of our perceptual awareness might come to bear for us on what
to think, such representings simply call for more of those same means (whatever
these might be) before they can have rational import for the perceiver on what to
think. For understanding the relation between our two forms of awareness, they
are merely Ptolemaic remedy.
Third remark. If, gazing at the rug, we are aware of things being such that
tonight’s T-bone is on it, and if, moreover, things are so represented in the repre-
senting then inflicted on us, this last part can be only incidental to the case. For
either this representing merely endorses what we can already recognize in seeing
what we thus do, the T-bone in plain sight, we knowing our meat, or we cannot
tell by looking whether it is T-bone on the rug, in which case it is not percep-
tion’s prerogative so to inform us, whether in some occurrence of representing or
otherwise.
Such awareness is not inflicted on us as our awareness of the presence of this
representing-as is. It is the awareness enjoyed in our taking it, knowledgeably, that
such-and-such is so (in our holding such a stance). (This should not be taken as
the discovery of a new, third, form of awareness.) Whether we are then aware
of things being any of the ways they were thus represented as being (assuming
they are any of these ways) depends on what we are aware of—notably in percep-
tual awareness—other than the presence, or occurrence, of that representing—of
what such other awareness positions us to judge knowledgeably. (If in that rep-
resenting the meat is represented as a T-bone, it is seeing the meat which allows
us to see whether this is so.) The excrescence on perceptual awareness that such
representing-as would be could only be at best annoying.

References

Chomsky, Noam (1959). Review of B. F. Skinner’s Verbal Behavior. Language, 35(1), 26–58.
Clarke, Thompson (1972). The legacy of skepticism. Journal of Philosophy, 69(20) (November
9), 754–769.
Descartes, René [1637] (2000). Discours de la méthode. Paris: Flammarion.
Diamond, Cora (1991). Frege against Fuzz. In id., The Realistic Spirit (pp. 145–178).
Cambridge, MA: MIT Press.
Frege, Gottlob (1882a). 17 Kernsätze zur Logik. In Nachgelassene Schriften (pp. 189–190).
Hamburg: Felix Meiner Verlag, 1983.
Frege, Gottlob (1882b). Letter to Anton Marty. In Gottlob Frege’s Briefwechsel (pp. 117–119).
Hamburg: Felix Meiner Verlag, 1980.
Frege, Gottlob (1892). Ueber Sinn und Bedeutung. Zeitschrift für Philosophie und philoso-
phische Kritik, NF 100, 25–50.
Frege, Gottlob (1892–95). Ausfuhrungen über Sinn und Bedeutung. Nachgelassene
Schriften, 128–136.
Frege, Gottlob (1918). Der Gedanke. Beiträge zur Philosophie des deutschen Idealismus, 1
(2), 58–77.
 Against Strong Content
178   

McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
Kaplan, David (1989). Demonstratives. In J. Almog et al. (Eds.), Themes from Kaplan (pp.
481–563). Oxford: Oxford University Press.
Putnam, Hilary (1988). Representation and Reality. Cambridge, MA: Harvard University
Press.
Waismann, Friedrich (1979). Ludwig Wittgenstein and the Vienna Circle. Oxford: Basil
Blackwell.
Wittgenstein, L. (1922). Tractatus Logico-Philosophicus. London: Routledge and Kegan Paul.
7

Disjunctivism, Discrimination, and Categorization

Diana Raffman

As most readers of this volume will be aware, the dispute between disjunctivism
and the common kind theory in the philosophy of mind centers on the metaphysics
of certain types of conscious events: veridical perceptions, illusions, and hallucina-
tions. Traditionally, these three types of events—sensory experiences—have been
taken to be of the same kind, in the strong sense of sharing the same fundamental
nature. What binds them together is said to be (for example) their intrinsic qual-
itative character, or their intentional content, or a certain adverbial analysis. In
contrast, the disjunctivist I will be discussing holds that veridical perceptions are
essentially different in kind from illusions and hallucinations.1 He thinks that the
mind-independent entities that are the intentional objects of our experiences are
constituents of the experiences that are veridical; and for obvious reasons, no such
constituency can obtain in the case of illusions or hallucinations.2 (Disjunctivism
is sometimes called a “relational” theory of perception, because of this constitutive
relation between perception and its objects.)
The disjunctivist acknowledges that veridical perceptions, illusions, and hal-
lucinations are all experiences, but he claims that what unites them as such is just
their indiscriminability from veridical perceptions—not their qualitative or inten-
tional or adverbial properties. Specifically, a mental event is an experience as of Φ
just in case it is indiscriminable from a veridical perception of Φ. The notion of
indiscriminability at work here, following Williamson (e.g., 1990), is meant to be
epistemological: experiences are indiscriminable if they cannot be told apart, i.e.,
cannot be known to be distinct, through introspection. M. G. F. Martin explains:
We need not look for some further characteristics in virtue of which an event
counts as [e.g.] an experience of a street scene, but rather take something
to be such an experience simply in virtue of its being indiscriminable from

1
Here I ignore many significant differences among disjunctivist positions; see, e.g., Haddock and
Macpherson (2008) and Soteriou (2009) for detailed surveys.
2
Actually disjunctivists disagree about the proper treatment of illusions; I skate over the issue
here, but see again Soteriou (2009). 179
 Against Strong Content
180   

a perception of a street scene. Nothing more is needed for something to be

an experience, according to this conception, than that it satisfy this episte-
mological condition. Rather than appealing to a substantive condition which
an event must meet to be an experience, and in addition ascribing to us cog-
nitive powers to recognise the presence of this substantive condition, it in-
stead emphasizes the limits of our powers of discrimination and the limits
of self-awareness: some event is an experience of a street scene just in case it
couldn’t be told apart through introspection from a veridical perception of
the street as the street (2004, p. 48). . . . There are certain mental events, at least
those hallucinations brought about through causal conditions matching those
of veridical perceptions, whose only positive mental characteristics are nega-
tive epistemological ones—that they cannot be told apart by the subject from
veridical perception. (2004, p. 74)
On the other side, the common kind theorist grants that perceptual experiences
are by their nature indiscriminable from veridical perceptions, but she insists that
they stand in the latter relation in virtue of possessing “substantive” or “robust”
qualitative or intentional properties. Thus disjunctivist and common kind theorist
agree that perceptual experiences (including veridical perceptions) are indiscrim-
inable from veridical perceptions, and to that extent are of the same kind, but
disagree as to whether any further attributes are essential to their nature. Susanna
Siegel emphasizes this point:
Martin takes it that the common-kind theorist will agree that for an event
to so much as count as a perceptual experience, it has to be indiscriminable
from a veridical perception. So they will agree, Martin thinks, that it’s a con-
ceptual truth that sensory experiences are indiscriminable from veridical per-
ceptions. The disagreement is supposed to concern whether anything else is
conceptually true of sensory experience. As Martin construes his opponent,
she says that something else is: it is part of the concept of perceptual experi-
ences that they instantiate mental properties that realize, or underlie, indis-
criminability from veridical perception. As to the metaphysical nature of the
common kind property, there are the options made familiar by the history of
the philosophy of perception so far: candidates include sense-data, being an
adverbial modification, having propositional content of some sort, and com-
binations thereof. (2004, pp. 92‒93)
The relation of perceptual or phenomenal indiscriminability, which plays a
central role in the disjunctivist’s theory, is widely considered to be nontransitive.3
The non-transitivity has seemed to pose problems for philosophical views on a
variety of topics. In the following passage, Martin voices concern on the disjunc-
tivist’s behalf:

3
Some exceptions are Hardin (e.g., 1988), Burns (1995), Raffman (2000), and Graff (2001).
 Disjunctivism, Discrimination, and Categorization   181

[A]‌challenge to the sufficiency of indiscriminability for identity of kind of

experience comes from the alleged nontransitivity of indiscriminability for
some observable properties. Certainly, given observers on particular occa-
sions may fail to detect the difference in shade between sample A and sample
B, and also fail to detect the difference between sample B and sample C, and
yet be able to detect the difference between sample A and sample C. If this
leads us to the conclusion that experiences of A are indiscriminable from
experiences of B, and experiences of B are indiscriminable from experiences
of C, then we face a problem supposing that there are kinds of event which
are sensory experiences of colour shades on the disjunctivist proposal. The
indiscriminability of experience of A and experience of B would require us
to suppose that these are just the same kind of experience; likewise for the
experience of B and of C. By transitivity of identity, this requires that the kind
of experience one has of A is of the same kind as the experience one has of C,
but this contradicts the observation that the experience of C is discriminable
from the experience of A since kinds of experience are discriminable only
where distinct. (2004, p. 76)
Martin appears to reason that if two stimuli are indiscriminable, then the expe-
riences they occasion are indiscriminable; and indiscriminable experiences are
identical. Hence if indiscriminability is nontransitive, something has to give.4
In what follows I want first to get clear just what indiscriminability consists
in, as Martin (and, following him, Siegel) construe it. Then, drawing upon some
experimental results, I’ll argue that when indiscriminability is thus construed, the
usual argument for nontransitivity falls apart. Even if the worries about nontransi-
tivity can be set aside, however, the relation of indiscriminability probably cannot
do the philosophical work that Martin’s disjunctivist assigns to it. At the end I will
suggest an alternative conception of the commonality among veridical percep-
tions, hallucinations, and illusions—a conception that may satisfy the disjunctiv-
ist’s needs without appealing to indiscriminability.

7.1. Indiscriminability: Some Clarifications

Unclarity about the relation of perceptual indiscriminability has been a source of

confusion in the philosophical literature. Philosophers working in the area have
achieved a better understanding over the last few years, but a brief review of the
issues will be helpful.
For starters, indiscriminability is often run together with the phenomenal
relation of appearing (looking, sounding, tasting, etc.) the same. The confusion

4
As Siegel (2004, p. 206, e.g.) points out, the alleged nontransitivity appears to pose comparable
problems for the common kind theory.
 Against Strong Content
182   

is understandable if one supposes that indiscriminable stimuli invariably appear

the same; but in fact there are no such stimuli: not even physically identical stim-
uli viewed under identical conditions by the same observer always appear the
same. (Psychophysicists use the term “false alarm” for instances in which phys-
ically identical stimuli are judged different.5) Rather, indiscriminability is best
understood as a statistical relation: roughly, stimuli are indiscriminable just in
case they appear the same (are judged the same) in 50 percent of pairwise same/
different comparisons, or in 75 percent, or in 60 percent, etc., depending upon
the experimenter’s explanatory goals (see any psychophysics textbook). Properly
construed as a statistical psychophysical relation, indiscriminability is plainly
nontransitive: there can be an observer O and series of stimuli s1. . .sn such that,
under some constant viewing conditions (suppose even that s1. . .sn are all in view
simultaneously), s1 appears the same to O as s2 in (e.g.) 75 percent of same/dif-
ferent trials, s2 appears the same as s3 in 75 percent . . . and sn–1 appears the same
as sn in 75 percent, but s1 and sn appear the same in only 30 percent of trials. For
convenience, call this kind of series an indiscriminability series.
Note that indiscriminability is not a phenomenal relation; you can’t tell
whether two stimuli are indiscriminable just by looking or listening or tasting.
On the other hand, appearing the same is a phenomenal relation—one that
holds between stimuli only occurrently, for a given observer at a given time. You
can tell just by looking, at a given time, whether two stimuli appear the same to
you at that time. In what follows I will use the expression “appear the same” to
refer to the relation of looking (sounding, tasting, etc.) the same to an observer
in a given act of comparison based upon inspection at a given time, where an
act of comparison is just a same/different judgment of stimuli presented either
simultaneously or in immediate succession. The statistical relation of indis-
criminability is defined in terms of the phenomenal relation of appearing the
same: indiscriminable stimuli appear the same in (e.g.) 75 percent of pairwise
comparisons. Statistical indiscriminability is a standing relationship, whereas
appearing the same is an occurrent one.
It is important here to distinguish same/different judgments from recogni-
tional judgments or type-identifications. For example, a subject may be able to
tell just by looking that a sample of red28 and a sample of red29 are different, i.e.,
are of different types (shades) of red, without being able to identify (recognize,
categorize) either of them as red28 or as red29.6 To put the point another way, she
may be able to discriminate the two samples, but not to type-identify them. (That

5
For present purposes I will suppose that a stimulus appears Φ to a subject at a time just in case
he judges it to be Φ upon inspection at that time; in particular, two stimuli appear the same in hue just
in case the subject judges them the same in hue upon inspection. Of course, in some contexts such a
supposition would be problematic, even question-begging, but it shouldn’t cause trouble here.
6
Discrimination and categorization are probably functions of distinct psychological mechanisms;
see, e.g., Pernet et al. (2004); Mirman et al. (2004).
 Disjunctivism, Discrimination, and Categorization   183

perceivers can discriminate more perceptual values than they can identify on most
stimulus dimensions is a basic tenet of psychophysics. Perceptual memory has its
limits.) Similarly, a subject may be able to tell by introspection that her experiences
of those samples of red are of different types (shades), without being able to iden-
tify either of them as an experience of red28 or as an experience of red29.
As far as I can tell, when philosophers of perception talk about the alleged
nontransitivity of indiscriminability, they almost never intend the statistical rela-
tion. Martin and Siegel certainly don’t; they have in mind a phenomenal relation.
In calling two stimuli indiscriminable they seem to mean either that the stimuli
invariably appear the same when compared or that they appear the same in some
same/different comparison at some time. Since there are no stimuli that invariably
appear the same, we need to read Martin and Siegel in the second way, as referring
to an occurrent relation of sameness of appearance.7 We will suppose it’s the latter
relation they believe to be nontransitive.
Perhaps the most compelling reason to think that appearing the same is non-
transitive is the existence, or at least the possibility, of phenomenal continua, viz.,
apparently continuous progressions in hue (loudness, pitch, etc.). In light of the
distinction between indiscriminability and appearing the same, how should we
characterize phenomenal continua? They are not plausibly identified with indis-
criminability series, in which neighboring members do, but the endpoints do
not, appear the same in a certain percentage of comparisons. A phenomenal con-
tinuum is supposed to be phenomenal, hence apparent upon inspection. As a first
approximation, a phenomenal continuum is a series of simultaneously presented
stimuli in which neighboring items appear the same but the endpoints appear dif-
ferent, at a given time, to a perceiver who proceeds along the series giving each
pair of stimuli a good straight look (listen, taste, etc.).8 More precisely, a phenom-
enal continuum is a continuous progression in appearance that is instantiated, for
an observer at a time, by such a series. The phenomenology of phenomenal conti-
nua is baffling because, given perfectly constant viewing conditions, the first and
last items appear different and yet nowhere between the two is any local difference
in appearance discerned.
Let me emphasize that stimuli that can instantiate a phenomenal continuum
need not be pairwise indiscriminable. Since discriminable stimuli sometimes ap-
pear the same (e.g., in 25 percent of same/different trials), there is nothing to pre-
vent a series of discriminably different stimuli from instantiating a phenomenal
continuum for a given observer on a given occasion. (Also there is nothing to
prevent an indiscriminability series from failing to instantiate one.) Let me say
that again: there is nothing to prevent a series of discriminably different stimuli from

7
Martin clearly has the occurrent relation in mind in the second and third sentences of the
passage from 2004, 76 cited above; but I cannot find a consistent usage in the texts.
8
I borrow the expression “good straight look” from C. L. Hardin (1988).
 Against Strong Content
184   

instantiating a phenomenal continuum. Any indiscriminability series can instan-

tiate a phenomenal continuum, but so can some series of discriminable stimuli. An
indiscriminability series and a phenomenal continuum are two different things.
Martin responds to the alleged nontransitivity problem this way:
[D]‌espite the appeal of apparent examples of indiscriminable but distinct
shades, one can seek to resist the argument and hold on to the idea of percep-
tual experience as forming kinds. One might follow Graff ’s suggestion that
there is simply no good reason to believe in the existence of phenomenal con-
tinua and hence insist that if two samples really do look alike then they share
a look. Even if a subject may on occasion fail to notice the difference in look
between adjacent samples, and indeed may be bound to fail to notice such
a difference, nonetheless there is a difference to be noticed and which could
be noticed. Alternatively, one may follow Williamson’s suggestion that while
in a given context a subject may fail to discriminate two samples, this does
not show that there is no context in which the samples are discriminable and
hence one can hold on to the claim that distinct samples are discriminable in
at least some context. By suitable application of the idea of impersonal indis-
criminability, we can then insist that the experiences of A and of B are in fact
discriminable, even if in the given context a subject fails to discriminate A
from B and consequently fails to discriminate the experience of A from the
experience of B. (2004, 77–78)
I am going to urge that these ad hoc, counterintuitive responses by Graff and
Williamson are unnecessary. First I need to present some experimental results that
cast doubt on the notion that appearing the same, specifically looking the same
in hue, is nontransitive. If I am right, the disjunctivist has nothing to fear from
nontransitivity.

7.2. Is Appearing the Same Nontransitive?

I wondered whether the mysterious phenomenology of phenomenal continua could

be explained by unnoticeable, or at least unnoticed, changes in the appearances of
the stimuli in the instantiating series—not unnoticeable differences of appearance
between distinct stimuli, but unnoticeable changes in the appearances of individual
stimuli. If such changes can occur, a claim of nontransitivity is undercut. For a
claim of nontransitivity to hold good, the stimuli in a phenomenal continuum must
remain constant in appearance throughout. A series in which the stimuli change
their appearance doesn’t show that appearing the same is nontransitive, any more
than the fact that Tom and Dick weigh the same, and Dick and Harry weigh the
same, but Tom and Harry have different weights, shows that identity is nontransi-
tive if we’ve weighed Tom and Dick in 2001, then Dick and Harry in 2002, then Tom
and Harry in 2003. To find out whether unnoticed changes in appearance occur in a
 Disjunctivism, Discrimination, and Categorization   185

x
x

FIGURE 7.1 Stimulus configuration for the first of two tasks in the experiment. Adjacent stimuli to
be judged (same/different) on a given trial are indicated by black dots. (For illustration in black
and white, I have marked the endpoints of the hue series with ‘x’.)

phenomenal continuum, I designed and ran an experiment with two psychologists

working on color vision, Delwin Lindsey and Angela Brown.9
The stimuli were a series of 41 patches of colored light that instantiated a phe-
nomenal continuum (on almost all trials) between two slightly but clearly different
shades of green. The stimuli were presented on a high-resolution color monitor in
the circular arrangement shown in ­figure 7.1. Nothing depended upon the loca-
tions of the endpoints. About half of the stimuli were redundant: roughly every
other patch in the circle was physically identical to its predecessor.10 Neighboring
physically different patches differed by less than the discrimination threshold or
just noticeable difference in hue of our most sensitive subject. (We had established
the thresholds of our subjects in an earlier experiment, requiring correct detection
on 75 percent of trials.) The subjects in the experiment were ten philosophy and
psychology faculty, students, and staff at Ohio State University, including several
faculty and graduate students in psychology of vision.
Each trial began with a same/different comparison of the hues of two neigh-
boring patches, indicated to the subject by two black dots as shown in ­figure 7.1.
If the subject made a judgment of “different” (which happened on only one trial
by one subject), the next trial began immediately and she was cued to judge the
next pair of patches. (If the patches are numbered #1‒#41, the order of the pairs
was #1/#2, #2/#3, #3/#4, etc. Consecutive pairs always shared a patch.) If the sub-
ject made a judgment of “same,” a disk of colored light appeared in the center of

9
Delwin Lindsey, Department of Psychology, Ohio State University; Angela Brown, School
of Optometry, Ohio State University. I present these results also in Raffman 2011, in a much more
detailed discussion of (in)discriminability.
10
There were 21 physically distinct stimulus values. If we label the 21 values as a‒u, their order in
the circle can be specified as a, a, b, b, c, c, and so on. Consecutive trials then involved the pairs a/a, a/b,
b/b, b/c, c/c, etc. The “redundant” pairs (a/a, b/b, etc.) tested for false alarms, viz., “different” responses
to identical stimuli. The latter data are irrelevant to the present discussion.
 Against Strong Content
186   

FIGURE 7.2 Stimulus configuration for the second task in the experiment, performed on trials in
which adjacent patches had been judged same in the first task. (This included all but one trial by
one subject.) Subjects adjusted the hue of the central disk to match the hue of the two patches
indicated by black dots.

the circle, as in ­figure 7.2. The subject then adjusted the hue of the disk by moving
the computer mouse back and forth until the hue of the disk matched the hue of
the two patches. The disk then disappeared and the next trial began. (Both the
initial hue of the disk and the relationship between the hue of the disk and the
mouse position were randomized from trial to trial.) Subjects went around the
circle twice. At the end of the experiment we asked roughly half of the subjects if
they had noticed any changes in the colors of the patches during the experiment.
All said “no.”
What we found was that even though all of the patches were in view
throughout, and the members of every pair were judged “same” by every subject
on almost every trial, subjects’ settings of the disk progressed more or less system-
atically with the physical reflectances of the patches (­figure 7.3). In other words,
subjects matched the pair #2/#3 to a longer wavelength than the pair #1/#2, the
pair #3/#4 to a longer wavelength than the pair #2/#3, and so on.11 This means that
patch #2 was matched to a different wavelength when it was compared to #1 than
when it was compared to #3; patch #3 was matched to a different wavelength when
compared to #2 than when compared to #4; and so forth—again, even though all
of the patches were in view. (The graphs contain more than 41 data points because
subjects went around the circle twice; hence, pairs that were judged “same” both
times received two disk settings.) The graphs show fairly steady progression of the
disk settings as subjects progressed through the pairs of patches, for both the phys-
ically identical and physically different pairs.12 Figure 7.4 shows the disk settings
averaged across all subjects.

11
For convenience I use the term “wavelength,” but strictly speaking it is incorrect. The stimuli were
mixtures of broadband lights, and neither the primaries nor the mixtures had a defined wavelength.
12
This result suggests that subjects may have been matching the hue of the disk to the mean
physical value of the two patches in each pair.
 Disjunctivism, Discrimination, and Categorization   187

(a) 1000
950 Physically same
900 Physically different
Disk setting in arbitrary units

850
800
750
700
650
600
550
500
450
400
0 5 10 15 20 25 30 35 40
Stimulus pair

(b)
1000
950
Same
900 Different
Disk setting in arbitrary units

850
800
750
700
650
600
550
500
450
400
0 5 10 15 20 25 30 35 40

Stimulus pair
FIGURE 7.3 Disk settings (in arbitrary units) for two subjects, on stimulus pairs judged “same”.

No doubt these data permit multiple interpretations. But they provide at least
some support for the idea that individual patches changed their hue appearances in
their different pairwise comparisons; the patches looked different from one com-
parison to another, but so subtly that subjects could not notice it. Plausibly, each
pairwise comparison required a distinct attentional act, and the stimuli changed
the way they looked with each shift of attentional focus. The change is so slight—
perhaps something like an extremely fine-grained Gestalt shift in hue—that sub-
jects could not notice it, and so could not report it.13 Although the phenomenology

13
Raffman (2000) contains a more detailed discussion of the possibility of Gestalt hue shifts in a
sorites series.
 Against Strong Content
188   

1000

900

800

700

600

500

0 5 10 15 20 25 30 35 40
FIGURE 7.4 Disk settings averaged across all subjects.

is very difficult to describe, it seems unlikely that subjects were undergoing merely
a subliminal or unconscious sort of perception, since making a same/different
comparison and setting the hue of the disk required sustained conscious attention.
Borrowing some nice terminology from Fred Dretske (2004), perhaps we can say
that subjects saw different hues, consciously saw them, but could not see that they
were different; or that subjects saw different hues, but could not see the difference
in hue. Or perhaps they saw different hues but could not see the changes in hue.14
However the phenomenology should be described, our results appear to
support the hypothesis that unnoticeable changes occur in the appearances of
individual stimuli in a phenomenal continuum. These subtle changes, not any
nontransitivity, make phenomenal continua possible. If that’s right, then neigh-
boring stimuli in a phenomenal continuum do indeed appear the same in pairwise
comparisons, but the instability of their individual appearances across different
pairwise comparisons defeats the nontransitivity claim. Although any indiscrim-
inability series can instantiate a phenomenal continuum, such a series is only
physically, not phenomenally, stable. A claim of nontransitivity is true of the statis-
tical relation of indiscriminability because the relevant physical properties of the
stimuli (wavelengths, frequencies, etc.) in an indiscriminability series are stable.

14
Dretske introduces this terminology in order to characterize the experience of change
blindness. The present results may be reminiscent of change blindness, but subjects experiencing the
latter effect are able to notice the change when their attention is explicitly directed to it, whereas that
is unlikely to be the case in our study. Also, our experimental condition involved no visual disruption.
In general, subjects who experience change blindness fail to notice large changes in visual scenes when
the changes occur during a visual disruption such as a saccade or blink or a cut in a film (though see
Simons et al. 2000). For example, viewers in one experiment failed to notice that two people in a scene
had exchanged heads (Grimes 1996).
 Disjunctivism, Discrimination, and Categorization   189

No instability analogous to that of the apparent hues in a phenomenal continuum

occurs in an indiscriminability series as such.
Now consider again the passage from Martin cited above (2004, 76). He writes
as if the expressions “experiences of A” and “experiences of B” have single, stable
referents, as if there is some one kind of experience—“the kind of experience one
has of A”—that could stand in stable relations of sameness and difference with
other kinds of experiences. In effect, he thinks of experiences as having a stability
commensurate with the stability of the stimuli of which they are experiences. No
doubt he would acknowledge the influence of contextual factors and refer more
explicitly to “the experience of A in context c” and “the kind of experience one has
of A in c.” Still, though, Martin would understand the latter descriptors as having
fixed, univocal referents. This conception is an error that goes hand in hand with
the error of assuming that A and B have single stable shades of color. Our exper-
imental results suggest that a locution like “the experience of B in context c” can
be applied only relative to a particular act of attention by a particular observer at
a particular time; all we can talk about is the shade B appears to have, or the shade
experience B engenders, when it is compared to A by a given observer in a given
context at a given time. The experiment suggests that the experience of B when B is
compared to A is different from the experience of B when B is compared to C.15 By
the same token, the shade B appears to have in the B/A comparison differs from
the shade B appears to have in the B/C comparison. Intuitively, it’s as if objects
change their shades like hats, wearing now one, and now another; for example, B
unnoticeably slips on a slightly different hat as the observer’s attention moves from
B/A to B/C. This changing of hats is what dissolves the mystery of the phenomenal
continuum, enabling the first and last items to appear different although nowhere
between the two is any local difference in appearance discerned.
If a nontransitivity claim is unjustified, the counterintuitive and ad hoc
responses by Graff and Williamson, invoked by Martin (above), are unnecessary.
Most importantly, we needn’t follow Graff in denying the existence of phenom-
enal continua. In addition, we can perfectly well say that if two samples “really do
look alike, then they share a look.” Samples do share a look when they are looking
alike, i.e., when they are being judged the same in a same/different comparison.
Williamson’s ad hoc suggestion is a nonstarter, viz., that phenomenal continua
do not (cannot?) exist because physically different samples are always “imperson-
ally” discriminable. On the contrary, all that is required for the existence of a phe-
nomenal continuum is that some series of stimuli instantiate the requisite sort of
progression for some observer at some time. It’s irrelevant whether someone else
could tell neighboring stimuli apart. At bottom, Graff, Williamson, and Martin
make the same mistake: they assume that a progression from one color to another
can occur only if a (perhaps not immediately detectable) color difference obtains

15
Of course, the experiment does not conclusively show this. Rather, it shows that we cannot
assume that subjects have a single, stable experience of B.
 Against Strong Content
190   

between some neighboring stimuli. They fail to consider the possibility suggested by
the experiment, viz., that such a progression can occur if individual stimuli subtly
shift their appearances, change hats, from one pairwise comparison to the next.
There need be no “discrimination,” explicit or implicit, between adjacent stimuli.16

7.3. On Comparing Token Experiences

Even if the worries about nontransitivity can be set aside, the relation of appearing
the same probably cannot do the philosophical work that Martin’s disjunctivist
requires of it. Let me try to say why.
Martin claims that “some [token] event is an experience of a street scene just
in case it couldn’t be told apart through introspection from a veridical percep-
tion of the street as the street.” What exactly is this supposed to mean? The locu-
tion “couldn’t be told apart through introspection from a veridical perception of
the street as the street” must be short for something like “couldn’t be told apart
through introspection from a veridical perception of the street as the street, were
the event in question to occur simultaneously with such a perception in a same/
different comparison.” Consider then an event e and some veridical perception v
of the street as the street. The trouble with Martin’s view is that, if the two events
do not actually occur simultaneously (or in immediate succession), the imagined
comparison would require a temporal relocation of at least one of the two. On the
plausible assumption that the time of occurrence of an experience is partly con-
stitutive of its being that very experience, the imagined comparison could not be
a comparison of e and v.17 Perhaps Martin imagines that two token experiences
couldn’t be told apart just in case, if two other token experiences of the same types
as e and v, call them e* and v*, were to occur simultaneously or in immediate
succession in a pairwise comparison, the subject would judge e* and v* to be the
same. But this strategy employs the very notion—that of experiences being of the
same type—that he is supposed to be explicating. Furthermore, the co-occurrence
of e* and v* would, or at least could, constitute a context relevantly different from
the contexts in which e and v occur, thereby threatening the type-identity of e* and
v* to e and v.
The upshot, I think, is that we cannot meaningfully talk about whether two
experiences could be told apart were they to occur simultaneously or in imme-
diate succession. Martin’s and Siegel’s talk of the (in)discriminability of experi-
ences is, in the loose and popular sense, a category mistake. Only physical stimuli,
whose identity conditions do not essentially involve states (e.g., attentional foci) of

16
Sturgeon (2006) and Conduct (2010), among others, offer alternative responses to the alleged
nontransitivity problem. Insofar as the approach I take is grounded in the perceptual psychology of the
situation, it may be better motivated than these others.
17
Siegel is aware of this problem (2004, p. 109).
 Disjunctivism, Discrimination, and Categorization   191

judging subjects, can sustain such subjunctive or counterfactual relationships. As

far as I can see, the most one can say about type-identity of experiences in terms
of same/different comparisons is that experiences are type-identical (-different),
i.e., they seem the same, when and only when the subject of those experiences is
undergoing them, simultaneously or in immediate succession, and judging them
the same (different). Only in this narrow range of circumstances is the relation of
seeming the same instantiated by our experiences. Outside this range, I think we
have to say that there is no fact of the matter as to whether experiences are type-
identical or -different—that is, no fact of the matter that can be expressed in terms
of same/different comparisons (discriminations).
Presumably the disjunctivist needs a more robust notion of sameness of expe-
rience. He needs a standing, non-occurrent relation that can hold between experi-
ences that don’t actually occur simultaneously and aren’t actually compared. Siegel
writes:
For all my complaints about the link Martin sees between indiscriminabil-
ity and the phenomenal, rejecting it leaves a serious question unanswered.
Even to state the debate between disjunctivism and its opponents, one needs
a way to characterize the relevant pairs of perceptions and hallucinations.
Only some such pairs raise the question whether they share a fundamental
property. Which pairs are these? The claim that they are the pairs that are
indiscriminable from the same veridical perception provides a simple answer.
If this answer is rejected, it’s not clear what to replace it with. Replacing it with
some other notion of indiscriminability, or with some notion of phenomenal
sameness, brings in weighty theoretical commitments at the outset—just as
Martin’s cognitive notion of indiscriminability does. The moral seems to be
that in this debate, it is difficult to escape making theoretical commitments
from the very start about the kind of access to experience that introspection
provides. (2004, p. 108)
If what I have just been saying is right, then in order to formulate sufficiently gen-
eral identity conditions for kinds of experience, resources other than, or at least
additional to, same/different judgments or discriminations will be needed. In the
final section of this paper, I will ask whether indiscriminability could be replaced
with a different relation in the disjunctivist’s account—a relation that may allow
a more workable answer to the question of what veridical perceptions, hallucina-
tions, and illusions have in common.

7.4. Discrimination vs. Categorization: A Speculative Proposal

My thought is that in explaining what makes a mental event an experience, Martin

and his interlocutors may have conflated two importantly different kinds of
judgments: discriminations or same/different judgments, on the one hand, and
 Against Strong Content
192   

categorizations or type-identifications, on the other.18 Martin says that an event

is an experience just in case it is “indiscriminable” by introspection from a verid-
ical perception. His concerns about nontransitivity, among other things, indicate
that he is thinking of (in)discriminability in terms of same/different judgments.
But why couldn’t he define an experience in terms of a recognitional judgment or
categorization, rather than a discrimination? Why couldn’t he say that an event
is an experience just in case the subject cannot tell—and here we mean “cannot
recognize”—whether the event is veridical or illusory? We might also say: the sub-
ject would categorize the event, with respect to its veridicality, in the same way he
would categorize a veridical perception.
Granted, if you can recognize (categorize, type-identify) one experience as
veridical and another as illusory, then trivially you can “discriminate” between
the two experiences with respect to veridicality; conversely, if you cannot rec-
ognize (tell) that the one is veridical and the other illusory, then in that sense
you cannot discriminate between them. They seem introspectively the same, i.e.,
category-identical: either both seem veridical or both illusory. But the latter spe-
cies of “(in)discriminability”—a misnomer, really—is not the relation defined in
terms of same/different judgments and reputed to be nontransitive.
For present purposes, the crucial difference between discrimination and
categorization is that a categorization, unlike a discrimination or same/different
judgment, would not require simultaneous instantiation of a veridical perception
and the mental event at issue. If there is any sense at all in which a categorization
effects a comparison, it is a comparison to a remembered standard—for example,
the subject’s memory, her knowledge, of what veridical perceptions (as opposed
to illusions) are like, or perhaps of the ways in which veridical perceptions differ
from illusions. M. D. Conduct expresses the disjunctivist stance this way:
[An hallucination] is subjectively indiscriminable from a [veridical perception]
if it is not possible to come to know, through reflection upon one’s experiential
situation, that it is [a hallucination]. (2010, p. 203; emphasis added)
When the disjunctivist’s view is expressed in this way, it is easy to see how a cate-
gorization, viz., the judgment that it is a hallucination, could be confused with a
discrimination or same/different judgment between an occurrent veridical per-
ception and a hallucination.
This much cannot be the whole story, of course, for Martin’s thesis is spe-
cific: an event is an experience as of Φ just in case it is indiscriminable from a
veridical perception of Φ; for instance, an event is an experience as of red28 just in
case it is indiscriminable from a veridical perception of red28. Can we replace the
reference to discrimination with a reference to categorization or recognition, even

18
See above. Lest there be any confusion: the distinction between discriminatory and categorical
judgments is wholly orthogonal to the distinction between statistical and “occurrent” (in)discriminability
relations (i.e., between [in]discriminability proper and appearing the same.)
 Disjunctivism, Discrimination, and Categorization   193

given the reference to a determinate shade? Given the limited nature of perceptual
memory, we cannot recognize or type-identify instances of (e.g.) red28 as such; but
can we nevertheless say that an event is an experience as of red28 just in case intro-
spectively it seems category-identical to a veridical perception of red28?
Consider that we already have in hand non-discriminatory identity condi-
tions for, and can recognize upon inspection, the four so-called unique hues.
These latter are pure instances of the four chromatic categories—red containing
no blue or yellow, green containing no blue or yellow, blue containing no red or
green, and yellow containing no red or green. In addition, a perfectly balanced
orange can be defined as the hue (determinate shade) containing equal amounts
of red and yellow, a perfectly balanced cyan as containing equal amounts of blue
and green, and so forth. Using a technique called magnitude estimation, subjects
can reliably identify the unique hues as the hues containing 0 percent of any other
chromatic category, the balanced binaries as containing 50 percent of each compo-
nent category, and so on (e.g., Ebenhoh and Hemminger 1981; Gordon et al. 1994).
Of course, the extent of our ability to identify or categorize determinate shades
in this way cannot be unlimited: for example, presumably we cannot recognize as
such a shade that is 43 percent blue and 57 percent green as opposed to 44 percent
blue and 56 percent green—much less 43.5 percent blue and 56.5 percent green,
etc. Presumably this is only for lack of the requisite resolving power, though.
Unless we have independent reason to think that our experiences of non-unique,
non-balanced determinate hues are relevantly different from our experiences of
the unique hues and balanced binaries, we can plausibly treat our experiences of
the non-unique hues as being of a kind with our experiences of the unique and
balanced ones, and regard them as similarly type-identifiable at least in principle.19
The idea then would be to say that an event is an experience as of red28 just in
case introspectively it seems category-identical to a veridical perception of a shade
that is (e.g.) 55 percent red and 45 percent yellow.20 Whether such an approach
would satisfy the disjunctivist, I am not certain. Even granting that we can iden-
tify determinate shades in the way just outlined, I am not sure whether identity
conditions expressed in terms of percentages of different chromatic components
can, as Siegel puts it, provide “a way to characterize the . . . pairs of perceptions
and hallucinations . . . [that] raise the question whether they share a fundamental
property” (2004, p. 108). While the proposed strategy might alleviate some of the
problems associated with identity conditions expressed in terms of same/different
judgments, its reliance on the subject’s ability to recognize determinate shades
may veer too close, for the disjunctivist’s liking, to injecting intentional contents

19
It must be remembered that, even if we can individuate determinate shades in the way I am
proposing, these qualities are highly unstable properties of objects; in particular, objects may change
their “hats” from one pairwise comparison of stimuli to another.
20
As before, this category identity would be something the subject can descry in judging a single
event; no hypothetical comparison of simultaneous events would be required.
 Against Strong Content
194   

into the definition of an experience.21 From the disjunctivist’s point of view, replac-
ing discrimination with categorization in the way I have suggested may be simply
replacing one set of problems with another.22

References

Burns, L. (1995). Something to do with vagueness. Southern Journal of Philosophy, 33(S1),

23–47.
Conduct, M. D. (2010). Naïve realism and extreme disjunctivism. Philosophical Explorations,
13(3), 201‒221.
Davidson, L., & Shaw, J. (2010). Perceptual illusions in non-native clusters are
context-dependent. Poster at the meetings of the Linguistics Society of America.
Dretske, F. (2004). Change blindness. Philosophical Studies, 120(1–3), 1–18.
Ebenhoh, H., & Hemminger, H. (1981). Scaling of color sensation by magnitude estima-
tion: A contribution to opponent-colors theory. Biological Cybernetics, 39(3), 227–237.
Gordon, J., Abramov, I., & Chan, Hoover (1994). Describing color appearance: Hue and
saturation scaling. Perception & Psychophysics, 56(1), 27–41.
Graff, D. (2001). Phenomenal continua and the sorites. Mind, 110(440), 905–935.
Grimes, J. (1996). On the failure to detect changes in scenes across saccades. In K. Akins (Ed.),
Perception, Vancouver Studies in Cognitive Science, 2 (pp. 89–110). New York: Oxford
University Press.
Haddock, A., & Macpherson, F. (Eds.) (2008). Introduction to Disjunctivism: Perception,
Action, Knowledge. Oxford: Oxford University Press.
Hardin, C. (1988). Phenomenal colors and sorites. Nous, 22, 213–234.
Horgan, T. (1994). Robust vagueness and the forced march sorites paradox. Philosophical
Perspectives 8: Logic and Language, 8, 159–188.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 120, 37–89.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). New York: Oxford University Press.
Mirman, D., Holt, L. L., & McLelland, J. L. (2004). Categorization and discrimination of
nonspeech sounds: Differences between steady-state and rapidly-changing acoustic
cues. Carnegie Mellon Research Showcase. Department of Psychology. Paper 158.
http://repository.cmu.edu/psychology/158.
Pernet, C., Franceries, X., Basan, S., Cassol, E., Démonet, J. F., & Celsis, P. (2004). Anatomy
and time course of discrimination and categorization processes in vision: An fMRI
study. Neuroimage, 4, 1563–1577.
Raffman, D. (2000). Is phenomenal indiscriminability nontransitive? In C. Hill (Ed.),
Vagueness, Philosophical Topics, vol. 28, 1 (pp. 153–175). Fayetteville: University of
Arkansas Press.

21
Another potential problem is that no analogous account could be given for pitch perception
(among other things). Listeners lacking perfect pitch, i.e., most listeners, have no mnemonic anchors
for pitch that are analogous to our mnemonic anchors for the unique hues. Perfect pitch just is the
ability to hear (remember) pitches the way any normally sighted person sees (remembers) colors.
22
I am grateful to Mohan Matthen and Susanna Siegel for extremely helpful comments.
 Disjunctivism, Discrimination, and Categorization   195

Raffman, D. (2011). Vagueness and cognitive science. In G. Ronzitti (Ed.), Vagueness: A Guide
(pp. 107–122). Dordrecht: Springer.
Siegel, S. (2004). Indiscriminability and the phenomenal. Philosophical Studies, 120(1–3),
91–112.
Simons, D., Franconeri, S., & Reimer, R. (2000). Change blindness in the absence of a visual
disruption. Perception, 29, 1143–1154.
Soteriou, M. (2009). The disjunctive theory of perception. In Stanford Encyclopedia of
Philosophy. Metaphysics Research Lab, CSLI, Stanford University.
Sturgeon, S. (2008). Disjunctivism about visual experience. In A. Haddock and F.
MacPherson (Eds.), Disjunctivism: Perception, Action, Knowledge. Oxford: Oxford
University Press.
Williamson, T. (1990). Identity and Discrimination. Oxford: Basil Blackwell.
PART THREE

Reconciliatory Views
8

The Relational and Representational Character of

Perceptual Experience
Susanna Schellenberg

The history of philosophy is a history of false dichotomies. The dichotomy

between relationalists and representationalists is one such false dichotomy.
Relationalists argue that perception is fundamentally a matter of a perceiver
being related to her environment. Representationalists argue that perception is
fundamentally a matter of a perceiver representing her environment. However,
the standard views in the debate are either austerely relationalist or austerely
representationalist. According to austere representationalists, perception is fun-
damentally representational but not fundamentally relational. According to aus-
tere relationalists, perception is fundamentally relational but not fundamentally
representational.1 Against both, I argue that perceptual relations to the environ-
ment and the content of experience should be recognized to be mutually de-
pendent in any explanation of what brings about perceptual awareness of the
environment.
Another way of expressing the idea is that being acquainted with particu-
lars in one’s environment is neither metaphysically nor explanatorily more basic

I am grateful to Bill Brewer, Todd Ganson, and an anonymous referee for Oxford University
Press for detailed written comments. An early version of this paper was presented at CUNY and the
University of Miami. Thanks also for many helpful discussions at those occasions.
1
For austere representationalist views, see McGinn (1982), Davies (1992), Tye (1995), Lycan
(1996), and Byrne (2001) among many others. For austere relationalist views, see Campbell (2002),
Travis (2004), Johnston (2004, 2006), Brewer (2006), Fish (2009), Genone (forthcoming), and Raleigh
(forthcoming) among others. Martin (2002a, 2004) argues against any view on which experience can be
analyzed in terms of a propositional attitude and a content, leaving open the possibility that experience
could have content without the subject standing in a propositional attitude to that content. Since he
does not outright deny that experience has content, I will discuss his view only to the extent that his
positive view of perceptual experience is structurally similar to that of austere relationalists. Campbell
(2002) calls his view the “relational view,” Martin (2002a, 2004) calls his “naïve realism,” while Brewer
(2006) calls his the “object view.” I will refer to the view with the label “austere relationalism” since
the most distinctive features of the view are arguably the central role of relations between perceiving
subjects and the world as well as its austerity: the view is austere insofar as it denies that experience has
any representational component. There is room in logical space to reject representationalism without
endorsing a relationalist view. For such a view, see Gupta (2012). 199
 Reconciliatory Views
200   

than representing one’s environment. As I argue, there is no such thing as being

brutally acquainted with one’s environment. In being perceptually related to one’s
environment one employs perceptual capacities that yield representational states.
So in contrast to austere relationalists, I argue that perceptual relations to particu-
lars neither ground nor explain perceptual representations. In contrast, to austere
representationalists I argue that perceptual representations neither ground not ex-
plain perceptual relations to particulars in the environment. Perceptual relations
and representations are mutually dependent.
In my (2010 and 2011), I have argued for the view that perceptual experience
is fundamentally both relational and representational. Brewer, one of the main
proponents of austere relationalism, has recently responded to my argument (see
his 2011, pp. 64–65). His response gets to the heart of the issues at stake in the
debate on whether experience has content. Therefore, it will serve as the foil for
the present discussion. In order to adequately respond to his argument, I will
clarify first what is at stake in the debate on whether experience has content by
discussing the idea that experience is a matter of representing (8.1) and the idea
that experience is a matter of being related to one’s environment (8.2). I will then
recapitulate my (2011) argument for the thesis that experience is fundamentally
relational and representational (8.3). Finally, I will respond to Brewer’s reply to
my argument (8.4).

8.1. Perception and Representation

Let the Content Thesis be the thesis that perceptual experience is fundamentally a
matter of representing the environment as being a certain way.
Content Thesis: Perceptual experience is fundamentally a matter of represent-
ing the environment as being a certain way.
As austere relationalists point out, this thesis is rarely argued for. While many
views have been defended that endorse it, more often than not such views simply
assume that experience is representational and proceed to argue for one particu-
lar way of understanding its content. Following Campbell (2002), I use the label
“the representational view” or “representationalism” for any view that endorses the
Content Thesis.2
There are three critical choice points for any representationalist view of per-
ceptual experience. One choice point is how to understand the relationship be-
tween the experiencing subject and the content of her experience. In order to

2
It is important to distinguish this view from the more specific view according to which the
sensory character of experience supervenes on or is identified with its content. Such views are
sometimes labeled “representationalism” rather than the more traditional “intentionalism.” I will
reserve “representationalism” for any view that endorses the Content Thesis. “Representationalism,” so
understood, is neutral on the relationship between content and sensory character.
 The Relational and Representational Character of Perceptual Experience   201

avoid terminological confusions, this choice point is critical in the discussion of

whether experience has content. Therefore, I will address the different options in
some detail.3 The Content Thesis must be distinguished from a thesis on which
the relation between the experiencing subject and the content of her experience
is one of mere association. I call this the Association Thesis.
Association Thesis: Every experience can be associated with (propositional)
content in the sense that sentences can be articulated that describe how the
environment seems to the subject, without the content expressed being a
proper part of the experience.
Any account of experience can accept the Association Thesis. After all, any ac-
count of experience can accept the fact that an experience can be (at least partially)
described. But this fact does not entail that the experience has the content that is
expressed with the description. Certainly, it does not entail that perceptual experi-
ence is fundamentally a matter of representing the environment as being a certain
way. So the Association Thesis does not entail the Content Thesis. To show why, con-
sider a painting. A painting can be described, but it does not follow from this that
the painting has the content that is expressed with the description.4 Similarly, while
describing what a subject experiences is informative, the fact that an experience can
be described does not entail that the relevant experience has the content expressed
with the description.5
The Content Thesis posits that the content of experience is an aspect of ex-
perience proper and not merely associated with the experience. So the Content
Thesis differs from the Association Thesis in kind. After all, if experience is fun-
damentally a matter of representing the environment in a certain way, then
experience will have content that is not merely associated with the experience.
It will have content that is a proper part of the experience. The relevant notion
of “fundamentality” in the Content Thesis marks a denial of the idea that con-
tent is merely associated with experience rather than being a proper part of
experience.
Now, a controversial version of the Content Thesis has it that the relation-
ship between the experiencing subject and the content of her experience is that
of a propositional attitude: the experiencing subject stands in a propositional
attitude to the content of her experience. This Propositional Attitude Thesis pos-
its both that the content of experience is a proposition and that experience is a
matter of standing in a certain attitudinal relation to this proposition, analogous

3
As I have argued elsewhere (see my 2011), some accounts of perceptual content fall prey to the
austere relationalist objections; others (in particular the view I defend which acknowledges mutual
dependence of the relational and representational character of perceptual experience) arguably do not.
4
For a detailed discussion of the relation between the content of pictures and the content of
experiences and mental states more generally, see Crane (2009).
5
Byrne (2009) and Siegel (2010) have presented arguments in support of the view that experience
has content, but arguably their arguments do not establish more than the Association Thesis.
 Reconciliatory Views
202   

to the sense in which one might say that belief is a matter of standing in the
believing relation to the content of the belief. English does not have a word to
denote such a perceptual attitudinal relation. Byrne (2009, p. 437) calls the rela-
tion the ex-ing relation; Pautz (2010, p. 54) calls it the sensorily entertaining re-
lation; Siegel (2010, p. 22) calls it the A-relation. The Propositional Attitude Thesis
is a version of the Content Thesis. However, it is important to keep in mind that
we can accept the Content Thesis without accepting the Propositional Attitude
Thesis: The Content Thesis is committed neither to the content of experience
being a proposition nor to the experiencing subject standing in a propositional
attitude to the content of her experience.
An even more controversial version of the Content Thesis has it that the relation
between the experiencing subject and the content of her experience is an awareness
relation: The experiencing subject stands in an awareness relation to the content
or its constituents, such that this awareness relation grounds the sensory character
of the experience. This Awareness Thesis can be traced back to Russell (1913), who
argued that an experiencing subject stands in acquaintance relations to particulars
that in turn can be understood as the constituents of the proposition that character-
izes her experience.6 In the tradition of Russell, some representationalist views are
formulated in a way that suggests a commitment to the Awareness Thesis. While
the Awareness Thesis entails the Content Thesis, the converse is not the case: we
can accept the Content Thesis without accepting that perceivers stand in any kind
of awareness relation to the content of their experience. Experience can be under-
stood to have content in that the experiencing subject represents her environment
by employing perceptual capacities without the subject standing in an awareness
relation to the content yielded by employing those capacities.
The Awareness Thesis and the Propositional Attitude Thesis carry controversial
commitments that the Content Thesis does not entail. As I will show, we can accept
that perceptual experience is fundamentally a matter of representing one’s envi-
ronment without being committed to these more contentious theses. The Content
Thesis is neither committed to the thesis that the content of experience is a prop-
osition, nor is it committed to the thesis that the experiencing subject stands in
an awareness relation to the content of her experience. This is important since at
least some arguments against the Content Thesis assume that this thesis entails
those more controversial theses.7 Such arguments lose their grip, if one recognizes
that the Content Thesis does not carry the commitments of the more controversial
theses.
So far, we have distinguished different ways of understanding the relation
between the experiencing subject and the content of her experience. A second

6
One could argue however that on Russell’s view, acquaintance with particulars and universals
is more basic than any contentful mental state in that acquaintance with particulars and universals
explains how it is possible to entertain the relevant contents. Thanks to Bill Brewer for pressing me on
this point.
7
See, for example, Travis (2004).
 The Relational and Representational Character of Perceptual Experience   203

choice point one faces is how to understand the relationship between the content
and the sensory character of perceptual experience. One might identify content
and sensory character. Alternatively, one might deny identity but maintain ei-
ther that content is grounded in sensory character or that sensory character is
grounded in content. Or, one might treat content and sensory character as inde-
pendent elements of experience, thereby denying that there is any identity or
grounding relation between content and sensory character.
A third choice point is how to understand the nature of perceptual content. This
choice point contains several levels. One is whether the content is understood in
terms of a Russellian proposition, a possible world proposition, a Fregean sense, or
in some other way. A second is whether perceptual content is conceptually or non-
conceptually structured.8 A third is whether or not the content of experience is prop-
ositionally structured. A fourth is whether the content is (at least in part) externally
individuated and so dependent on the experiencer’s environment, or alternatively
only internally individuated and so independent of the experiencer’s environment.
I will consider the final point in more detail, since the question of whether
perceptual content is environment-dependent is crucial in the debate on whether
experience is fundamentally representational, relational, or fundamentally both
relational and representational.9 According to austere representationalism, the
content of experience is internally individuated in the sense that it is indepen-
dent of the environment of the experiencing subject.10 The view is austere since
it leaves no significant room for a relational component. The only difference be-
tween subjectively indistinguishable experiences in distinct environments is a
difference in the causal relation between the experiencing subject and her envi-
ronment. According to austere representationalists, this difference in causal re-
lation has repercussions neither for the content of the experiences nor for their
sensory character. On such a view, the content of experience can be analyzed in
terms of existentially quantified content of the form that there is an object x that
instantiates a certain property F: (∃x)Fx. So experience represents only that there
is an object with the relevant properties in the external world. No element of the
content depends on whether there is in fact such an object present. Austere repre-
sentationalism has it that the content lays down a condition that something must

8
The debate on whether perceptual content is conceptually or nonconceptually structured is
sometimes understood as a debate about whether perceptual content is structured by Fregean concepts
and not just properties and objects. On this understanding, the first and second level distinguished
above collapses. However, there are ways to understand perceptual content as structured by modes of
presentation without committing oneself to the idea that the content is thereby conceptually structured
(see, for example, Schellenberg 2013). Therefore, I distinguish the two debates. Thanks to Todd Ganson
for pressing me on this point.
9
Nanay (forthcoming) argues that the debate between representationalists and relationalists is
best understood as a debate not about what is fundamental in an account of perceptual experience but
rather as a debate about the individuation of perceptual states.
10
McGinn (1982), Davies (1992), Tye (1995), Lycan (1996), and Byrne (2001) among others
have defended views that are committed to perceptual content being independent of the experiencer’s
environment.
 Reconciliatory Views
204   

satisfy to be the object determined by the content. The condition to be satisfied

does not depend on the object that satisfies it. So the relation between content
and object is simply the semantic relation of satisfaction. Of course, the object of
the experience does not fall out of the picture altogether on an austere represen-
tationalist view. The content of the experience is accurate only if there is an object
at the relevant location that instantiates the properties specified by the content.
But the important point is that whether an object of the right kind is present has
a bearing only on the accuracy of the content, not on the content itself.
We can contrast such a non-relational view with a relational view of perceptual
content, that is, a view on which perceptual content is understood to be inherently
relational and so at least in part environment-dependent. For something to be the
object of a relational content, the content must constitutively depend at least in part
on that very object. So while a non-relational content is the very same regardless of
the situation in which the subject experiences, a relational content differ depending
on what environmental particular (if any) the subject is related to. The token rela-
tional content co-varies with the environment of the experiencing subject. In the
case of a successful perceptual experience, the token content determines a referent.
Insofar as the token relational content is individuated in part by the environmental
particulars perceived, it is at least in part environment-dependent.

8.2. Perception and Relations

Austere relationalists have formulated at least six different objections to the

Content Thesis. They can be stated as follows:

The Particularity Objection: Representationalist views cannot adequately

account for the fact that we see particulars and have perceptual
knowledge of particulars (e.g., Campbell 2002; Martin 2002b).
The Indeterminacy Objection: If perception has representational content,
then the way an object looks on a given occasion must fix what rep-
resentational content the perception has. However, the way an object
looks on a given occasion does not fix what representational content
the perception has. Therefore, perception does not have representa-
tional content (e.g., Travis 2004).
The Accuracy Condition Objection: Perception is a relation between a per-
ceiving subject and her environment or alternatively an event in which
such a relation obtains. Relations and events do not have accuracy
conditions. So perception is not the kind of thing that can be accu-
rate or inaccurate. If accounting for accuracy conditions is the reason
for introducing content, then denying that experience has accuracy
conditions undermines at least this reason for the Content Thesis (e.g.,
Brewer 2006).
 The Relational and Representational Character of Perceptual Experience   205

The Phenomenological Objection: Representationalist views misconstrue

the phenomenological basis of perceptual experience insofar as they
detach the sensory character of experience from relations to qualita-
tive features of the environment (e.g., Campbell 2002; Martin 2002a;
Brewer 2007).
The Epistemological Objection: Representationalist views do not properly
account for the epistemological role of perceptual experience. Only
if perceptual experience is itself not representational can it constitute
the evidential basis for demonstrative thoughts and ultimately percep-
tual knowledge (e.g., Campbell 2002).
The Grounding Objection: Representationalist views cannot adequately
account for the fact that perceptual relations to the environment
provide the ground for the possibility of thought and language (e.g.,
Campbell 2002; Brewer 2006).

I have shown elsewhere that a representationalist view on which perceptual con-

tent is understood to be inherently relational does not fall prey to these objections
and will not rehearse those arguments here (see my 2010, 2011). I will proceed
immediately to stating the relationalist view that has been formulated in response
to these objections.
The central positive idea of relationalism is that perceptual experience is not
fundamentally a matter of representing, but rather fundamentally a matter of a
subject standing in a perceptual relation to a material, mind-independent object, a
property that this object instantiates, an event, or a combination thereof (Campbell
2002; Brewer 2006). Alternatively, experience is thought of as an event in which such
a relation obtains (Martin 2002a). Views differ further on whether subjects are per-
ceptually related only to objects in an environment (Brewer 2006, 2011) or whether
they are related also to the properties that these objects instantiate (Campbell 2002).
Views differ moreover on how the perceptual relation is understood: It can be un-
derstood as a causal relation, a sensory relation such as an awareness relation, or as
an epistemic relation such as an acquaintance relation. Finally, views differ on how
they oppose representationalism: While all austere relationalists agree that no appeal
to content is necessary to give a good account of perceptual experience, some go a
step further in arguing that representationalism is less attractive than relationalism
in explaining certain phenomena, or indeed that representationalism cannot explain
certain phenomena (e.g., Brewer 2006, 2011).
What austere relationalist views have in common is that they endorse the negative
thesis that no appeal to representational content is necessary in a philosophical account
of perceptual experience, in conjunction with the positive thesis that any perception es-
sentially involves at least three components: a subject, the environment of the subject,
and a perceptual relation between the subject and certain elements of her environment.
For the sake of definiteness, I will work with the case of a subject being percep-
tually related to a mind-independent object that instantiates a perceivable property.
 Reconciliatory Views
206   

Everything I will say about this case needs to be modified only slightly to fit with other
versions of austere relationalism. Given this case, the austere relationalist thesis can be
articulated in the following way: A subject perceives a particular white cup only if she
is perceptually related to that particular white cup; no appeal to content is necessary
to fully explain the nature of the subject’s perceptual experience. Being perceptually
related to a white cup may in turn be analyzed in terms of being perceptually related
to a cup instantiating whiteness, where the relevant object and property-instance are
collocated. More generally, subject S perceives object o as instantiating property F
only if S is perceptually related to o and an instance of F, where o and the instance of
F are collocated.
It will be helpful to make three clarifications about the view at stake. First, aus-
tere relationalists do not deny that beliefs and judgments are formed on the basis of
perception. So what is contentious is not whether perception brings about mental
states with content. The questions at stake are rather whether this content is an as-
pect of perception proper and whether the thesis that experience is representational
is fundamental in an account of perceptual experience. Second, austere relational-
ists do not contest that perception involves cognitive or neural processing that can
be characterized in terms of representations (Campbell 2002, 2010), but insist rather
that no appeal to content is necessary to explain the nature of the awareness of our
surroundings that we have as a consequence of this cognitive processing. So while for
example Campbell allows that representations play a role on a subpersonal level, he
denies that any appeal to representations is necessary to explain perception on a per-
sonal level. Finally, austere relationalists need not deny that we can articulate proposi-
tions to express what we experience. Acknowledging that a subject can articulate such
propositions entails no commitment to positing that her experience itself has the con-
tent articulated. It might just be that the propositions articulated are merely associ-
ated with the experience. Austere relationalists can accept the Association Thesis. So in
order to establish the Content Thesis, we cannot simply appeal to the fact that we can
articulate propositions to express what we experience. Appealing to such a fact would
merely establish the Association Thesis. We need to show that these propositions or
contents are a proper aspect of experience and indeed that they are a fundamental
aspect of experience.11

8.3. Perceptual Content Defended Again

I will present my argument for the Content Thesis in two stages. I will first put
forward the Master Argument for perceptual content. The Master Argument is

11
For an argument that disposing of perceptual representations is inconsistent with empirical
findings about dorsal perception and about the multimodality of perception, see Nanay (this volume).
For a discussion of how the ventral and the dorsal stream work together in visual experience, see also
Wu (forthcoming). For a critical discussion of recent representationalist views on empirical grounds,
see Bronner, Kerr, and Ganson (forthcoming).
 The Relational and Representational Character of Perceptual Experience   207

compatible with accepting only the Association Thesis, and so is not sufficient for
establishing the Content Thesis. However, we will need it to establish that thesis,
since it clarifies the relevant notion of content. In the next section, I will put for-
ward the Argument for Relational Content, which builds on the notion of content
established by the Master Argument. By arguing that perception is fundamentally
a matter of representing one’s environment, I will be in a position to conclude the
Content Thesis. By arguing that perceptual content is best understood to be inher-
ently relational, I will be in a position to conclude that perceptual experience is
fundamentally both relational and representational.

8.3.1. THE MASTER ARGUMENT FOR PERCEPTUAL CONTENT

The mere fact that the environment sensorily seems a certain way when one per-
ceives supports a standard notion of perceptual content. The Master Argument for
perceptual content goes as follows:

P1: If a subject is perceptually related to her environment (while not suf-

fering from blindsight or any other form of unconscious perception),
then she is sensorily aware of her environment.
P2: If a subject is sensorily aware of her environment, then her environ-
ment sensorily seems a certain way to her.
P3: If her environment sensorily seems a certain way to her, then she has
an experience with content C, where C corresponds to the way her
environment sensorily seems to her.
Conclusion 1: If a subject is perceptually related to her environment (while
not suffering from blindsight or any other form of unconscious percep-
tion), then she has an experience with content C, where C corresponds
to the way her environment sensorily seems to her.
P4: Her environment is either the way it sensorily seems to her or it is dif-
ferent from the way it sensorily seems to her.
P5: If a subject has an experience with content C, then C is either accurate
(if her environment is the way it sensorily seems to her) or inaccurate
(if her environment is not the way it sensorily seems to her).
Conclusion 2: If a subject is perceptually related to her environment
(while not suffering from blindsight or any other form of unconscious
perception), then the content of her experience is either accurate or
inaccurate.12

12
While in my (2011) I formulate this argument in terms of “aware” and “seems,” I here speak
of “sensorily aware” and “sensorily seems.” This clarifies the relevant notion of “aware” and “seems.”
Furthermore, while in my (2011) I formulate this argument by saying that the subject is perceptually
related to the world, I here formulate the same argument by saying that the subject is perceptually
related to her environment. This clarifies which part of the world the subject is perceptually related to.
Brewer’s response to my argument is in no way affected by these two clarifications.
 Reconciliatory Views
208   

Brewer accepts the first two premises of the argument. It is not contentious that
if we are perceptually related to our environment, then we will be sensorily aware
of that environment (P1). Moreover, it is not contentious that if we are sensorily
aware of our environment, then our environment will sensorily seem a certain way
to us (P2).
We can moreover recognize P3 to be true, if we recognize that there is a
notion of content on which the content of experience corresponds to the way the
environment sensorily seems to the experiencing subject. Let’s call this connec-
tion between content and the way the environment seems the seems-content link.
Since we are talking only of sensory seemings the relevant cases are constrained to
those in which a subject is hearing, seeing, smelling, touching, or experiencing the
environment in some other sensory mode, or a combination of sensory modes. If
we recognize the seems-content link, then the idea that the environment can seem
a certain way to a subject without her being in a contentful mental state becomes
impossible. But the idea that the environment can seem a certain way to a subject
without her being in a contentful mental state is precisely the idea that austere
relationalism relies on.
Recognizing the seems-content link is compatible with accepting that any given
scene can be perceived in many different ways. The way the environment seems
to the perceiver may change from moment to moment even as her gaze remains
steady. Say she is looking at a pig. She can direct her attention at its shape, its color,
the texture of its skin, or any combination of these features. As her attention shifts,
her sensory character will change. One or more propositions can be associated with
every one of these phenomenal states and thus with every one of these ways that the
environment may seem to her. All of these propositions or sets of propositions are
equally legitimate contents of possible experiences she may enjoy while beholding
the pig. Nevertheless, at any given moment the environment will seem to her to be
one single way. This is all that we need to establish the seems-content link.13
Once one has recognized the seems-content link, only minor further com-
mitments are necessary to establish that the way the environment seems to an
experiencing subject is assessable for accuracy. In virtue of a subject perceiving
the environment, it seems a certain way to her. The way the environment seems to
a subject determines the way the environment would have to be for the content of
her experience to be accurate. The environment is either the way it seems to her
or it is different from the way it seems to her (P4). If the environment is the way it
seems to her, then the content of the experience is accurate. In all other cases, the
content of the very same experience is inaccurate. So if a subject has an experience

13
For a detailed discussion of how these premises must be interpreted depending on whether
one understands the seemings in question comparatively or noncomparatively, see my (2011). For
present purposes we can ignore this detail, since the Master Argument can be accepted regardless of
whether seemings are understood comparatively or noncomparatively. For the distinction between
the comparative and noncomparative use of appearance words, see Chisholm (1957, pp. 50–53) and
Jackson (1977, pp. 30ff.).
 The Relational and Representational Character of Perceptual Experience   209

with a particular content, then this content is either accurate or inaccurate (P5). It
follows from this, together with P1 and P2 of the Master Argument, that if a subject
is perceptually related to the environment, then the way the environment seems to
her is assessable for accuracy. Together with P3, it follows that if a subject is per-
ceptually related to the environment, then the content of her experience is either
accurate or inaccurate.14

8.3.2. THE ARGUMENT FOR RELATIONAL CONTENT

The Master Argument does not on its own establish the Content Thesis. After all, it
makes no claims about whether experience is fundamentally a matter of represent-
ing. On a weak reading of the Master Argument, it establishes only the Association
Thesis. In order to establish the Content Thesis, we need an additional argument.
One such additional argument is the Argument for Relational Content:

From P1 and P2: If a subject S is perceptually related to her environment

(while not suffering from blindsight or any other form of unconscious
perception), then S’s environment sensorily seems a certain way to her.
P6: If a subject S’s environment sensorily seems a certain way to her, then S is
employing perceptual capacities that constitute the way her environment
sensorily seems to her.
P7: If S is employing perceptual capacities that constitute the way her envi-
ronment sensorily seems to her, then S is representing her environment
in virtue of employing perceptual capacities.
P8: S is representing her environment in virtue of employing perceptual
capacities.
P9: If S is representing her environment in virtue of employing perceptual
capacities, then S has a perceptual experience that is fundamentally a
matter of representing her environment as being a certain way.

14
It will be helpful to make two clarifications about the thesis that the way the environment seems
to one determines accuracy conditions. First, there can be phenomenal differences between experiences
that are not a matter of how the environment seems to one, but rather a matter of how one experiences.
If I am shortsighted, my experience may be blurry, but I need not perceive the environment as being
blurry. I have argued that perceptual content corresponds to the way the environment seems to the
perceiver. This seems-content link is neutral on how those aspects of sensory character are accounted
for that do not pertain to the way the environment seems to the perceiver. Second, the environment is
arguably rarely and perhaps never the way it seems to us to be. We perceive plates to be round, although
their shapes are much more complicated. We see surfaces to be colored, but it has been argued that
surfaces do not have color properties. We see our environment to be populated by objects, but it has
been argued that there really are no objects or at least not the kind of objects that we seem to see. In
order to accommodate these phenomena, we need to loosen the notion of accuracy conditions in play
or alternatively we need to accept widespread but explicable perceptual error. For a detailed discussion
of this set of issues, see Pautz (2009) and Siegel (2010). If my argument for the thesis that experience
has accuracy conditions holds, then it holds regardless of what stance one takes on this set of issues.
 Reconciliatory Views
210   

Conclusion 3: S has a perceptual experience that is fundamentally a matter

of representing her environment as being a certain way. (from P8 and P9)
P10: Perceptual capacities are by their nature linked to what they single
out in the good case.
P11: If S is representing her environment in virtue of employing percep-
tual capacities, then S has a perceptual experience that is fundamen-
tally a matter of being related to her environment in a certain way.
Conclusion 4: S has a perceptual experience that is fundamentally a matter
of being related to her environment in a certain way. (from P8 and P11)

The basic idea in support of this argument is that when we perceive, we employ
perceptual capacities by means of which we differentiate and single out particulars
in our environment. The relevant particulars are external and mind-independent
objects, events, property-instances, and instances of relations. Sensory seemings
are understood as individuated by employing such perceptual capacities in a sen-
sory mode, that is, modes such as seeing, hearing, touching, smelling, or tasting.
Employing such perceptual capacities in turn constitutes a mental state with
content.
Say we perceive a white cup on a desk. We employ our perceptual capacity to
discriminate white from other colors and to single out white in our environment.
Similarly we employ our capacity to differentiate and single out cup-shapes from,
say, computer-shapes and lamp-shapes. It is not clear what it would be to single
out an object in our environment without employing capacities of this kind. If this
is right, then it is in virtue of employing such capacities that we are in a sensory
state that is of a white cup.
Now how should we understand the capacities in play? They can be under-
stood as conceptual capacities or as nonconceptual capacities. There are pow-
erful reasons to understand perceptual content as nonconceptually structured.15
Therefore, I will focus on nonconceptual perceptual capacities. Indeed, I will focus
on the cognitively most low-level nonconceptual perceptual capacities, namely,
discriminatory, selective capacities. A discriminatory, selective capacity functions
to differentiate and single out, where singling out a particular is a proto-conceptual
analog of referring to a particular.16 So if we possess the discriminatory, selective
capacity that functions to differentiate and single out red, we are in a position to
differentiate instances of red from other colors in our environment and to single
out instances of red. More generally, to possess a discriminatory, selective capacity
is to be in a position to differentiate and single out the type of particulars that the
capacity concerns, were one related to such a particular. So if we possess such a

15
For discussion of nonconceptual content, see Peacocke (1998), Heck (2000), and Speaks (2005).
For a recent defense of the idea that perceptual content is conceptually structured, see Glüer (2009)
and Bengson et al. (2011).
16
In some cases, a discriminatory capacity may also function to type the kind of particulars that
the capacity concerns, but this is not an essential feature of the capacities in play.
 The Relational and Representational Character of Perceptual Experience   211

capacity, then—assuming no finking, masking, or other exotic case is involved (see

Lewis 1997)—the following counterfactual should hold: If we were perceptually
related to a particular that the capacity functions to single out, then we would be
in a position to single out such a particular.
What happens in hallucination? Although such capacities are determined by
functional connections between perceivers and their environment, arguably they
can be employed even if one is misperceiving or hallucinating. After all, one could
be prompted to employ the capacities due to nonstandard circumstances, such as
unusual brain stimulation or misleading distal input. If this is right, then we can
employ a perceptual capacity even if a relevant particular is not present—where a
relevant particular is a particular of the type that the capacity functions to single
out. The capacities employed account for the fact that in hallucinations we purport
to single out particulars. Since in hallucination, we are not perceptually related
to a particular, we fail to single out a particular in our environment. We merely
purport to single out a particular. As a consequence, the capacities employed are
baseless. They are baseless in the sense that the usual target of discrimination and
selection—external, mind-independent particulars—are absent. Analogously, if
we employ concepts, but fail to refer, the concepts employed remain empty. So if
we hallucinate a white cup on a desk, we employ the capacity to discriminate and
single out white from other colors and we employ the capacity to differentiate and
single out cup-shapes from, say, computer-shapes and lamp-shapes. Since we are
hallucinating rather than perceiving and so not perceptually related to a white cup,
the capacities we employ are baseless. Yet even though we fail to single out any
white cup, we are in a sensory state that is as of a white cup in virtue of employing
the capacity to discriminate and single out white from other colors and cup-shapes
from other shapes.
If this is right, then the very same perceptual capacity can be employed
such that a particular is successfully singled out or employed without success-
fully singling out any particular. In this sense, employing perceptual capacities
not only yields sensory seemings but moreover constitutes accuracy conditions.
So employing perceptual capacities has all the hallmarks of content insofar as it
yields something that is entertainable and that can be accurate or inaccurate. So if
S is employing perceptual capacities thereby constituting the way her environment
sensorily seems to her, then S is representing her environment in virtue of employ-
ing perceptual capacities. Indeed, insofar as employing perceptual capacities yields
content and the employment of the perceptual capacities constitute the experience,
the content is a proper part of experience rather than merely associated with the
experience. If S is representing her environment in virtue of employing percep-
tual capacities, then S has a perceptual experience that is fundamentally a matter
of representing her environment as being a certain way. So the subject bears the
representation relation to the content rather than the mere association relation. As
a consequence, the notion of content established by P6‒P9 goes beyond that estab-
lished by the Master Argument. Since S is arbitrarily chosen, Conclusion 3 holds for
 Reconciliatory Views
212   

any perceiver and so characterizes perceptual experience generally. Therefore, the

Content Thesis follows from Conclusion 3. So by building on the Master Argument,
P6‒P9 establish the Content Thesis.
In order to show why the idea that perceptual experience is a matter of employ-
ing perceptual capacities supports the view that experience is fundamentally not
just representational but moreover relational, it is crucial to take a closer look at
perceptual capacities. The function of perceptual capacities is to differentiate and
single out the type of particulars that the capacity is of. It would be unclear what
it would mean to possess a perceptual capacity, the very function of which is to
single out a type of particular, without being in a position to single out such a par-
ticular when perceptually related to one. An example will help illustrate the point.
If we possess the capacity to discriminate and single out white from other colors,
we can use this capacity to single out white in our environment. Were we not in a
position to use our capacity in this way, when perceptually related to an instance
of white, we would not count as possessing the capacity. If this is right and percep-
tual content is yielded by employing such capacities, then relations to objects and
property-instances are implicated in the very nature of perceptual content.
If the fact that perceptual capacities single out particulars in some situations
but not in others has any semantic significance, then the content ensuing from
employing these capacities will depend at least in part on the environment in
which they are employed. After all, relations to particulars are implicated in the
very nature of perceptual content, if perceptual content is yielded by employing
perceptual capacities and such capacities function to single out particulars, then
relations to particulars are a fundamental part of perceptual content. So insofar as
the perceptual capacities that yield content function to single out particulars, per-
ceptual experience is fundamentally both relational and representational.17
The notion of perceptual content in play can be specified more specifically
as follows: Employing perceptual capacities yields a content type that subjectively
indistinguishable experiences have in common. The token content ensues from
employing perceptual capacities in a particular environment such that the token
content co-varies with the environment of the experiencing subject. Since the per-
ceptual capacities employed are the very same in subjectively indistinguishable
experiences, such experiences have the same content type. Individuating expe-
riences by a content type amounts to individuating experiences with regard to
the experiencing subject’s sensory state. In virtue of recognizing that perceptual
experience is fundamentally both relational and representational, the suggested
approach rejects all ways of factorizing perceptual content into internal and
external components.18

17
For an alternative way of avoiding the pitfalls of both austere representationalism and austere
relationalism, see Dorsch (2013).
18
For a detailed development of the semantic nature of such token contents, see my (2010). For a
helpful discussion of the problems of factorizing mental content into internal and external components,
see Williamson (2000, 2006). See also Burge (2010).
 The Relational and Representational Character of Perceptual Experience   213

It is important to note that the content yielded by employing perceptual

capacities need neither be conceptually nor propositionally structured. Indeed,
the thesis that perceptual experience has content in virtue of experiencing subjects
employing perceptual capacities allows for a substantive a way of understanding
perceptual content as non-conceptual and non-propositional. Moreover, this
thesis neither implies that the experiencing subject stand in a propositional atti-
tude to the content of her experience nor does it rely on there being such a relation
between the subject and content of an experience. So there is no need to say that
the experiencing subject ‘exes’ that p—to use Byrne’s (2009) phrase.
A further advantage of the suggested view is that it does not imply that expe-
riences have an attributional structure, such as object o is F. After all, the view
does not depend on the idea that we are perceptually related to objects. It depends
only on the idea that we are perceptually related to particulars. Those particulars
need not be objects. They could be events or property-instances. So we may well be
perceiving only property-instances and so not be attributing any property to a per-
ceived object. Olfactory, gustatory, and tactile experiences may not have an attribu-
tional structure or at least not typically have such a structure.19 Since the suggested
view neither implies nor presupposes that perception has an attributional structure,
the view applies not just to visual and auditory experiences, but also olfactory, gus-
tatory, and tactile experiences among other kinds of experiences.
Austere relationalism has it that for perceptual experience to ground per-
ceptual knowledge of particular objects, there must be a phenomenal difference
between experiences of qualitatively indistinguishable but numerically distinct
objects. This is an unfortunate consequence of austere relationalism. On the view of
content developed, we can avoid this unfortunate consequence. I have argued that
the content of experience is in part dependent on the experiencer’s environment.
By arguing that only the part of the experience that is not environment-dependent
grounds the sensory character of the experience, the provided view allows that
experiences of numerically distinct but qualitatively indistinguishable objects
differ in content, while having the same sensory character.
The view suggested is fundamentally representational, insofar as perceptual
content is yielded by employing perceptual capacities and the employment of per-
ceptual capacities constitute the experience. It is fundamentally relational insofar
as a perception has the particular content it has because the experiencing subject is
perceptually related to an environmental particular. Moreover, insofar as the content
of experience is yielded by employing perceptual capacities the possession of which
grounds the ability to single out objects and property-instances, relations to objects
and property-instances are implicated in the very nature of experiential content.

19
For discussion, see Smith (2007), Batty (2010, 2011), and Fulkerson (2011). For discussions
of whether auditory experiences have attributional structure, see Nudds (2001), O’Callaghan (2010),
Ivanov (2011), Phillips (2013), and Matthen (forthcoming). See Macpherson (2011) for different ways
of individuating the senses.
 Reconciliatory Views
214   

8.4. Rejoinder to Relationalist Response

In his book Perception and its Objects, Brewer responds to my argument. His
focus is on the first part of my Master Argument. He accepts the first two prem-
ises, but rejects the third premise of the argument. Recall that P3 has it that if a
subject’s environment seems a certain way to her, then she has an experience with
content C, where C corresponds to the way her environment seems to her.
Brewer notes that the Master Argument can be understood in ways compatible
with his view. As I have been careful to note, the Master Argument itself is com-
patible with just the Association Thesis. Accordingly, Brewer considers a strength-
ening of the Master Argument on which its conclusion is in genuine tension with
his view. He does so by replacing my P3 with a stronger thesis, that I will call P3*.
P3*: “the idea of a person having an experience whose most fundamental
nature is to be elucidated in terms of some kind of representational content
C” (2011, p. 61).
My P6‒P9 establish P3*, so my Argument for Relational Content is built into P3*.
Brewer rejects P3* on the following grounds:
I simply deny that it follows from the fact that there are truths of the form ‘o
looks F’ that apply to a person S in virtue of her perceptual relation with o,
that the most fundamental nature of that experience with the representational
content (of some kind) that o is F. On my interpretation of Schellenberg’s
Master Argument, this is the transition articulated explicitly by [P3*]. The
account of looks offered in ch. 5 proves that [P3*] is false on this interpre-
tation. For I explain there precisely how various looks claims apply to S in
virtue of her perceptual relation with the world around her without assuming
that the very nature of that perceptual relation is itself to be characterized in
terms of any corresponding worldly representational content. The perceptual
relation between perceivers and the mind-independent physical objects in
the world around them is on that account more basic than any such repre-
sentational contents and grounds the truth of the looks claims that perfectly
reasonably inspire talk of perceptual representation (2011, p. 60).
In a nutshell, his objection to my argument is that the nature of the perceptual
relation should not be understood as fundamentally involving representational
content. On Brewer’s view, the nature of perceptual experience is fundamentally
only relational.20 Brewer does not consider the prospect that perceptual experience
could be fundamentally both relational and representational. He seems to assume
that these options are exclusive. One of my main points is that these options are

20
Brewer focuses exclusively on the objects to which the perceiver is perceptually related, while
I focus on the environment to which the perceiver is perceptually related. The environment contains
objects, property-instances, relations, and events. This is a significant difference between Brewer’s
approach and my own. It can however be ignored for present purposes.
 The Relational and Representational Character of Perceptual Experience   215

not exclusive but complementary. I argue that perceptual experience is fundamen-

tally both relational and representational. So Brewer and I agree that experience
is fundamentally relational. What we disagree about is whether it is in addition
fundamentally representational.21
Brewer’s central argument against experience being fundamentally represen-
tational goes as follows:
If S sees a mind-independent physical object o, then there are certainly
(perhaps indefinitely) many true sentences of the form ‘o looks F’, but
I would . . . deny that S’s seeing o itself consists in the truth of those sentences
or can be fruitfully illuminated by listing the facts that o looks F1, o looks
F2, . . . , o looks Fi, etc., or the fact that is visually seems to S that o is F1, o is
F2, . . . , o is Fi, etc. S’s seeing o, her perceptual experiential relation with that
particular mind-independent physical object is more basic than any such facts
and is what grounds the truth of all those sentences. (Brewer 2011, p. 62f.)
Brewer argues here that perceptual relations between a perceiver and an object
are more basic than sentences expressing how the object looks to the perceiver.
We can agree with Brewer that the truth of sentences is derivative of their truth-
makers. So we can agree with him that “S’s seeing o, her perceptual experiential
relation with that particular mind-independent physical object is more basic than
any such facts and is what grounds the truth of all those sentences,” where these
sentences express how the environment looks to a person. However, what is at
issue in the debate on whether experience has content is not the relation between
sentences and their truthmakers. The thesis that experience is fundamentally a
matter of representing the environment is neither a thesis about sentences nor
a thesis about sentential truth. It is a thesis about mental content. No one thinks
that perception is fundamentally a matter of sentences that express how the object
looks to perceivers being true.22
While everyone can accept that perception is not fundamentally a matter of
such sentences being true, there is an argument in close vicinity of Brewer’s argu-
ment that strengthens his case against representationalism. This argument has the
same form as Brewer’s argument and preserves the intuitions guiding his argu-
ment, but is about mental content rather than sentences:
If S sees a mind-independent physical object o, then there are certainly (per-
haps indefinitely) many accurate mental contents of the form ‘o looks F’, but S’s

21
I would also quibble with Brewer’s wording when he characterizes representationalism as
endorsing the thesis that “the very nature of that perceptual relation is itself to be characterized in
terms of any corresponding worldly representational content” (2011, p. 60). I would not say that the
perceptual relation can be understood in terms of representational content, but only that perceptual
experience—rather than the perceptual relation—can be understood (at least in part) in terms of
representational content.
22
For a discussion of the relation between mental content and linguistic meaning, see Speaks
(2006). On perceptual reports, see Brogaard (forthcoming).
 Reconciliatory Views
216   

seeing o itself does not consist in the accuracy of those mental contents or can be
fruitfully illuminated by listing the facts that o looks F1, o looks F2, . . . ., o looks Fi,
etc., or the fact that it visually seems to S that o is F1, o is F2, . . . , o is Fi, etc. S’s see-
ing o, her perceptual experiential relation with that particular mind-independent
physical object is more basic than any such facts and is what grounds the accuracy
of all those mental contents.
In response to this rephrased version of Brewer’s argument, we can say that we
can accept that S seeing o is more basic than the accuracy of the mental content.
However, we can accept this without accepting that S seeing o is more basic than the
fact that S has an experience, which is fundamentally a matter of representing o. So
we can accept that S seeing o is more basic than the accuracy of the mental content,
while acknowledging that S seeing o is not more basic than S representing o. I argue
that S being perceptually related to o and S representing o are equally fundamental.
A further central reason for Brewer to reject representationalism is that
according to him representationalism cannot account for illusions. Brewer focuses
on the Müller-Lyer illusion, an illusion prompted by two lines that in fact have
the same length, but that seem to have different lengths due to outward looking
hashes on one line and inward looking hashes on the other line. He discusses sev-
eral ways a representationalist could analyze what the content of such an illusion
could be (2011, pp. 65–69). For the sake of argument, I will assume that the ways
of accounting for illusions in a representationalist framework that Brewer con-
siders do not work for the very reasons that Brewer cites. Rather than take issue
with Brewer’s arguments, I will put forward a way that a representationalist could
account for illusions that Brewer does not consider.
First, consider perceiving a cup at an angle. One way of analyzing what we rep-
resent when we perceive a cup at an angle is that we represent the shape of the cup
in two ways: <the cup has shapeʹ, the cup has shapeʹʹ>. The single primed property
is an intrinsic property and the double primed property is a situation-dependent
property, that is, it is a property that is determined by one’s location and the intrinsic
shape of the cup. More specifically, a situation-dependent property is a (noncon-
stant) function of an intrinsic property and one or more situational features, that
is, features of the environment that determine how objects are presented such as
the lighting conditions and the subject’s location in relation to perceived objects.
This means that fixing the intrinsic properties of an object and the situational fea-
tures fixes the situation-dependent properties. Furthermore, situation-dependent
properties are ontologically dependent on and exclusively sensitive to intrinsic
properties and situational features. I understand the intrinsic properties, of say a
white cup, to include among other properties, the shape and size of the cup. More
generally, intrinsic properties of an object are the properties that an object has re-
gardless of the situational features. They are the properties that an object has that
do not depend on the object’s relations to other individuals distinct from itself. 23

23
For a development of the notion of situation-dependent properties, see my (2008). For a critical
discussion, see Jagnow (2012) and Madary (2012).
 The Relational and Representational Character of Perceptual Experience   217

While the single and double primed properties are different kinds of proper-
ties, a perceiving subject need not represent these metaphysical facts. The perceiv-
ing subject may just represent that the cup has a certain shape in one respect and a
different shape in another respect. So perception need not represent the metaphys-
ical basis of the distinction between shapeʹ and shapeʹʹ. For perceptual content to
be consistent, it is sufficient that a distinction is represented. It is not necessary
that it is represented what the metaphysical basis for that distinction is.
Now consider the Müller-Lyer illusion. One could say that the two lines that
prompt the Müller-Lyer illusion have the same length in one respect and have
different lengths in another respect. So one can distinguish two conceptions of
length: Call these lengthsʹ and lengthsʹʹ. In light of this distinction, we can say that
when we are perceptually related to the two lines, we represent <A and B have
the same lengths′, A and B have different lengthsʹ′>. The primes mark that one
distinguishes between different respects. Due to the primes, the content of our
experience is not inconsistent. In the case of the Müller-Lyer illusion, the sense
in which the lines look different in length cannot in any straightforward way be
analyzed in terms of situation-dependent properties. After all, there is no situ-
ational feature in the environment in virtue of which the lines look different in
length. However, the way in which we represent the difference between intrinsic
and situation-dependent properties can be exploited for an analysis of how we
represent the lines. As in the case of seeing the cup, we can say that the single and
double primed properties lengthsʹ and lengthsʹʹ are different kinds of properties,
but that a perceiving subject need not represent these metaphysical facts. For per-
ceptual content to be consistent, it is sufficient that a distinction is represented.
If this is right, then we have no reason to reject that experience is fundamentally
a matter of representing one’s environment to account for illusions. So again, we
have no reason to reject the Content Thesis.

References

Batty, C. (2010). A representational account of olfactory experience. Canadian Journal of

Philosophy, 40, 511–538.
Batty, C. (2011). Smelling lessons. Philosophical Studies, 153, 161–174.
Bengson, J., Grube, E., & Korman, D. (2011). A new framework for conceptualism. Noûs,
45, 167–189.
Brewer, B. (2006). Perception and content. European Journal of Philosophy, 14, 165–181.
Brewer, B. (2007). Perception and its object. Philosophical Studies, 132, 87–97.
Brewer, B. (2011). Perception and its Objects. Oxford: Oxford University Press.
Brogaard, B. (forthcoming). Perceptual reports. In M. Matthen (Ed.), Oxford Handbook of
the Philosophy of Perception. Oxford: Oxford University Press.
Bronner, B., Kerr, A., & Ganson, T. (forthcoming). Burge’s defense of perceptual content.
Philosophy and Phenomenological Research.
Burge, T. (2010). Origins of Objectivity. Oxford: Oxford University Press.
Byrne, A. (2001). Intentionalism defended. The Philosophical Review, 110, 199–240.
 Reconciliatory Views
218   

Byrne, A. (2009). Experience and content. Philosophical Quarterly, 59, 429–451.

Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press.
Campbell, J. (2010). Demonstrative reference: The relational view of experience and the
proximality principle. In R. Jeshion (Ed.), New Essays on Singular Thought (pp. 193–
212). Oxford: Oxford University Press.
Chisholm, R. (1957). Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press.
Crane, T. (2009). Is perception a propositional attitude? Philosophical Quarterly, 59,
452–469.
Davies, M. (1992). Perceptual content and local supervenience. Proceedings of the Aristotelian
Society, 92, 21–45.
Dorsch, F. (2013). Experience and introspection. In F. Macpherson & D. Platchias (Eds.),
Hallucination (pp. 69–165). Cambridge, MA: MIT Press.
Fish, W. (2009). Perception, Hallucination, and Illusion. New York: Oxford University Press.
Fulkerson, M. (2011). The unity of haptic touch. Philosophical Psychology, 24, 493–516.
Genone, J. (forthcoming). Appearance and illusion. Mind.
Glüer, K. (2009). In defence of a doxastic account of experience. Mind and Language, 24,
297–327.
Gupta, A. (2012). An account of conscious experience. Analytic Philosophy, 53, 1–29.
Heck, R. (2000). Nonconceptual content and the space of reasons. The Philosophical Review,
109, 483–523.
Ivanov, I. (2011). Pains and sounds. Journal of Consciousness Studies, 18, 143–163.
Jackson, F. (1977). Perception. Cambridge: Cambridge University Press.
Jagnow, R. (2012). Representationalism and the perspectival character of perceptual experi-
ence. Philosophical Studies, 157, 227–249.
Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 120, 113–183.
Johnston, M. (2006). Better than mere knowledge? The function of sensory awareness. In T.
Gendler & J. Hawthorne (Eds.), Perceptual Experience (pp. 260–290). Oxford: Oxford
University Press.
Lewis, D. (1997). Finkish Dispositions. Philosophical Quarterly, 47, 143–158.
Lycan, W. G. (1996). Consciousness and Experience. Cambridge, MA: MIT Press.
Macpherson, F. (2011). Individuating the senses. In F. Macpherson (Ed.), The Senses:
Classical and Contemporary Readings (pp. 3–46). Oxford: Oxford University Press.
Madary, M. (2012). Anticipation and variation in visual content. Philosophical Studies, 165,
335–347.
Martin, M. G. F. (2002a). The transparency of experience. Mind and Language, 17, 376–425.
Martin, M. G. F. (2002b). Particular thoughts and singular thoughts. In A. O’Hear (Ed.),
Logic, Thought and Language (pp. 173–214). Cambridge: Cambridge University Press.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 103, 37–89.
Matthen, M. (forthcoming). On the diversity of auditory objects. Review of Philosophy and
Psychology.
McGinn, C. (1982). The Character of Mind. Oxford: Oxford University Press.
Nanay, B. (forthcoming). The representationalism versus relationalism debate: Explanatory
contextualism about perception. European Journal of Philosophy.
Nudds, M. (2001). Experiencing the production of sounds. European Journal of Philosophy,
9, 210–229.
O’Callaghan, C. (2010). Sounds: A Philosophical Theory. Oxford: Oxford University Press.
 The Relational and Representational Character of Perceptual Experience   219

Pautz, A. (2009). What are the contents of experiences? Philosophical Quarterly, 59, 483–507.
Pautz, A. (2010). Why explain visual experience in terms of content? In B. Nanay (Ed.),
Perceiving the World (pp. 254–310). Oxford: Oxford University Press.
Peacocke, C. (1998). Nonconceptual content defended. Philosophy and Phenomenological
Research, 63, 381–388.
Phillips, I. (2013). Hearing and hallucinating silence. In F. Macpherson & D. Platchias
(Eds.), Hallucination (pp. 333–360). Cambridge, MA: MIT Press.
Raleigh, T. (forthcoming). Phenomenology without representation. European Journal of
Philosophy.
Russell, B. (1913). Theory of Knowledge. London: Routledge.
Schellenberg, S. (2008). The situation-dependency of perception. Journal of Philosophy, 105,
55–84.
Schellenberg, S. (2010). The particularity and phenomenology of perceptual experience.
Philosophical Studies, 149, 19–48.
Schellenberg, S. (2011). Perceptual content defended. Noûs, 45, 714–750.
Schellenberg, S. (2013). Experience and Evidence. Mind, 487, 699–747.
Siegel, S. (2010). Do experiences have content? In B. Nanay (Ed.), Perceiving the World (pp.
333–368). Oxford: Oxford University Press.
Smith, B. (2007). The objectivity of taste and tasting. In B. Smith (Ed.), Questions of
Taste: The Philosophy of Wine (pp. 41–77). Oxford: Oxford University Press.
Speaks, J. (2005). Is there a problem about nonconceptual content? Philosophical Review,
114, 359–398.
Speaks, J. (2006). Is mental content prior to linguistic meaning? Noûs, 40, 428–467.
Travis, C. (2004). Silence of the senses. Mind, 113, 57–94.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.
Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.
Williamson, T. (2006). Can cognition be factorized into internal and external components?
In R. Stainton (Ed.), Contemporary Debates in Cognitive Science (pp. 291–306). Oxford:
Blackwell.
Wu, W. (forthcoming). Against division: Consciousness, information, and visual stream.
Mind and Language.
 9

Experiential Content and Naïve Realism

A RECONCILIATION
Heather Logue

I’m currently having an experience of a yellow crescent-shaped banana on my

desk. As a result, I come to believe that there is a yellow crescent-shaped ba-
nana before me. The proposition that there is a yellow crescent-shaped banana
is the content of my belief. Does my experience have content too? An affirma-
tive answer to this question opens a rather large can of worms. To identify just
a few: Is the content of an experience a proposition, like the content of a belief?
If so, which proposition is it, exactly (e.g., the proposition that there is a yellow
crescent-shaped thing before me)? And what’s the nature of this proposition (e.g.,
is it Fregean)?
These questions have been subjects of debate for quite some time, and these
debates were enabled by the practically universal assumption that experiences
have content. Recently, however, this assumption has come under fire (see, e.g.,
Travis 2004; Brewer 2006; Johnston 2006)—mainly from those who are attracted
to a view called Naïve Realism. A rough statement of the view is that certain expe-
riences (namely, those in which one perceives one’s environment as it is) funda-
mentally consist in perceiving things in one’s environment. In other words, their
most basic psychological nature is given in the description just used to pick them
out. For example, Naïve Realism holds that my experience of the banana on my
desk fundamentally consists in my perceiving the banana.
Naïve Realists tend to hold that their view is incompatible with the claim that
experiences have content. I think this is incorrect. My view on this matter isn’t
novel; some have argued that there is a relatively weak interpretation of the claim
that experience has content that Naïve Realists can and should accept (see Siegel
2010 and Schellenberg 2011). But I differ with previous “compatibilists” on two is-
sues. First, pace Siegel and Schellenberg, I think there is an argument for the claim
that experience has content in the weak sense that is more effective than the ones

Thanks to Adam Pautz, as well as audiences at York, Edinburgh, and Nottingham for helpful
220 questions and comments.
 Experiential Content and Naïve Realism   221

hitherto offered. Second, pace Siegel, I think that Naïve Realism is compatible with
much stronger interpretations of the claim that experience has content.
In the first section of this chapter, after briefly arguing for the assumption that
experiential content is propositional, I’ll distinguish three interpretations of the
claim that experience has content (the Mild, Medium, and Spicy Content Views). In
the second section, I’ll flesh out Naïve Realism in greater detail, and I’ll reconstruct
what I take to be the main argument for its incompatibility with the Content Views.
The third section will be devoted to evaluation of existing arguments for the Mild
Content View (the arguments from accuracy and appearing), and the development
of what I take to be a stronger argument (the argument from belief generation). In
the final section, I’ll identify a flaw in the argument for the incompatibility of Naïve
Realism and the Content Views, which opens the door to a reconciliation.

9.1. What Does It Mean to Say That Experience Has Content?

Before we attempt to reconcile experiential content with Naïve Realism, we must

first clarify what it means to say that perceptual experience has content in the first
place. In my view, part of what it means is that there is a proposition associated
with the experience—i.e., that the content of an experience is a proposition. This
is a controversial claim, as some philosophers who maintain that experience has
content claim that it’s non-propositional.
If the content of a perceptual experience isn’t a proposition, then what is it?
One possibility is that it’s an object of some sort (e.g., a banana, or a sense-datum),
or a state of affairs (e.g., a banana’s being yellow, or a sense-datum’s being yel-
low′). But this can’t be what is meant—after all, everybody thinks that experiences
that involve perceiving things have content in this sense, and practically no one
(besides sense-datum theorists) thinks that hallucinations have content in this
sense. So this understanding of non-propositional content is one part trivially true
and one part plausibly false.
Another candidate put forward for non-propositional experiential content is
the way the subject perceptually represents her environment as being (see Crane 2009,
p. 456). However, on a relatively uncontroversial understanding of what a proposi-
tion is, it’s just a way the world might be—e.g., the sort of thing that can be true or
false, expressed by a sentence, and (most importantly for our purposes) perceptually
represented by a subject. One way the world might be is for it to be the case that there
is a yellow crescent-shaped thing before me. That there is a yellow crescent-shaped
thing before me is a proposition, and something I can perceptually represent. So we
don’t really have a candidate for non-propositional perceptual content after all.
Why the insistence to the contrary? Plausibly, one idea in the background is
that since experience is very different from belief, the content of experience (what-
ever it is) must be rather different from the content of belief. But, of course, this idea
can be accommodated without denying that experiences have propositional content
 Reconciliatory Views
222   

(e.g., by claiming that experiential contents are typically much more specific or
finely-grained than the contents of beliefs).1 Another possibility is that the advocates
of non-propositional perceptual content have a particularly demanding conception of
what a proposition is in mind (as suggested in Byrne 2001, p. 201)—e.g., that they are
composed of Fregean senses. However, it is important to distinguish two questions:
(1) Do experiences have propositional content?
(2) If experiences have propositional content, what is the metaphysical struc-
ture of those propositions (e.g., Fregean, Russellian . . . )?
We must be careful not to simply assume an answer to (2) that supports a negative
answer to (1). One can show that experiences don’t have propositional content on
the basis of a particular answer to (2) only if one has an argument for giving that
answer rather than another—otherwise, it’s epistemically possible that experiences
have propositional content given a different theory of the metaphysical structure
of the relevant propositions.
Yet another potential source of resistance to identifying the way the subject of an
experience represents her environment as being with a proposition is the idea that an
experiencing subject represents things in the world, not propositions (cf. Crane 2009,
pp. 464–465). For example, one might insist that in having an experience of a yellow
crescent-shaped banana on my desk, I’m perceptually representing a banana (as well
as its color, shape, and location), not a proposition. However, this is a false contrast.
It’s not clear that there’s any daylight between, say, representing something as being
yellow crescent-shaped, and to my left, on the one hand, and representing the prop-
osition that something is yellow crescent-shaped, and to my left, on the other. Given
that a proposition is basically a way for the world to be, representing things in the
world as being certain ways is tantamount to representing a proposition.
I don’t take this brief case for skepticism about non-propositional experiential
content to be decisive, but unfortunately I don’t have the space to elaborate and
defend it. In any case, if you think there is such a thing as non-propositional expe-
riential content, then you should think of my aim as being to establish that Naïve
Realism is compatible with propositional experiential content.
Supposing that the content of experience is a proposition, which proposition
is the content of a given experience—say, my experience as of a yellow crescent-
shaped banana? As I see it, to say that an experience has content is to say at least
the following:
for any perceptual experience E,
(i) there is a proposition associated with E, and
(ii) this proposition captures the way things perceptually appear to the subject
in virtue of having E.

1
Thanks to Matt Nudds for pressing me to mention this potential motivation for denying
experiential content.
 Experiential Content and Naïve Realism   223

For example, the content of my experience is (something along the lines of) the
proposition that there is a yellow crescent-shaped thing before me. These two
claims, (i) and (ii), are the bare minimum that one is committed to when one
says that experience has content. Let us call these two claims the Mild Content
View.2
Note that the Mild Content View is silent on whether experience is a prop-
ositional attitude—one could hold that there is a proposition associated with an
experience, but deny that the experience consists in the subject perceptually rep-
resenting it. On this sort of position, although the proposition figures in the char-
acterization of the experience from the theorist’s point of view, the subject of the
experience doesn’t bear any distinctively experiential psychological relation to it.
Alternatively, one could endorse the following claim:
(iii) perceptual experience consists in the subject perceptually representing
her environment as being a certain way.
Let us call claims (i)–(iii) the Medium Content View.
Note that the Medium Content View is silent on whether perceptual ex-
perience fundamentally consists in the subject perceptually representing her
environment as being a certain way—one could hold that experience is a prop-
ositional attitude, but not fundamentally so. To say that perceptual experience
fundamentally consists in personal-level psychological feature x is to say that
it has some or all of its other personal-level psychological features ultimately
in virtue of x. So, for example, one might hold that my experience of a yellow
crescent-shaped banana involves my perceptually representing my environment
as containing a yellow crescent-shaped thing, but that this fact isn’t the ultimate
personal-level psychological explanation of the why this experience naturally
generates the belief that there’s a yellow crescent-shaped thing before me, or
the phenomenal character associated with this experience, or any of the other
psychological features we’re trying to give an account of when we’re giving a

2
The Mild Content View adopts what Adam Pautz calls the appears-looks conception of
experiential content, and he argues that this conception trivializes the debate over whether experience
has content, as well as other related debates (2009, pp. 485–486). I won’t go into Pautz’s arguments here;
the crucial point is that (as he recognizes) one could try to avoid them by insisting on a distinction
between perceptual and epistemic appearances (more on this distinction in section 9.3). He seems to
think that this insistence must amount to holding that locutions of the form ‘it appears to S that p’
always pick out epistemic appearances, while perceptual appearances are only picked out by locutions
of the form ‘o appears F to S’. If this is right, then the appears-looks conception would still trivialize
the relevant debates (e.g., hallucinations would trivially lack perceptual content). However, one could
hold that locutions of the form ‘it appears to S that p’ are potentially ambiguous, sometimes picking
out epistemic appearances and sometimes picking out perceptual appearances. If that’s right, then the
debate is not trivial—as I will argue in section 9.3, it boils down to a debate over whether there is a
kind of perceptual appearance that ‘it appears to S that p’ is used to pick out. (Thanks to Adam Pautz
for pressing me to clarify this point.)
 Reconciliatory Views
224   

philosophical theory of perceptual experience. Alternatively, one could endorse

the following claim:
(iv) perceptual experience fundamentally consists in the subject perceptually
representing her environment as being a certain way.
Claim (iv) entails that at least some psychological feature of perceptual experi-
ence (e.g., its phenomenal character, or some aspect of its epistemological role)
is ultimately grounded in the subject perceptually representing her environ-
ment as being a certain way. This view composed of claims (i)–(iv) is known as
Intentionalism about perceptual experience, but for consistency’s sake, let’s call it
the Spicy Content View.3
If perceptual experience has content in any of these three senses, there are a
number of matters arising. To name just a few: first, as I hinted in the discussion
of the notion of non-propositional experiential content, we must determine the
nature of the propositions that can be experiential contents—e.g., whether they are
Fregean-style propositions composed of senses, Russellian-style ordered pairs of
objects and relations, or perhaps coarse-grained sets of possible worlds.4 Second,
we must determine the relationship between the content of an experience and its
phenomenal character. Does the phenomenal character of an experience super-
vene on its content? And does the content of an experience supervene on its phe-
nomenology?5 Third, we must determine which sorts of properties can figure in
experiential contents. For example, do visual experiences have contents that take
a stand only on the presence of properties like color, shape, and location, or can they

3
Here’s how the taxonomy of views I’ve just offered relates to some of the others in the
literature: Susanna Siegel (2010) distinguishes between the Content View and the Strong Content
View; the latter is more-or-less my Spicy Content View, and the former is essentially my Mild
Content View. (Siegel doesn’t single out what I’ve called ‘the Medium Content View’.) Pautz (2009)
distinguishes between appears-looks, accuracy, and identity conceptions of experiential contents;
the first is more-or-less my Mild Content View, and the last is basically my Spicy Content View. The
accuracy conception is an alternative—and in my view, inferior—way of formulating a Mild Content
View. (I won’t defend this claim here, although the reader might be able to discern my reservations
about the accuracy conception on the basis of the discussion of the argument from accuracy in section
9.3.) Finally, Susanna Schellenberg (2011, pp. 15–16) distinguishes between the association thesis and
the representation thesis. The former is basically my Mild Content View, while my Medium and Spicy
Content Views are different ways of spelling out Schellenberg’s representation thesis (both of which
are different from Schellenberg’s preferred way of spelling it out, which construes representation in
terms of employing concepts—my Medium and Spicy Content Views make no such commitment).
Schellenberg also identifies a view she calls the awareness thesis: basically, a view on which the content
of experience is a Russellian proposition, and the subject is aware of it in the sense that she literally
perceives its constituents (presumably, this is the only way to make sense of the idea that one could
literally perceive the propositional content of an experience). Since I will remain neutral in this chapter
on the nature of the propositions that are potential contents of experience, this view is beyond the
scope of this chapter.
4
See, e.g., Tye 2000 (Russellian content), Burge 1991 (Fregean content), Stalnaker 1984 (possible
worlds content), and Chalmers 2006 (a pluralistic view).
5
For negative answers to these questions, see (e.g.) Block 1990 and 1996; for affirmative answers,
see (e.g.) Tye 2000.
 Experiential Content and Naïve Realism   225

take a stand on the presence of so-called “high-level” properties, such as natural kind
properties (e.g., being a banana)?6 I’m not going delve into any of these thorny issues
here, since doing so isn’t required in order to establish the main theses of this chapter;
namely, that experience has content in the weak sense, and that all three Content
Views are compatible with Naïve Realism.

9.2. Naïve Realism and the Content Views

According to Naïve Realism, some perceptual experiences fundamentally consist

in the subject perceiving entities in her environment.7 For example, according to
the Naïve Realist, the experience I’m having of the banana on my desk fundamen-
tally consists in my perceiving the banana. On this view, the ultimate psycholog-
ical explanation of at least some of the features of my experience is in terms of my
bearing the perceptual relation to the banana. Note that Naïve Realism is structur-
ally similar to a version of the sense-datum theory: both views hold that at least
some perceptual experiences fundamentally consist in the subject bearing the per-
ceptual relation to something. They just give a different account of what that some-
thing is. According to the sense-datum theory, it’s a mind-dependent, immaterial
sense-datum; while according to the Naïve Realist, it’s a mind-independent, mate-
rial object in the subject’s environment.
Another difference between Naïve Realism and the version of the sense-datum
theory just sketched is that the latter is a theory about all perceptual experiences. By
contrast, Naïve Realism is a claim about only some perceptual experiences. Which
ones? Well, it’s certainly not a claim about hallucinatory experiences. By definition,
such experiences don’t involve the subject perceiving things in her environment;
a fortiori, they can’t fundamentally consist in the subject perceiving things in her
environment.
What about the experiences that do involve the subject perceiving things in her
environment? This class of experiences divides into two types: veridical experiences
and illusions. Veridical experiences are those in which the subject perceives a thing o,
and it appears to have a property F in virtue of the subject perceiving o’s F-ness. For
example, when I’m having a veridical experience of the banana on my desk, it looks
yellow to me in virtue of my perceiving the banana’s yellowness.8 By contrast, illusions

6
For a case for the claim that natural kind properties (and other high-level properties) can figure
in the content of experience, see Siegel 2006; for objections, see (e.g.) Price 2009 and Logue 2013b.
7
This characterization of Naïve Realism is superficially different from some others found in the
literature (e.g., Fish 2009). See Logue (2013a) for discussion of how the characterization in the main
text captures the content of other typical formulations of the view.
8
One might suggest that the very notion of a veridical experience smuggles in an affirmative
answer to the question of whether experience has content, and so isn’t a notion I’m entitled to in this
dialectical context. The line of thought is this: A veridical experience is one in which things in the
subject’s environment appear to her to be a certain way, and they are that way (as the subject perceives
 Reconciliatory Views
226   

are experiences in which the subject perceives a thing o, and it appears to have a
property F even though the subject doesn’t perceive o’s F-ness. For example, consider
an illusion in which a green banana looks yellow as a result of unusual lighting condi-
tions. The subject of this illusion sees the banana, and it looks yellow to her, but not
in virtue of perceiving the banana’s yellowness—indeed, the banana isn’t yellow, so it
doesn’t instantiate any yellowness for her to perceive.
All Naïve Realists take their theory to apply to veridical experiences. However,
they divide when it comes to illusions. Some Naïve Realists restrict their theory
to veridical experience (e.g., Martin 2006). Others think that the theory can be
extended to illusions: one way to do this (very roughly) is to claim that illusion
fundamentally consists in the subject perceiving things and some of their proper-
ties, and that something’s illusorily appearing F consists in perceiving a property
distinct from F-ness that the thing does instantiate.9 The details of such an account
of illusion are beyond the scope of this chapter. For simplicity’s sake, I will restrict
my focus to Naïve Realism about veridical experience.
Why might a Naïve Realist be hostile to the claim that experience has con-
tent? Naïve Realism holds that veridical experience fundamentally consists in the
subject perceiving things in her environment; but on the face of it, this is perfectly
compatible with it having content as well. Just because Naïve Realists don’t typically
characterize veridical experiences as having experiential content, it doesn’t follow
that they couldn’t. Why would the Naïve Realist insist that veridical experience
consists in nothing more than the subject perceiving things in her environment?
I suspect that much of the Naïve Realist resistance to experiential content is rooted
in an argument of M. G. F Martin’s regarding a distinct but related issue—namely,
what a Naïve Realist should say about hallucinations. Let’s sketch this argument,
and explore its implications for the question of whether experience has content.
In his paper “The Limits of Self-Awareness” (2004), Martin argues that certain
accounts of hallucination don’t go well with Naïve Realism—in particular, what he
calls positive accounts. A positive account of hallucination is one that characterizes
it in terms that are independent of veridical experience. One example is the claim
that hallucination fundamentally consists in the subject perceptually representing
her environment as being a certain way (i.e., the Spicy Content View restricted to
hallucination). Note that this claim makes no reference at all to veridical experi-
ence. By contrast, a negative account of hallucination characterizes it in terms of a

the properties that appear to her to be instantiated by things in her environment). But, given what I’ve
said in the previous section, doesn’t the claim that things appear to her to be a certain way amount
to saying that her experience has content (the proposition that specifies how the things appear to her
to be)? In a word, no. As I’ll argue in the following section, the talk of appearance that figures in the
characterization of veridical experience can be understood in a distinctively perceptual sense or in an
epistemic sense. Only the former entails a commitment to experiential content, so one who denies
experiential content can make sense of the notion of veridical experience in terms of the latter.
9
For views roughly along these lines, see Brewer 2008, Fish 2009, Antony 2011, and Kalderon
2011.
 Experiential Content and Naïve Realism   227

relation it bears to veridical experience. For example, Martin’s preferred account

of hallucination is that it fundamentally consists in nothing more than the subject
being in a state she can’t tell apart by reflection alone from a veridical experience
of a certain kind (e.g., a veridical experience of a yellow crescent-shaped banana).
This account characterizes hallucination as simply seeming like something it’s
not—there’s nothing to hallucination beyond this relation to veridical experience.
Martin’s argument against combining Naïve Realism with a positive account
of hallucination has two stages. First, he argues that any feature of a hallucination
will be also had by veridical experience that has the same proximate cause (2004,
pp. 52–58). Second, he argues that such a commonality across a veridical experi-
ence and a hallucination will render Naïve Realism largely explanatorily redun-
dant (2004, pp. 58–68).10
I have discussed Martin’s argument against positive accounts of hallucination
at length elsewhere (Logue 2013a). However, a close cousin of the second part of
his argument could be employed in the debate over whether experience has con-
tent, and so deserves investigation in this context. The argument is as follows:

1. Suppose (for the sake of reductio) that veridical experiences have contents
(in the Mild sense that there are propositions that specify the way things
perceptually appear to the subject).
2. Naïve Realism is not explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (a
commitment of Naïve Realism).
3. The fact that a veridical experience has a given content is sufficient to
explain its epistemological, behavioral, and phenomenal features.
4. If veridical experiences have contents, Naïve Realism is explanatorily
redundant with respect to a veridical experience’s epistemological,
behavioral, and phenomenal features (from 3).
5. Naïve Realism is explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (from 1
and 4)
6. Contradiction (2 and 5); veridical experiences don’t have contents.

Of course, one could just as well re-frame this argument as a reductio of Naïve
Realism (by supposing premise 2 for the sake of reductio instead). But since our
issue is whether Naïve Realism is compatible with the Mild Content View, I will
grant premise 2 for the sake of argument. That leaves premises 3 and 4—what can
be said for them?

10
Strictly speaking, it’s not Naïve Realism that’s incompatible with a positive account of
hallucination—in principle, one could endorse Naïve Realism along with a positive account of
hallucination, and concede that Naïve Realism plays a very limited role in explaining the phenomena
that philosophical theories of experience are supposed to explain (e.g., the epistemological and
phenomenal aspects of experience). But it’s not clear why anyone would want to do that.
 Reconciliatory Views
228   

Premise 3 says that the fact that a veridical experience has a certain content
(e.g., that there is a yellow crescent-shaped thing before me) is sufficient to explain
the experience’s epistemological features (e.g., its tendency to generate the belief
that there’s a yellow crescent-shaped thing before me), its behavioral features (its
tendency to cause me to reach for the banana if I want to eat one), and its phenom-
enal features (i.e., what it’s like for me to see a yellow crescent-shaped thing). This
last plank of the premise is particularly controversial—it will be rejected by those
who hold that the phenomenal character of an experience doesn’t supervene on
its content. A popular reason for this rejection is the combination of externalism
about perceptual content with the alleged possibility of spectrum inversion. Given
externalism about perceptual content, one’s experience represents that something
is yellow in virtue of being a type of experience that is typically caused by yellow
things. Given the possibility of spectrum inversion, what it is like for one subject
to experience yellow things could be what it’s like for another subject to experi-
ence purple things. Putting the two together, two subjects could have experiences
with the content that there’s something yellow before them (because they’re hav-
ing experiences of the sort typically caused by yellow things), but what it’s like for
the subjects to have the experiences differs (because they are spectrally inverted
with respect to each other). If this is a genuine possibility, the phenomenal char-
acter of an experience doesn’t supervene on its representational content, and an
experience’s content isn’t sufficient to explain why it has the phenomenal character
it does—something more has to be said. For example, one might think we have
to appeal to intrinsically non-intentional qualia to fully capture experiential phe-
nomenology (Block 1996).
However, Naïve Realists are no friends of this kind of qualia, at least when it
comes to veridical experience. They typically hold that the phenomenal character
of a veridical experience is determined by the way the subject’s environment is—
that its phenomenal character is “constituted by the actual layout of the [environ-
ment] itself: which particular objects are there, their intrinsic properties, such as
color and shape, and how they are arranged in relation to one another and to [the
subject]” (Campbell 2002, p. 116). In short, standard Naïve Realism simply leaves
no room for phenomena like spectrum inversion that motivate the rejection of
the claim that phenomenology supervenes on content.11 Thus, most Naïve Realists
would be hostile to a rejection of premise 3 on the grounds just outlined. Again,
since our issue is whether Naïve Realism is compatible with the Mild Content
View, I will set aside this kind of reason for rejecting premise 3, and assume along
with most Naïve Realists that spectrum inversion isn’t a genuine possibility.
One might worry that no Naïve Realist would endorse premise 3. For such
an endorsement seems tantamount to admitting that there’s no motivation for
Naïve Realism—if experiential content really is sufficient to explain the episte-
mological, behavioral, and phenomenal features of experience, then we might as

11
For a version of Naïve Realism that does leave room for such phenomena, see Logue 2012b.
 Experiential Content and Naïve Realism   229

well all embrace the Spicy Content View and call it a day.12 However, this worry
doesn’t withstand scrutiny. First, not all motivations for Naïve Realism claim that
it can explain a feature of veridical experience that the Spicy Content View cannot.
For example, Martin argues that Naïve Realism is required to account for sensory
imagination—he says that Naïve Realism and the Spicy Content View are on a par
with respect to accounting for perceptual experience (2002, p. 402). Premise 3 is
perfectly compatible with a motivation for Naïve Realism of this sort. Second, the
Naïve Realists who do think that there’s some feature of experience that only Naïve
Realism can explain don’t seem to have realized that a claim along the lines of
premise 3 is incompatible with their motivations for Naïve Realism. For example,
Bill Fish (2009) argues that only Naïve Realism can properly account for percep-
tual phenomenal character, but nevertheless wrongly takes his view to be subject
to Martin’s screening off argument. In short, some Naïve Realists can and do accept
premise 3, and at least some of those who shouldn’t don’t seem to have realized that
they shouldn’t. So the argument presented above is still a plausible reconstruction
of the Naïve Realist argument against experiential content.
So much for premise 3 (for the time being). What about premise 4? This
premise is a plausible consequence of premise 3. If the fact that a veridical experi-
ence has a given content is sufficient to explain its epistemological, behavioral, and
phenomenal features, then Naïve Realism drops out as redundant with respect to
accounting for veridical experience—if we can explain everything about the expe-
rience that a philosophical theory of perceptual experience is supposed to explain
just by saying that it has a certain content, then the Naïve Realist claim that it also
involves the obtaining of the perceptual relation between the subject and things in
her environment needlessly complicates our account of veridical experience. Of
course, it’s open to the Naïve Realist to insist that we should jettison the notion of
experiential content from our account instead.13 But the point is that it’s one or the
other—the marriage of experiential content and Naïve Realism appears to be an
unhappy one, since each steals the other’s explanatory thunder.
As I said before, I take it that something like the argument just presented
underlies much of the Naïve Realist hostility to experiential content.14 And on the

12
Thanks to Dave Ward for raising this important issue.
13
This is a disanalogy with Martin’s argument against positive disjunctivism. In that case, there is
a reason to prefer explanations in terms of experiential content: such explanations can provide a unified
account of the epistemological, behavioral, and phenomenal features shared by a veridical experience
and a hallucination. By contrast, a Naïve Realist explanation doesn’t apply in the case of hallucination.
Given that we should give a unified explanation of a phenomenon whenever possible, once we let
perceptual content into our account of perceptual experience, it affords the superior explanation of
the relevant phenomena. (In order to protect the explanatory power of Naïve Realism, Martin eschews
accounting for hallucination in terms of any positive experiential features like perceptual content—see
Martin 2004, pp. 71–72.)
14
This isn’t the only source of resistance—e.g., some have expressed doubts about whether we can
non-arbitrarily pin down the content of an experience (see Travis 2004 and Brewer 2006). Since my
main aim is to reconcile the Content Views with Naïve Realism, rather than to defend the former from
objections, I’ll set these doubts aside.
 Reconciliatory Views
230   

face of it, the argument is plausible—if you’re persuaded by Martin’s argument

against positive disjunctivism (as many are), you’re likely to be persuaded by this
argument against experiential content. Nevertheless, I don’t think that this argu-
ment is sound. But before criticizing it, let us turn our attention to arguments in
favor of experiential content. We can fully appreciate the need for a reconciliation
of Naïve Realism and the Mild Content View only once we recognize that simply
denying the latter isn’t a viable option.

9.3. Arguments for the Mild Content View

In this section, I will outline two arguments for the Mild Content View. I will argue
that, although they are sound, they aren’t likely to persuade Naïve Realists who are
convinced that their view is incompatible with the Mild Content View. So I will
offer a different argument for the Mild Content View that I take to be more dia-
lectically effective.
One argument for the Mild Content View is the argument from accuracy
(Siegel 2010, pp. 337–343; 2011, pp. 33–42). It runs as follows:

1. Intuitively, experiences are assessable for accuracy (e.g., an experience

as of a yellow crescent-shaped thing had in the presence of yellow
crescent-shaped thing is accurate, whereas such an experience had in the
absence of any yellow crescent-shaped thing whatsoever is inaccurate).
2. Hence, there are conditions under which an experience is accurate (e.g.,
an experience as of a yellow crescent-shaped thing is accurate only if there
is a yellow crescent-shaped thing before one).
3. The conditions under which an experience is accurate specify a
proposition (e.g., that there is a yellow crescent-shaped thing before one).
4. The proposition specified by an experience’s accuracy conditions captures
the way things perceptually appear to the subject in virtue of having the
experience.
5. Hence, for any given experience E, there is a proposition associated with E
that captures the way things perceptually appear to the subject in virtue of
having E (i.e., the Mild Content View).

Siegel doesn’t rest her case for the Mild Content View on the argument from
accuracy, as she’s skeptical of premise 4. She notes that there’s no guarantee that an
experience’s accuracy conditions are conveyed to the subject, i.e., that the accuracy
conditions capture how things perceptually appear to her. As Siegel points out,
not all of the conditions under which an experience is accurate specify how things
perceptually appear to the subject. For example, take the trivial accuracy condition
of the experience’s being accurate—that one’s experience is accurate isn’t among
the ways things can perceptually appear to be (Siegel 2010, p. 344). Nevertheless,
given the claim that things perceptually appear to the subject of an experience to
 Experiential Content and Naïve Realism   231

be a certain way, and that the experience has accuracy conditions, it’s natural to
identify the way things perceptually appear to the subject with at least some subset
of its accuracy conditions. That is, once we’ve recognized that experiences have
accuracy conditions, the burden is on the opponent of the Mild Content View to
explain why the way things perceptually appear to the subject isn’t identical to a
proposition specified by some subset of those conditions.15
Once the premises are tweaked so that the content of experience is specified
by a subset of an experience’s accuracy conditions, I believe the resulting argu-
ment is sound. But I don’t think it’s dialectically effective. Consider the situation
from the perspective of a Naïve Realist who’s convinced that her view is incom-
patible with the Mild Content View. The argument starts off with an appeal to
an alleged intuition—viz., that experiences are assessable for accuracy. Our Naïve
Realist thinks that endorsing this intuition is tantamount to giving up on her view,
so she’s well-advised to scrutinize this intuition carefully. And although I’m sym-
pathetic to the intuition, it’s not obvious that it’s worth hanging onto at any cost. If
I took myself to I have excellent reasons for endorsing Naïve Realism, and excel-
lent reasons to believe that Naïve Realism is incompatible with the Mild Content
View, I’d be willing to argue in the opposite direction from the falsity of the Mild
Content View to the falsity of the intuition. Philosophical reflection sometimes
recommends forsaking intuitions in favor of an error theory, and for our Naïve
Realist, this could be one of those cases. So, given the present state of the dialectic
(in which Naïve Realism is often taken by its proponents to be incompatible with
the Mild Content View), the argument from accuracy isn’t persuasive.
The route to the Mild Content View that Siegel does endorse is the argument
from appearing (Siegel 2010, pp. 345–354; see also Schellenberg 2011, pp. 718–720).
The argument goes roughly as follows:

1. In having an experience, things perceptually appear to the subject to be

some way (e.g., it perceptually appears to the subject of an experience of
a yellow crescent-shaped banana that there is a yellow crescent-shaped
thing before her).
2. The way things perceptually appear to a subject to be specifies a
proposition (e.g., that there is a yellow crescent-shaped thing before one).
3. Hence, for any given experience E, there is a proposition associated with E
that captures the way things perceptually appear to the subject in virtue of
having E (i.e., the Mild Content View).16

In short, things perceptually appear to the subject of an experience to be some way,

and the proposition that things are that way is the content of the experience (in the

15
I suspect Siegel would agree; it seems that her further argument from appearing (discussed
below) is intended to provide a way of specifying which subset of an experience’s accuracy conditions
yields its content.
16
Siegel and Schellenberg present much more detailed variants of this argument, but the details
wouldn’t affect the point I want to make here.
 Reconciliatory Views
232   

Mild sense). Prima facie, this argument is much more promising than the argu-
ment from accuracy. First, there’s no room to argue that the proposition specified
doesn’t capture the way things perceptually appear to the subject, since that’s what
specifies the proposition in the first place. Second, the starting point is much more
difficult to reject—one might be willing to reject the intuition that experiences
have accuracy conditions, but how can one deny that things perceptually appear
to be some way to one when one has an experience? Isn’t that a conceptual truth
if there ever was one?
Just as with the argument from accuracy, I think that the argument from
appearing is sound—it’s just not dialectically effective. Again, let’s think of the sit-
uation from the perspective of a Naïve Realist who takes her view to be incompat-
ible with the Mild Content View. If she’s right, the argument from appearing entails
the falsity of her view, so she’s well-advised to scrutinize its starting point. And (as
Siegel recognizes) there is some wiggle room here. The Naïve Realist might agree
that things appear to be some way to one when one has an experience, but only in
a sense of ‘appear’ that doesn’t entail the truth of premise 1.
It’s uncontroversial that things appear to the subject of an experience to be
some way in the sense that, in normal circumstances, the experience generates
the belief that things are that way. For example, when I have an experience of a
yellow crescent-shaped banana on my desk, there appears to me to be a yellow
crescent-shaped thing before me in the sense that I will form the belief that there
is a yellow crescent-shaped thing before me—at least as long as I’m rational, and
I don’t suspect that my experience has been generated in a non-standard way (such
as being the result of a hallucinogen). Such an appearance is an epistemic one. Let’s
say that it epistemically appears to a subject S that p just in case S is disposed to
believe that p solely on the basis of evidence (e.g., perceptual evidence), given that
S is rational and doesn’t suspect that the evidence is misleading. For example, it
epistemically appears to a subject of the Müller-Lyer illusion that the lines she’s
seeing are different lengths. Although a subject who’s in the know isn’t disposed
to believe that the lines are different lengths, she would be disposed to believe this
solely on the basis of her perceptual evidence if she didn’t know that her experi-
ence was misleading.
Now, the fact that things epistemically appear to the subject of an experience
to be some way doesn’t obviously entail that there is some other sense in which
things appear to the subject to be some way, associated with experiences instead
of beliefs—i.e., a distinctively perceptual appearance (see, e.g., Travis 2004; Brewer
2008 for claims along these lines). This is particularly clear on Naïve Realism, on
which veridical experience is fundamentally a relation to objects in one’s environ-
ment, unlike belief, which is fundamentally a relation to a proposition. According
to a Naïve Realist, the subject of a veridical experience perceives entities in her
environment (e.g., a yellow crescent-shaped banana), and this disposes her
to believe that her environment is a certain way (e.g., that there’s a yellow
crescent-shaped thing before her). And although there’s an epistemic appearance,
 Experiential Content and Naïve Realism   233

in that the subject is disposed to believe on the basis of perceptual evidence that
there’s a yellow crescent-shaped thing before her, the Naïve Realist might insist
that that there’s no distinctively perceptual appearance in addition—that there is a
proposition that captures the way things perceptually, as opposed to epistemically,
appear to the subject. In short, the idea is that things appear to be some way to the
subject in virtue of having the experience, but the proposition that things are that
way is the content of perceptually-based belief, not experience proper—veridical
experience is a relation to concrete things in one’s environment, not abstract prop-
ositional contents.
Given that the claim that it perceptually appears to S that o is F is equivalent
to the claim that o perceptually appears F to S, to deny the former is to deny the
latter.17 So the Naïve Realist maneuver sketched in the previous paragraph would
be tantamount to denying that things perceptually appear to have properties—e.g.,
that the banana on my desk perceptually appears yellow. Furthermore, one might
think that a necessary condition of S perceiving o’s F-ness is that o perceptually
appears F to S—I wouldn’t be perceiving the yellowness of the banana on my desk
if it didn’t look (i.e., visually appear) yellow to me. If that’s right, the Naïve Realist
maneuver just outlined entails that the subject of a veridical experience simply
perceives things in her environment (e.g., bananas); strictly speaking, she doesn’t
perceive any the properties they instantiate (e.g., yellowness). The resulting view
is what Siegel calls ‘Radical Naïve Realism’: veridical experience “ . . . consists in a
perceptual relation to a worldly item, and properties are not among the things the
subject is perceptually related to” (2010, p. 358).18
Siegel argues that Radical Naïve Realism is implausible, mainly because our
theory of veridical experience should reflect which properties of a perceived object
make a difference to phenomenal character and which ones don’t (2010, p. 359).
For example, in having a veridical experience of the banana on my desk, the color
of its rind contributes to the phenomenal character of my experience, but the color
of the fruit within does not. Just saying that my experience fundamentally consists
in my perceiving the banana doesn’t capture this obvious phenomenological fact.
It seems that we need to specify which of the banana’s properties I perceive in order
to fully account for the phenomenal character of my experience, contra Radical
Naïve Realism.

17
One might think that o could perceptually appear F to S without it perceptually appearing to
S that o is F on the grounds that only the latter requires that S has the concept of F-ness. However, its
perceptually appearing to S that o is F requires that S has the concept of F-ness only if having experiences
of F-ness requires that S has the concept of F-ness. And if having experiences of F-ness requires that S
has the concept of F-ness, then so does o’s appearing F to S—which means that the claims don’t come
apart after all.
18
Siegel formulates Radical Naïve Realism as a thesis about all non-hallucinatory experiences,
not just veridical experiences. But since I’m concerned only with what the Naïve Realist says about
the latter sort of experience here, I’ve weakened Radical Naïve Realism accordingly. Also, note that if
the Naïve Realist gives up on the claim that we perceive properties, she’ll have to draw the distinction
between veridical and illusory experiences in a different way than I drew it above (a task I’ll leave to
the reader).
 Reconciliatory Views
234   

While I’m sympathetic to this objection, I can (dimly!) see a way out for the
Radical Naïve Realist. It’s uncontroversial that, when I have a veridical experience
of a banana, my visual system registers some of its features (e.g., the bright yellow-
ness of its rind) but not others (e.g., the yellowish-whiteness of the fruit within).
But it’s not obvious that this uncontroversial fact amounts to perception of some of
the banana’s properties, or to the banana perceptually appearing to be a certain way
to me. For example, one might think that there are subpersonal perceptual states
that carry information about a perceived object’s properties, but the personal-level
upshot of such information processing is a unified experience of an object, as
opposed to one that “carves it up” in terms of its properties (so to speak).19
I’m not confident that this line of thought, when fully spelled out, will vindi-
cate Radical Naïve Realism and the rejection of the argument from appearing it’s
supposed to enable—just as I’m not sure that the denial that experiences are assess-
able for accuracy is a defensible way out of the argument from accuracy. But one
thing I am sure of is that this debate has gotten more complicated than it needs to
be. The driving idea behind these arguments for experiential content is that there
are distinctively perceptual appearances—that we’re failing to capture something
about veridical experience if we confine appearances to the post-perceptual dox-
astic domain. The path of least resistance would be an argument for distinctively
perceptual appearances from a starting point that even the most radical Naïve
Realist would have to accept. Fortunately, I think there is such an argument—let’s
call it the argument from belief generation.
If there are distinctively perceptual appearances, then things perceptually ap-
pear to a subject of an experience to be a certain way (e.g., that there is a yellow
crescent-shaped thing before her), and the proposition that things are that way is
at least necessary for the accuracy of the experience. In other words, if there are
distinctively perceptual appearances, then premise 2 of the argument from accu-
racy and premise 1 of the argument from appearing are true. The argument for
belief generation is essentially a case for the antecedent.
The argument begins with the truism that a given experience naturally gives
rise to particular beliefs rather than others. For example, my experience of a yel-
low crescent-shaped banana naturally gives rise to the belief that there’s a yel-
low crescent-shaped thing before me, but not the belief that there is a purple,
star-shaped thing before me. More precisely, a given experience E is associated
with a particular epistemic appearance that p—the subject is disposed to believe
that p solely on the basis of E, given that she is rational and doesn’t suspect that
E is misleading. Since I am rational and have no suspicions that my experience
of a yellow crescent-shaped banana is misleading, I am disposed to believe (and
indeed do believe) that there is a yellow crescent-shaped thing before me solely on
the basis of my experience.

19
I suspect that something like this line of thought is behind Brewer’s Object View (see his 2008,
especially pp. 171–172).
 Experiential Content and Naïve Realism   235

At this point, one should wonder: what grounds the association between E
(say, my experience of a yellow crescent-shaped banana) and the epistemic appear-
ance that p (say, that there is a yellow crescent-shaped thing before me)? What does
my experience have to do with proposition that there is a yellow crescent-shaped
thing before me? How do we get from the former to the latter? A plausible answer
is that it perceptually appears to me that p—i.e., that the epistemic appearance asso-
ciated with E is just the proposition specified by how things perceptually appear to
me. And from here we can establish the Mild Content View. To summarize:

1. Any given experience E is associated with a particular epistemic

appearance that p.
2. The best explanation of (1) is that there is a proposition associated with E
that captures the way things perceptually appear to the subject in virtue
of having E.
3. Hence, for any given experience E, there is a proposition associated with E
that captures the way things perceptually appear to the subject in virtue of
having E (i.e., the Mild Content View).

I submit that this argument for the Mild Content View is more dialectically effective
than the arguments from accuracy and appearing. For while the Naïve Realist might
be willing to deny that experiences are assessable for accuracy, or that things per-
ceptually appear to a subject to be a certain way, it would be sheer madness to deny
that a given experience naturally gives rise to particular beliefs about one’s surround-
ings—e.g., to deny that a veridical experience of a yellow banana naturally gives
rise to the belief that there is a yellow crescent-shaped thing before one. (Indeed,
recall that the Naïve Realist had to appeal to epistemic appearances associated with
experiences in order to render the denial of perceptual appearances even remotely
plausible.)
So the argument’s starting point is non-negotiable. But what about its second
premise? There are two broad ways of resisting it: one could put forward an equally
good alternative explanation of the association between experiences and epistemic
appearances, or one could deny that this association requires psychological expla-
nation. As for the first option: the task is to explain, e.g., why the experience I’m
having now gives rise to the epistemic appearance that there is a yellow crescent-
shaped thing before me—in short, the relationship between this proposition and
my perceptual experience. I have no idea what this relationship would be if not
that of the former being the way things perceptually appear to me in virtue of hav-
ing the latter. Any other candidate that comes to mind is more controversial—e.g.,
the proposition being the way I perceptually represent things as being in virtue of
having the experience (as on the Medium Content View). So the proposed expla-
nation seems to be the least we can get away with.
The more promising option is to challenge whether we even need a psycho-
logical explanation of the connection in the first place. One might think that it’s
just a brute psychological fact that the experience I’m having right now gives rise
 Reconciliatory Views
236   

to the epistemic appearance that there’s a yellow crescent-shaped thing before

me—we don’t need to insert a perceptual appearance into the picture to explain
what’s going on here. An experience of a certain kind reliably produces a disposi-
tion to believe a particular proposition, and that’s all there is to it.
First, note that this response to premise (2) commits us to a broadly reliabilist
picture of how experiences justify beliefs. The connection between an experience
and an epistemic appearance is primarily causal, and only rational in virtue of
being a component of a reliable belief generating process. Many might find this
too bitter a pill to swallow to avoid the Mild Content View; but I won’t pursue this
worry here. A more fundamental problem with this response is that it is in tension
with the project of giving a philosophical theory of perceptual experience.
Presumably, in carrying out this project, our aim is to give an account of the
metaphysical structure of perceptual experience that explains certain of its fea-
tures—the features of most interest to philosophers being its epistemological role,
its phenomenal character, and its role in facilitating action. Now, if we can fully
explain these aspects of experience without an account of its metaphysical struc-
ture, then the whole point of this project goes out the window. It’s a short step
from the claim that we don’t need an explanation of the epistemological role of
perceptual experience in terms of its metaphysical structure to the claim that we
don’t need a philosophical theory of perceptual experience.20 And a consequence
of this would be that the debate over whether experience has content is pointless.
In short, one can avoid the Mild Content View by rejecting premise (2) on these
grounds, but the victory would be hollow—it would come at the cost of devaluing
the very debate one is engaged in.
Tim Crane suggests that “it is a mistake to read back from the content of a per-
ceptual judgement a hypothesis about the structure of experience on the basis of
which it is made” (Crane 2009, p. 465). This is essentially what the argument from
belief generation does—it moves from an epistemic appearance (a disposition to
judge that p solely on the basis of experience) to the claim that the associated expe-
rience has content in the Mild sense (there is a proposition that captures how things
perceptually appear to the subject). But far from being a mistake, I think this kind
of approach is the way forward. For how else are we supposed to figure out what the
structure of perceptual experience is, if not by looking to the roles experience plays
for constraints on that structure? Otherwise, it’s just not clear what is at stake in
arguments about the metaphysical structure of perceptual experience.
In summary, although the Naïve Realist has room to resist the arguments
from accuracy and appearing for the Mild Content View, it’s hard to see how the

20
To be fair, one might hold that although we don’t need an explanation of the epistemological
role of perceptual experience in terms of its metaphysical structure, we do need such an explanation
of its phenomenal character and/or its role in facilitating action. However, if one wants explanations
of the latter in terms of the metaphysical structure of perceptual experience, why wouldn’t one want
analogous explanations of the former? The burden is on the proponent of such a view to explain why
the epistemological role of experience differs from the other features in this respect.
 Experiential Content and Naïve Realism   237

Naïve Realist could escape the argument from belief generation with her credi-
bility intact. So if Naïve Realism is incompatible with the Mild Content View, then
so much the worse for Naïve Realism. Fortunately for the Naïve Realist, the ante-
cedent is false—establishing this is the first task of the next section.

9.4. Reconciling Naïve Realism and the Content Views

The conclusion of the argument from belief generation is that the Mild Content
View is true—that for any given experience E, there is a proposition associated
with it that specifies how things perceptually appear to the subject in virtue of
having E. In section 2, I outlined the following argument for the incompatibility of
Naïve Realism with the Mild Content View:

1. Suppose (for the sake of reductio) that veridical experiences have contents
(in the Mild sense that there are propositions that specify the way things
perceptually appear to the subject).
2. Naïve Realism is not explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (a
commitment of Naïve Realism).
3. The fact that a veridical experience has a given content is sufficient to
explain its epistemological, behavioral, and phenomenal features.
4. If veridical experiences have contents, Naïve Realism is explanatorily
redundant with respect to a veridical experience’s epistemological,
behavioral, and phenomenal features (from 3).
5. Naïve Realism is explanatorily redundant with respect to a veridical
experience’s epistemological, behavioral, and phenomenal features (from 1
and 4)
6. Contradiction (2 and 5); veridical experiences don’t have contents.

If Naïve Realism is to be reconciled with the Mild Content View, this argument
must be unsound. I submit that the culprit is premise 3.
To see this, consider the Mild Content View in isolation. All it says is that,
for any given experience, there’s a proposition associated with it that specifies how
things perceptually appear to the subject in virtue of having it. But to say only that
is to leave an important question unanswered: what makes it the case that that prop-
osition specifies how things perceptually appear to the subject? For example, the
proposition that there is a yellow crescent-shaped thing before me specifies how
things perceptually appear to me right now. How did this proposition end up
being cast in this role? Why does it perceptually appear to me that there is a yellow
crescent-shaped thing before me, as opposed to a red, round thing? Since the Mild
Content View doesn’t yield an answer to such questions, the fact that a veridical
experience has a given content in the Mild sense isn’t sufficient on its own to explain
 Reconciliatory Views
238   

its epistemological, behavioral, and phenomenal features. The Mild Content View
has to be supplemented with further claims that will yield an answer.
And Naïve Realism can come to the rescue. As I previously characterized it,
Naïve Realism is the view that veridical experience fundamentally consists in per-
ception of things in one’s environment. Some Naïve Realists (such as Johnston
2006 and Fish 2009) take this to amount to perception of facts or truthmakers.
Roughly, these are entities constituted by things and their properties, entities of
the form o’s being F (e.g., this banana’s being yellow, Mark’s being to the left of
Bill). As the second label suggests, such entities are what make propositions true.
This banana’s being yellow and before me makes it true that there is a yellow thing
before me; Mark’s being to the left of Bill makes it true that Mark is to the left of
Bill.
Fleshing out Naïve Realism in terms of perception of truthmakers affords a
handy explanation of why things perceptually appear to be the way they do in the
case of veridical experience. We can say that the proposition associated with an
experience that specifies how things perceptually appear to its subject is the one
such that the truthmakers perceived are necessary and sufficient for its truth. For ex-
ample, I perceive this banana’s being yellow crescent-shaped, and to my left, and this
truthmaker is necessary and sufficient for the truth of the proposition that there is
a yellow crescent-shaped thing to my left. It perceptually appears to me that there
is a yellow crescent-shaped thing to my left because the truthmaker I perceive in
the course of my current experience is necessary and sufficient for the truth of this
proposition. In short, the proposal is that the truthmakers the subject perceives
determine which proposition is the content of her experience (in the Mild sense).21
So we should reject premise (3) of the argument for the incompatibility of
Naïve Realism and the Mild Content View. The latter is insufficient to account for
the epistemological, behavioral, and phenomenal features of a veridical experi-
ence, because it is silent on the facts in virtue of which things perceptually appear
to a veridically perceiving subject as they do. Furthermore, Naïve Realism has a
story to tell here—the Naïve Realist can offer an account of the content of a verid-
ical experience in terms of the truthmakers the subject perceives in the course of
having it.

21
The “necessary and sufficient” restriction is required to avoid unwelcome consequences like the
following (respectively): that it perceptually appears to me that either there is a yellow crescent-shaped
thing to my left or 2 + 2 = 4, and that it perceptually appears to me that Heather’s banana is yellow.
This way of specifying the content of veridical experience has a controversial consequence.
The truthmaker this banana’s being yellow and before me is necessary and sufficient for the truth of
the proposition that there is a banana before me. And, as I briefly mentioned at the end of section
9.1, it’s controversial whether the content of experience takes a stand on matters like whether there
are bananas before one. But there is a formulation that is neutral on this issue, namely: the content
of a veridical experience is a proposition concerning which perceptible properties are instantiated in
the subject’s environment, which is such that the perceived truthmakers are necessary and sufficient
for its truth. If it turns out that the property of being a banana isn’t a perceptible property, then this
formulation excludes it from the content of experience.
 Experiential Content and Naïve Realism   239

One can use broadly the same kind of reasoning to reconcile Naïve Realism
with the Medium Content View. Recall that the Medium Content View is the Mild
Content View plus the claim that perceptual experience consists in the subject
perceptually representing her environment as being a certain way. On this view,
perceptual representation isn’t what experience fundamentally consists in—this is
what differentiates the Medium and the Spicy Content Views. Since the Medium
Content View is silent on what experience fundamentally consists in, it cannot
provide an exhaustive personal-level psychological explanation of the epistemo-
logical, behavioral, and phenomenological features of an experience. We’re left
without an account of what experience fundamentally consists in, if not percep-
tual representation.
As in the previous case, Naïve Realism can be wheeled in at this point. Notice
that Naïve Realism and the Medium Content View aren’t competitors: the former
is an account of the fundamental nature of veridical experience, while the latter is
not. Hence, it’s possible (at least in principle) to combine them. For example, we
can say that the subject of a veridical experience perceptually represents her envi-
ronment as being a certain way in virtue of perceiving things in her environment—
e.g., that I perceptually represent the proposition that there is a yellow thing before
me in virtue of perceiving the banana’s being yellow. The idea is that in some sense
(which I won’t attempt to spell out here) the subject perceiving things in her envi-
ronment is more basic than the representational state; something about the latter is
explained in terms of the former (see Logue 2013a). Alternatively, we can say that
the representational state is a constituent of the subject perceiving things in her
environment—e.g., that my perceptually representing the proposition that there is
a yellow thing before me is a constituent of my perceiving the banana’s being yel-
low. The idea here is that the representational state is but one part of what veridical
experience fundamentally consists in (see Logue 2012a).22
What about Naïve Realism and the Spicy Content View? Surely, one might
think, the claim that veridical experience fundamentally consists in the subject
perceptually representing her environment as being a certain way renders Naïve
Realism explanatorily redundant. When it comes to veridical experience, Naïve
Realism and the Spicy Content View seem to be competitors—how could verid-
ical experience fundamentally consist in perceptual representation and perceiving
things in one’s environment?
As it happens, this rhetorical question has an answer. Recall that a philosoph-
ical theory of perceptual experience has several explanatory tasks: in particular,
it’s supposed to explain the epistemological, behavioral, and phenomenological
aspects of experience. Thus, it’s in principle possible to divide the labor across Naïve
Realism and the Spicy Content View: say, the latter explains the epistemological

22
Of course, a proponent of the Naïve Realism/Medium Content View package would have to
tell a different story about what non-veridical experiences fundamentally consist in. I’ll set this issue
aside since veridical experience is the main focus of this chapter, but see Logue 2012a for a suggestion.
 Reconciliatory Views
240   

role of experience while the former yields an account of the phenomenal character
of experience and the role it plays in facilitating action.23 The upshot is that Naïve
Realism and the Spicy Content View need not be in competition with each other.
The fundamental structure of veridical experience could be a composite of a prop-
ositional attitude and a perceptual relation, and it could be that this is the best way
to explain everything that needs explaining.24
Note that I’ve merely offered a template for reconciliation of Naïve Realism
with the stronger Content Views. I’ve said nothing about the benefits we would get
(if any) from combining Naïve Realism with either of them. My aim in this chapter
is simply to show that Naïve Realism is in principle compatible with these views.
Whether either of these combinations is well-motivated is a question that must be
left to another paper.

9.5. Conclusion

Naïve Realists have continued to resist the claim that experience has content, de-
spite compelling arguments in its favor (the arguments from accuracy and appear-
ing). I have proposed a two-prong strategy for talking the Naïve Realist down from
the ledge. First, I offered an argument for the Mild Content View that even the
most radical Naïve Realist wouldn’t reject. Second, I reconstructed what seems to
be the primary argument for the incompatibility of Naïve Realism and experiential
content, and identified a flaw in it. In particular, regardless of which of the Content
Views a Naïve Realist adopts, a story about why a crucial premise in the incom-
patibility argument is false is at least in principle available to her. Hence, Naïve
Realism can be reconciled with experiential content after all.

References

Antony, L. (2011). The openness of illusions. Philosophical Issues, 21, 25–44.

Block, N. (1990). Inverted earth. Philosophical Perspectives, 4, 53–79.
Block, N. (1996). Mental paint and mental latex. Philosophical Issues, 7, 19–49.
Brewer, B. (2006). Perception and content. European Journal of Philosophy, 14, 165–181.
Brewer, B. (2008). How to account for illusion. In A. Haddock & F. Macpherson (Eds.),
Disjunctivism: Perception, Action, Knowledge (pp. 168–180). Oxford: Oxford University
Press.

23
One could strengthen the Spicy Content View into the claim that perceptual experience
fundamentally consists in the subject perceptually representing her environment as being a certain
way, and nothing else. Of course, this type of content view is incompatible with Naïve Realism, but in
an uninteresting way (viz., by definition). Thanks to an anonymous referee for pointing out the need
to mention this.
24
I take it that the account of experience defended in Schellenberg 2011 is one version of this
view.
 Experiential Content and Naïve Realism   241

Burge, T. (1991). Vision and intentional content. In E. LePore & R. van Gulick (Eds.), John
Searle and his Critics (pp. 195–214). Oxford: Blackwell.
Byrne, A. (2001). Intentionalism defended. Philosophical Review, 110, 119–240.
Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press.
Chalmers, D. (2006). Perception and the fall from Eden. In T. Gendler & J. Hawthorne
(Eds.), Perceptual Experience (pp. 49–125). Oxford: Oxford University Press.
Crane, T. (2009). Is perception a propositional attitude? Philosophical Quarterly, 59,
452–469.
Fish, W. (2009). Perception, Hallucination, and Illusion. Oxford: Oxford University Press.
Johnston, M. (2006). Better than mere knowledge? The function of sensory awareness. In T.
Gendler & J. Hawthorne (Eds.), Perceptual Experience (pp. 260–290). Oxford: Oxford
University Press.
Kalderon, M. E. (2011). Color illusion. Nous, 45, 751–775.
Logue, H. (2012a). What should the Naïve Realist say about total hallucinations?
Philosophical Perspectives, 26, 173–199.
Logue, H. (2012b). Why Naïve Realism? Proceedings of the Aristotelian Society, 112, 211–237.
Logue, H. (2013a). Good news for the disjunctivist about (one of) the bad cases. Philosophy
and Phenomenological Research, 86, 105–133.
Logue, H. (2013b). Visual experience of natural kind properties: Is there any fact of the
matter? Philosophical Studies, 162, 1–12.
Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17, 376–425.
Martin, M. G. F. (2004). The limits of self-awareness. Philosophical Studies, 120, 37–89.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). Oxford: Oxford University Press.
Pautz, A. (2009). What are the contents of experiences? Philosophical Quarterly, 59, 483–507.
Price, R. (2009). Aspect-switching and visual phenomenal character. Philosophical
Quarterly, 59, 508–518.
Schellenberg, S. (2011). Perceptual content defended. Nous, 45, 714–750.
Siegel, S. (2006). Which properties are represented in perception? In T. Gendler & J.
Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Siegel, S. (2010). Do visual experiences have contents? In B. Nanay (Ed.), Perceiving the
World (pp. 333–368). Oxford: Oxford University Press.
Siegel, S. (2011). The Contents of Visual Experience. Oxford: Oxford University Press.
Stalnaker, R. (1984). Inquiry. Cambridge, MA: MIT Press.
Travis, C. (2004). The silence of the senses. Mind, 113, 57–94.
Tye, M. (2000). Consciousness, Color, and Content. Cambridge, MA: MIT Press.
 10

Love in the Time of Cholera

Benj Hellie

10.1 Does Perception Have Content?

David Lewis thought so. In his view, “someone sees if and only if the scene before
his eyes causes matching visual experience,” where “visual experience has infor-
mational content about the scene before the eyes, and it matches the scene to the
extent that its content is correct.”1 “Visual experience”? Lewis presupposes that
this is a sort of state that “goes on in the brain” (Lewis 1980, p. 239). And he states
that “the content of the experience is, roughly, the content of the belief it tends to
produce”—more precisely, “only if a certain belief would be produced in almost
every case may we take its content as part of the content of the visual experience”
(p. 240).
To see the relevance of these views for the question of this volume, let us
generalize. Seeing is a kind of “perceiving” or “perception”: other kinds include,
at least, hearing, feeling, smelling, and tasting. Visually experiencing—perhaps—is
a kind of “sensorily experiencing”: if so, other kinds include, at least, auditorily
experiencing, tactually experiencing, olfactorily experiencing, and gustatorily expe-
riencing. A scene before the eyes is, perhaps, a kind of “perceptual surround”: if so,
other kinds include, at least, the sounds around the ears, the impingements in and
on the body, the aromas drawn in through the nose, and the flavors in the mouth.

Thanks first and foremost to Jessica Wilson. Thanks also to, in no particular order, Susanna
Siegel, Berit Brogaard, Geoff Lee, David Chalmers, Nico Silins, Jeff Speaks, Heather Logue, Nate
Charlow, Seth Yalcin, Cian Dorr, Tim Button, Laurie Paul, Shamik Dasgupta, Jessica Moss, Ned Block,
John Morrison, Dominic Alford-Duguid, Luke Roelofs, Adam Murray, David Balcarras, Alex Byrne,
Brie Gertler, Jacob Berger, Casey O’Callaghan, Herman Cappelen, Sebastian Becker, Andrew Sepielli,
and many others. Distinctive recognition is due, finally, to Mohan Matthen: for it is through Mohan’s
agency that the lion’s share of the material in this chapter was in a position to take wing.
1
Actually, he thinks the biconditional is “not far wrong.” First, causal chains come cheap: an
intricately refined replacement for “causes” is required to secure sufficiency—the “veridical
hallucination” of the title is a case Lewis takes to show this. Second, eyes are not required: rather, only
some sort of “optical transducer”—as, Lewis thinks, is shown by the “prosthetic vision” of the title. Our
complaints about Lewis’ view will target issues unaffected by these complications, so that ascribing to
242 him the cruder formulation in the body text buys simplicity without the cost of significant misdirection.
 Love in the Time of Cholera   243

So Lewis thinks that for Sam to be in an environmental state of perceiving is

for her to be in a brain state of sensorily experiencing which is caused by her per-
ceptual surround, and the content of which—the content of the belief caused by
almost every case of which—is correct. The brain state state has content. The brain
state is a part of the environmental state. So the environmental state of perceiving
has content “derivatively” by having a part, the brain state of sensorily experienc-
ing, which has content “more directly.” So says Lewis.
My story is a mixture of agreement and disagreement with Lewis’. Lewis is
right to think that perceiving involves causal impingement by the perceptual sur-
round on the organism. And he is right to think that “perceptual experience” has
content (or, I should say, there is content to the phenomenon coming closest to
deserving that vexed name). But he is wrong to think that “perceptual experi-
ence” is a part of perception: the relationship of these two is entirely different, in
a way incompatible with even the indirect assignment of content to perception.
And he is wrong to think it at all helpful to speak of “perceptual experience” using
efficient-causal idioms.
Now in slightly more detail.

• The point of agreement

Perception is something along the lines of a certain baseline of organ-

ismal sensitivity to the perceptual surround. It makes sense to use
“efficient-causal” idioms in discussing perception. And, indeed,
the ordinary case of perception surely does involve something like
“standard-causation” of a neural state by a perceptual surround.
There is something that comes close to deserving the name “perceptual
experience.” There is a certain aspect of the stream of consciousness
which stands in an intimate relation to perception. That aspect of the
stream of consciousness has content. It must, because it rationalizes
beliefs—and rationalization requires content (Lewis 1994).

• The first point of disagreement

Aspects of the stream of consciousness are not states or processes

“down there” in the “objective world.” Rather, my stream of conscious-
ness is something more like a window past which aspects of the objec-
tive world pass.2
If we think of the objective world as, so to speak, “lashed together in an
efficient-causal nexus,” then we might imagine that the sort of expla-
nation applying to the stream of consciousness is not efficient-causal.
Aspects of the stream of consciousness are explicable, if at all, rational-
ized by other aspects of the stream of consciousness.

2
For more on this, see Hellie 2013.
 Reconciliatory Views
244   

If perception is part of the efficient-causal nexus but the stream of

consciousness is not, then it sows confusion to think of perception
as “psychological”: we should think of it, rather, as organismal or
organismo-environmental or ecological. If perception is nonpsychologi-
cal, then, because only the psychological can have content,3 perception
doesn’t have content. And conversely, because “perceptual experience”
is most certainly psychological, and has content, “perceptual experi-
ence” can’t be part of perception.

• The second point of disagreement

Explaining the nature of the relationship between consciousness and

perception requires abandoning the Carnapian notion of “perceptual
experience” in favor of a notion with a greater degree of phenomeno-
logical veracity.
Namely, attention. Attention is, at the very least, the targeting of aspects
of the perceptual surround for inclusion within the stream of con-
sciousness. Recast in this idiom, the nature of the relation between per-
ception and “perceptual experience”—ahem, attention—springs into
focus. Attention is not a part of perception. Rather, perception supplies
and constrains attention.

• The third point of disagreement

While attention has content, it cannot fail to “match” the perceptual sur-
round—cannot fail to be “correct.” And moreover, the content is not the
“narrow” sort embraced by Lewis (1994). When an aspect of the per-
ceptual surround is targeted by attention, it is present within the stream
of consciousness: the stream of consciousness is characterized by a dis-
tinctively perceptual sort of assurance, or certainty, of the existence of
the target—has the content, “coded” in a distinctive way, that the target
exists. And, more alluringly still, the “nature” of the target is simply
“revealed,” in a way that leaves (at a certain level) no room for doubt
about what it is like.4 While content in general is used to model error
and ignorance, any mistakes or uncertainty I may make are “wrapped
around” a point of certainty about a minor but substantial matter: my
certainty that this exists as such. This certainty is the condition I am in
thanks to having a certain target of attention.

3
Well, can only have “original” content, as against the “derived” content of written messages, and
as against the “contentful stance” we sometimes take toward bread-baking machines.
4
Attention is in this way therefore similar to the “relational states” proposed by a number of
contemporary direct realists—although many of these theorists paint a view incompatible with the
ascription of content.
 Love in the Time of Cholera   245

Presentation within attention therefore provides me with my “cognitive

toehold” on reality: it is what distinguishes conscious life from a “fric-
tionless spinning in the void.”

The remainder of this chapter will flesh out this story in detail.
Doing so, regrettably, will require some terminological innovation. For the
ordinary notion of perception is too protean to be useful in philosophical theory, while
the philosopher’s notion of experience carries doctrinal baggage I reject. So the broad
phenomenon of the chapter will be labeled sensory consciousness (sometimes “senso-
rimotor consciousness”), by which we will mean, roughly, those aspects of a creature’s
conscious life that pertain to its “sense-perceptual” or “sensorimotor” condition.
These sensory aspects of consciousness are phenomenologically distinc-
tive: have a character that is immediately striking upon first-person contemplation
of what it is like to undergo them. This distinctive character is often thought to
involve a sort of presentation within sensorimotor consciousness of ingredients of
the objective world (Martin 2004): things around one; one’s own body; the motor
activity of one’s body in relation to the things around one.
If we set sensory consciousness in its broader phenomenological context, it
is this presentational aspect that uniquely qualifies sensorimotor consciousness
to perform a variety of “rational-psychological” duties: duties of a semantic, epis-
temological, or praxeological sort. These include: advancing ingredients of the
objective world as topics for thought and talk (Snowdon 1992); opening a source
of evidence about the objective world (Hellie 2011); providing a sink for agency
in regard to the objective world. So if conscious life in the objective world makes
any sense, the presentational capacities of sensory consciousness must be secure.
Unfortunately, philosophical challenges to presentation remove this security. For
it can seem that what it is like for one can remain fixed over an interval during which
consciousness becomes “disengaged” from the objective world (Valberg 1992); and it
can seem that the subject’s contribution to what sensory consciousness is like threat-
ens to overwhelm any contribution of what is allegedly presented (Hellie 2010).
This chapter will follow out this dialectic. We turn immediately to a theory of
the structure of sensory consciousness; the phenomenon of presentation can be
clearly located within this structure. We then defend the rational-psychological
necessity of presentation. We conclude with discussion of these philosophical
challenges to the possibility of presentation. A crucial aspect of the discussion
will be recognition of the deep nonobjectivity of consciousness, a notion expanded
upon in the technical appendix.

10.2. Presentation within Sensory Consciousness

The theory of sensory consciousness used in this chapter is, in outline, the fol-
lowing. At each moment of a creature’s life, the creature and its environment
 Reconciliatory Views
246   

are aspects of a particular extremely rich and intricate course of “sensorimotor”

interaction. This sensorimotor process is “objective”: neither it nor any of its
aspects is essentially a part of the creature’s “subjective” conscious life; it can be
fully understood from the “third-person” point of view. Still, at each (waking)
moment of the creature’s life, various aspects of the sensorimotor state are “drawn
up within” conscious life: are presented (or given) within the creature’s conscious
picture of the world as the momentary “anchor” dropped by the objective world
within conscious life. This drawing up/presentation/givenness of a particular is
what we colloquially call attention to that particular.
We now expand on six points of detail. The first concerns our talk of “objective
things” and “the objective world.” The chapter adopts this manner of speaking for
economy and vividness of expression; the subjective/objective distinction applies
more literally to modes of presentation (a more rigorous statement of this idea is
found in the technical appendix). To illustrate. When Mo studies a creature—such
as Sam—as a physiological system (or as an abstraction from an ecological system),
Mo’s manner of understanding is “third-person”: is attained within a perspective
on Sam. By contrast, Sam herself, in her conscious life, understands herself in a
manner that is “first-person”: is attained within a perspective from Sam—within
her conscious life (Harman 1990a). The specific character of a certain episode of
Sam’s understanding of the latter sort is the specific way Sam’s conscious life, or
existence, or being, is for her: is “what it is like to be Sam.” Mo, like every creature
distinct from Sam, cannot adopt the very perspective from Sam: after all, doing
so requires being Sam. And yet, a phenomenological manner of understanding
Sam—a “knowledge of what it is like” for Sam—is not restricted to Sam. Through
sympathy—through temporarily making himself (as a creature) more like Sam (as
a creature)—Mo’s own temporary first-person understanding becomes more like
Sam’s; Mo attains a “second-person” perspective in which he attains some knowl-
edge of what it is like to be Sam (Heal 2003).
The second point concerns the general structure of the sensorimotor state.
The sensory, recall, is objective: accordingly, a theory of its structure should not
involve phenomenological notions (“represents,” “aware,” “attention,” “looks”);
and while phenomenological appeals are legitimate in the “context of discovery,”
they should be treated with great delicacy in the “context of justification.” Research
keeping these points firmly in mind (Gibson 1979; Thompson 2007; Matthen 2005;
Noë 2006) coalesces around a vision of an ecological process: a “feedback loop”
from which the creature (Sam) and her environment are not genuinely dissociable,
and of which no momentary snapshot can be understood outside of its enduring
context. Much of the literature resists the “broad” and “holistic” aspects of this
vision: each sensory state is dissociated from the broader process; creature is dis-
sociated from environment; motor “output” is dissociated from sensory “input.”
Within these parameters, the sensory is sometimes treated as a relation, often a
relation of “awareness” (Moore 1903): perhaps to the environment (Campbell
2002); perhaps to “internal sense-data” (Russell 1910–11). But the sensory might
 Love in the Time of Cholera   247

also be treated as monadic (Ducasse 1942): perhaps as a physiological condition

of “irritation” (Quine 1960); perhaps as a semantic condition of “representation”
(Harman 1990b). Perhaps resistance to the ecological vision sometimes stems from
the phenomenological considerations to be discussed in section 4. But because the
sensory is objective, establishing the pertinence to it of those considerations would
require extensive elaboration.
The third point concerns the “constituency” of the sensorimotor state: that
which can be found in it, or abstracted from it. Among the sensorimotor state’s
central jobs is that of providing the subject-matter for ordinary judgments of per-
ception and sensation (along with motor behavior). So if we think Sam sees a truck
or a dog or a book or a rainbow or a mirror image, smells an aroma of chili, or
suffers a pain in her shoulder or a spell of double vision—then Sam’s sensorimotor
state should accordingly embed her in some sort of visual condition in regard to
truck or dog or book; or in regard to “rainbow-relevant phenomena”; or in regard
to the mirror and what it reflects; and should embed her in an olfactory condition
in regard to some “chili-scent-waft” phenomenon; and should locate “painfulness”
somehow in her shoulder; and should somehow qualify her visual condition with
some sort of “doubledness” somehow in her eyes and brain. Conversely, what Sam
does not perceive or sense—the remote, the tiny, the subtle, the obscured, the
occluded—should in general be absent from her sensorimotor state. Judgments of
what is perceived or sensed are often highly indeterminate; a full theory of sen-
sorimotor processes should reflect and explain that. Resolving the vexation stem-
ming from such phenomena as rainbows, mirrors, and double vision will require
delicate balancing of methodological and ontological issues. Again, care should be
taken to avoid phenomenological notions like the looks, sounds, smells, tastes, and
feels of things; and, conversely, to avoid contaminating solutions to the phenom-
enological puzzle of “distinguishing the senses” (Grice 1962) with considerations
more appropriate to the objective characterization of sensorimotor processes.
Fourth, we elaborate the notion of “presentation.” To fix ideas, stare at this
page: there it is. When the page is, in this way, there, we shall say it is “present
within conscious life” (“present,” “presented”). Philosophical traditions have
labeled this phenomenon in various ways: the page is “up against you,” “given,”
“apprehended in intuition.” That this phenomenon of presentation is at least
prima facie genuine is widely acknowledged (Price 1932–50; Valberg 1992; Hellie
2011); indeed, its first-blush allure is arguably the central source of dialectical
tension in the analytic philosophy of perception (Martin 2000). Presentation
is, moreover, of philosophical interest because if it is genuine, it would involve
an incursion or intrusion of the objective within the nonobjective—of “brute”
nonconscious matter within conscious life. Presentation does not seem to be
restricted to objects, in a strict sense: the features of objects (the white of the page)
are candidates, as are events (the utterance of a sentence; the throbbing of a pain),
as are courses of motor activity; and as are, perhaps, whatever smells or rainbows
may be.
 Reconciliatory Views
248   

Fifth, we elaborate the connections between attention and our other notions.
Attention and presentation are linked: what is (or are) present is exactly what
is (or are) the target of attention. The concept of attention is phenomenological
in nature: known and understood ultimately from within conscious life as the
visage of however one comports oneself such that something becomes present.
Nevertheless, attention involves an admixture of the objective: what is a candi-
date target of Sam’s attention is exhausted by what is a constituent of her sensori-
motor state. Still, while the sensorimotor state is objective, that one turns attention
on some aspect of it is not. Attention is therefore the “porthole” in conscious life
through which the objective world drops its anchor; or perhaps the “lashings”
with which conscious life stays moored to the objective world. (Perhaps there is
some objective “realization” of attention: if so, the risk of terminological confusion
would be reduced by calling it “centering” or “tracking.”)
And sixth and finally, we remark on a range of phenomenological features of
attention. Ordinary discourse recognizes looking at x, feeling x, tasting x, smelling
x, and listening to x; these are all varieties of targeting attention on x. Each of these
varieties itself doubtless comes in still more determinate varieties: for example,
staring and luxuriating. Typically we find attention used in conscious life as an
inextricable part of an activity. Consider reading, chasing down a fly ball, assem-
bling a ship in a bottle, dancing, conversing, searching for one’s keys, analyzing a
piece of music or a wine: one performs such an activity attentively just when one
turns attention to those aspects of the sensorimotor process in which the action
unfolds within the objective world. What it is like is not exhausted by what is the
target of attention: intense focus on a tomato and a passing glance differ phenome-
nologically. Moreover, which activity one is performing attentively seems to make
a distinctive phenomenological contribution: an artist making a final survey of a
painting and a gallery visitor studying that painting might glance over the very
same regions of the painting, but what it is like for them in doing so would differ
dramatically.
To summarize. (A) What it is like for Sam—the distinctive character of her
conscious life—is a nonobjective matter. (B) Aspects of what it is like for her
include (but are not exhausted by) every fact concerning what is present within
conscious life. (C) An entity x is present within Sam’s conscious life (and that x
is so present is part of what it is like for her) at a moment just if Sam then targets
x with attention; and (D1) if she does target x with attention, x is a constituent of
Sam’s sensorimotor state. But not conversely: (D2) most constituents of Sam’s sen-
sorimotor state are not targets of attention. While (E1) Sam’s sensorimotor state is
composed in part of what she perceives or senses, (E2) much remains beyond the
scope of Sam’s sensorimotor state. Finally, (F) Sam’s sensorimotor state is an ingre-
dient of the objective world, so that its constituency is an objective matter.
Distinguishing the objective phenomenon of the sensorimotor from the non-
objective phenomenon of conscious life permits an attractive description of the
following sort of case:
 Love in the Time of Cholera   249

Fred’s copy of Being and Time: it is on his bookcase somewhere. But where?
Fred combs every inch of the bookcase furiously, repeatedly, unsuccessfully.
His frustration mounts. Until, at last—there it is. Right in front of Fred’s nose
the whole time, he saw it but did not notice it—a source of great consternation.

Recent literature (Block 2011) draws a conundrum from the following assumptions.
(1) Fred sees whatever is right in front of his nose; (2) if Fred sees something right
in front of his nose, that the thing indeed is right in front of his nose is part of what
it is like for him; (3) what it is like for Fred explains what Fred thinks and does.
The conundrum is drawn out as follows. Being and Time is right in front of
Fred’s nose; so by (1), he sees it; so by (2), that Being and Time is right in front of his
nose is part of what it is like for Fred; so by (3), what Fred thinks and does is made
sense of by the fact that part of what it is like for him is that Being and Time is right
in front of his nose. But it isn’t: if that is part of what it is like for Fred, he should
reach out and grab the book rather than continuing the search.
The literature presents a choice between poverty and excessive wealth: some
deny (1), concluding that what we see is impoverished relative to what we think we
see; others deny (3), concluding that what it is like for one is enriched relative to what
we think it is like for one. But the poverty response loses the distinction between
Fred’s case and a search for something simply unseen: the latter should not provoke
the consternation Fred displays in the example. And the wealth response severs the
evident connection between consciousness and rationality: if the location of Being
and Time is within Fred’s conscious life and yet he acts in a way that (we would have
thought) makes no sense in light of that, the rational role of consciousness is cast
into obscurity.
The theory of this section allows the following story. By (3), the rationality
of Fred’s search depends on what it is like for him. By (B), it makes sense to
assume that Fred’s search is rational just if Being and Time is not present within
conscious life. By (C), the search is rational just until Fred’s attention alights on
the book. By (D1), ending the search is only rational if the book is an aspect of
Fred’s sensorimotor state. So, by (E1), at least when the search ends, Fred sees
the book; but moreover, continuing the search can be rational even if the book is
an aspect of Fred’s sensorimotor state: after all, (D2) means that even if attention
has not alit on the book, the book can nonetheless be an aspect Fred’s sensori-
motor state and therefore (E1) seen—preserving consistency with (1). Wrapping
up, (E2) preserves the consternation-free case in which Being and Time is at
home.
So, by asserting both (D1) and (D2), we drive a wedge between the targets of
attention (and onward to presence, what it is like, and rationality) and the constit-
uents of the sensorimotor state (and onward to what is seen)—and are therefore
in a position to reject (2).
But how do we have the right to (D1) and (D2) simultaneously? The literature
embeds a widespread presupposition that consciousness is objective, involving
 Reconciliatory Views
250   

“subjects of experience” instancing “sense-perceptual qualia”—where states of

such instancing are rather like narrow and monadic sensory states (Block 1995;
Chalmers 2010). But this picture makes (D1) a trivial consequence of (B) and
(C), while engendering a very strong tension among (B), (C), and (D2): (B) and
(C) mean that all qualia are targets of attention; but (D2) means that some qualia
are not targets of attention. The “higher-order” maneuver (Rosenthal 2005) of
making consciousness into attended qualia, where attention is an objective prop-
ositional attitude, restores the significance of (D1) and the coherence of (D2); but
the cost is to raise the vexing question of how an objective propositional attitude
and qualia, neither by itself an article of consciousness, can collectively amount to
consciousness.
With (F), our theory acknowledges the objectivity of the sensory. But with
(A), it rejects the objectivity of consciousness—and is therefore compatible with
neither the qualia nor the “higher-order” approach. The right to (D1) and (D2) is
secured by abandoning the objectivity of consciousness.

10.3. The Importance of Presentation

Attention to a tomato drops the tomato as an anchor of the objective world within
Sam’s conscious life. To appropriate John McDowell’s vivid metaphor (McDowell
1994), it is this anchoring that distinguishes conscious life from a “frictionless
spinning in the void.” As discussed above, a central aim of the philosophy of per-
ception has long been to secure the apparent friction against the concerns to be
discussed in section 4. In this section, we discuss an explanation of why presenta-
tion is worth the bother.
In a nutshell: without presentation, rational psychology—and probably con-
scious life—in an objective world would be impossible. In outline, a sort of “tran-
scendental argument”: (1) Rational psychology is about picture of the world and
stock of actions, both of them grasped through understanding how they evolve
intelligibly (Stalnaker 1984; Anscombe 1963). In particular, (2) one’s picture of
the world evolves through the accumulation of evidence (Lewis 1973), while one’s
stock of actions evolves through the discharging of plans (Bratman 2000); where,
still more specifically (in a way apparently required by embodiment), (3) one accu-
mulates evidence by gradually making more precise one’s certainties about the ob-
jective world as regards the evolving sensorimotor processes of a certain creature
and discharges one’s plans by contouring those same sensorimotor process. So un-
less the link in (3) is intelligible, (RD) our understanding of rational psychology is
thoroughgoingly “semantically defective.” But intelligibility is a phenomenological
notion, in at least the weak sense that (4W) whether someone evolves intelligibly
over an interval is determined by what it is like to be them over the interval; and
perhaps also in the strong sense that (4S) there is nothing to what it is like beyond
that which is relevant to intelligibility. So, by (4W), unless (L) some aspect of what
 Love in the Time of Cholera   251

it is like makes the link in (3) intelligible, (RD) follows; and perhaps by (4S), unless
(L), (CD) our understanding of consciousness is also thoroughgoingly semanti-
cally defective. But (RD) is perilously close to the baffling claim that there are no
truths of rational psychology, and (CD) is perilously close to the absurd claim that
there are no truths about consciousness. Fortunately, we can avoid (RD) and (CD),
for (5) presentation within conscious life can, and can alone, suffice for (L)—can
be the aspect of what it is like that makes the link in (3) intelligible.
Now in a bit more detail. Principles (1) and (2) are ancient framework doc-
trines best explored more deeply in another forum. Principle (3) is obvious: for
each of us, there is a creature about which we care in a manner that is absolutely
sui generis, the death of which would extinguish consciousness; it is this creature’s
sensorimotor processes which serve as evidence source and agency sink.
Now to (4W). Intelligible evolution is one with the availability of rationaliz-
ing explanation; of answers to “why” asked with a distinctively rationalizing spirit
(Anscombe 1963). For example: suppose that Fred has leapt to his feet, and that we
wonder why. An “efficient-causal” explanation of the sort offered by physiology is
not what we want: we don’t know any physiology, so such an explanation would be
so much gibberish to us. What we wonder, rather, is what Fred saw in leaping to his
feet at that moment. Citing facts utterly beyond Fred’s ken would therefore be of
no assistance: if Fred thereby narrowly avoided being struck by a flying bottle, that
would be of no explanatory force unless that he did so was part of his picture of the
world. Nor would some sort of “intentional stance”-type story in which some part
of Fred’s brain is treated as performing a calculation on representations (Burge
2005): an explanation offered in the course of Chomsky-type syntactic research
may elucidate how it comes about that a sentence strikes one as structured in this
way rather than that, but it offers no insight into what one sees in being struck by
the sentence in this way rather than that (indeed, one sees nothing in doing so: one
is simply so struck). Instead, what we want to know was what it was like for Fred
in the interval during which he leapt to his feet: what his conscious picture of the
world and conscious aims were such that leaping to his feet was the best action in
his repertoire for achieving those aims in a world like that. If we are told that, in
Fred’s view, the Mayor had just entered the room, that Fred seeks always to obey
protocol, and that Fred’s conception of how to do obey protocol when a high po-
litical figure enters a room calls for leaping to one’s feet, this gives us a sense of
what it was like for Fred; and we do find that if this is what it was like for us, we too
would leap to our feet. This may not be what it was like for us: we think it wasn’t
the mayor, are not especially concerned to obey protocol regarding this mayor, and
think the protocol for mayors doesn’t require leaping to one’s feet anyway. So that
we did not leap to our feet was overdetermined. Nevertheless, when we sympa-
thize with Fred, we understand why he did so.
Now, somewhat more speculatively, to (4S). This principle is in the spirit of,
and inherits the plausibility of, the widely discussed doctrine of “representation-
alism” (Harman 1990b; Chalmers 2004). Separated from its focus on exclusively
 Reconciliatory Views
252   

sensory consciousness and its commitment to objective “phenomenal properties”

as characteristic of conscious life, we are left with the reasonably salutary doctrine
that what it is like for one is just what (for one) the world is like—that conscious
life involves no “subjective qualities,” does not outstrip one’s picture of the world.
A still more salutary doctrine would reflect the practical side to conscious life; but
if so, we would be left with the view that all there is to know about what it is like
is what the objective world is like and what is to be done in it. (This leaves room
for presentation: the presence of something objective is not a kind of subjective
quality.)
Finally to (5). What is presented is, and is phenomenologically manifest as,
singled out as certain to be part of the objective world. Paraphrasing a famous
passage by Price (1950, p. 3): if one is presented with a certain aspect of a senso-
rimotor state, then—while one may be uncertain about the exact nature of that
aspect, or about the broader objective world “wrapped around” the aspect—one’s
conscious picture of the world displays certainty that the aspect is an ingredient of
the objective world and that the aspect is singled out as present within conscious
life. This phenomenologically manifest certainty simply is the accumulation of
evidence; grounded in presentation, this certainty faces no further demand for
justification. And while explaining the credibility of discharging plans on a single
objective sensorimotor process would be a more subtle matter, the manifest sin-
gling out of such a process may at least serve as a toehold. We see a way forward
on the links in (3).
Conversely, without presentation, these links become unintelligible. There
are two alternatives to the relevant sort of privileging of a single particular: the
privileging of many things; the privileging of nothing. For Sam’s conscious life to
privilege many objective particulars would be for Sam to be uncertain about which
location in the objective world is hers (creature Sam now? creature s' then? crea-
ture s" at some other moment?) for conscious life to leave it open which of many
candidate positions Sam occupies (our ordinary predicament, according to Lewis
(1979)). This would be phenomenologically distinct from Sam’s actual condition
just if the positions are objectively discriminable. So suppose that creature Sam
focuses on red and creature s' focuses on green. By the hypothesis, Sam’s conscious
life involves the presentation of both the former focus on red and the latter focus
on green. But the data is that Sam’s evidence about the objective world records
only what happens to creature Sam. So, phenomenologically, it is as if the presen-
tations of creature s' have been thrown out, and only the presentations of creature
Sam are recorded in the picture of the objective world. That would seem to be an
imposition of an alien intelligence: a decision that makes no sense in light of the
multiplicity of presentations. Similarly, for Sam’s plans to discharge only in the
creature Sam would seem to make no sense in light of the multiplicity of presenta-
tion. Why go all in with this creature when she might be that creature? Or why not
use both bodies? Again, this would seem to be an imposition of an alien conscious-
ness, would not make sense.
 Love in the Time of Cholera   253

At the other pole, for Sam’s conscious life to privilege no objective particular
would be for Sam to be forever ignorant about which location in the objective
world is hers (the situation of the “two gods” in Lewis 1979). Sam’s objective evi-
dence would, she would recognize, update in accord with the peregrinations of
creature Sam; and Sam’s plans would, she would recognize, discharge in a way
mirrored exactly by the motor behavior of creature Sam. Evidence would come out
of nowhere; plans would discharge into nothing. Sam would find herself simply
“saddled” with the evidence, would find her plans simply “falling away”: nothing
in conscious life would be present to make sense of all this coming and going. It
would be cold comfort to superadd the certainty that some creature is, bizarrely,
comporting itself exactly as if a source of Sam’s evidence and sink for her agency.

10.4. Challenges to Presentation

10.4.1. DISENGAGEMENT

In cases of “illusion” and “hallucination,” one does not recognize that one has
become somehow “out of touch” with one’s environment. Such cases have been
widely thought to demonstrate the impossibility of presentation, on grounds like
the following (Martin 2004):
(1) If anything is ever presented, what it is (perhaps in particular, perhaps
in kind) is then part of the character of conscious life. (2) For any course of
ordinary waking life, one could dream in such a way that what the course of
dreaming is like is no different from what the course of ordinary waking life
is like. Of course (3) the character of conscious life is just exhausted by what
it is like, so that (appealing to (2)) (4) the conscious lives of the ordinary sub-
ject and the dreamer do not differ in character. And if so (appealing to (1)),
(5) whatever is presented to either is presented to both. But surely (6) nothing
(either in particular or in kind) need be presented to both subjects. And if not
(appealing to (5)), (7) nothing is presented to either.
The premisses are (1)–(3) and (6). We have defended (1) and rejected (7) in the
previous section, and advocated (3) in the first section. Little light would emanate
from a challenge to the validity of the argument. So we face a choice of challenging
(2) and challenging (6).
Challenging (6) is the way of Russell (1910–11): even if nothing familiar is
presented to both, perhaps something unfamiliar is—an “internal” sort of
“sense-datum.” A first problem. The unfamiliarity of these internal sense-data is
a perpetual source of widespread philosophical distaste for this approach. This
distaste is a sure sign that we think that what is presented in ordinary waking life
is almost always something “external”; so Russell’s view means that, in ordinary
waking life (and also when we are taken in by a dream), we are almost always
 Reconciliatory Views
254   

mistaken about what is presented. The alternative this chapter will press predicts
that mistake is (strictly more) rare. In this respect, Russell’s approach is strictly
worse. This leads to a second problem: the stating of the approach undermines its
own solution; so the approach embeds a sort of “pragmatic contradiction.” For if
we come to understand what sense-data are, we will find it easy enough to concoct
a scenario in which what it is like is preserved while the availability of sense-data
is scrambled—destroying the solution. Any difficulty imagining such a scenario is
rooted in our cluelessness about what sense-data are. But that is also bad: for it is
cluelessness about what the theory says. This suggests a third, and fundamental,
problem. Russell’s proposed solution is, ultimately, utterly superficial: the dialec-
tical tension is rooted not in details of the constitution of the objective world, but
rather in our capacity to dissociate the first—and third-person perspectives on
our own lives. (From the first-person point of view, Sam regards herself as seeing
an anteater; when she later learns she was only dreaming, her sense of the situa-
tion from the third-person no longer coheres with her sense from the first-person;
bringing this sense into the “here and now,” one may temporarily adopt an alien-
ated perspective regarding one’s ordinary view as mistaken.) Pushed to its limit,
this strategy leads to Cartesian skeptical hypotheses: and presumably an evil
genius could blast out of the picture any sense-data (or other unfamiliar objective
things) that might have been lying around.
So we need an alternative to (2). Fortunately, the alternative is obvious
(Snowdon 1980–1; Martin 2004; Hellie 2011): when one is taken in by a dream of
looking at a tomato, one mistakenly thinks that one is looking at a tomato; and so
because looking at a tomato is the sort of activity that can be part of what it is like
for one, one therein mistakenly thinks that part of what it is like for one is that one
is looking at a tomato. If one is mistaken in this, then it is no part of what it is like
for one that one is looking at a tomato. And more generally, there is no course of
ordinary waking life for which there is a dream such that what they are like is the
same (though one is always at risk of being taken in).
One might well wonder what it could possibly be to be mistaken about what
it is like. The apparatus of the previous sections sheds light on the phenomenon;
we will develop the position in the course of the following dialogue with a skeptic
about such mistakes (“S” for skeptic; “U” for us):

S. One can’t be mistaken about what it is like for one: one always knows
exactly that it is just like this.
U. Even if so, this infallible but inarticulate sense for what it is like is
irrelevant. For (2) is about cross-comparisons; so unless (2) begs the
question, these must involve some “interpretive grain” mixed in with the
bare “this.” Where there is interpretation, there is misinterpretation. And
where there is misinterpretation, there is a mistaken picture of the world.
S. What is the mistake? (And what is it like when we don’t make the
mistake?)
 Love in the Time of Cholera   255

U. The literature focuses too extensively on good cases in which one

knows oneself to be seeing and bad cases in which one is taken in by
hallucination. But surely one can believe oneself to be hallucinating, as
during lucid dreaming (Hellie 2011): and once hallucination and belief
that one is seeing are prised apart, it becomes clear that one might
mistakenly believe waking life to be a dream. In addition to good and
bad cases, we need therefore to recognize also infragood cases (like
good cases in involving seeing; like bad cases in involving mistake about
whether one is seeing) and suprabad cases (like bad cases in involving
hallucinating; like good cases in involving correct belief about whether
one is seeing).

(What it is like when we don’t make the mistake is what lucid dreaming
is like.)
Having made this distinction, the mistake in the bad case is just
that one is seeing and that therefore the presented entity is a physical
object rather than a mental figment. But because any mental figment is
presumably essentially that way and any physical object is presumably
essentially that way, the bad-case judgment this is a physical object is
counteressential and therefore fails to draw up a coherent picture of
the world. The bad case is a case of a bad, because false, presupposi-
tion:5 nothing in the mind is “intrinsically bad”; badness is a Frege case,
resulting from unobvious misalignment of various components of one’s
picture of the world.

S. Are we allowing picture of the world to fix what it is like? If so, I applaud.
(A) One can’t be mistaken about one’s picture of the world, and
(B) presentation seems to have been cut out of the story—so what it is
like is fixed by a trouble-free “narrow” feature about which one can’t
be mistaken (compare the doctrine of “representationalism,” discussed
above).
U. (A) is mistaken. To see this, recall that David Lewis once believed that
Nassau Street runs roughly (to within 10 degrees) north–south, the train
runs roughly east–west, and the two are roughly parallel (Lewis 1982): his
picture of the world was inconsistent. Obviously he spent some time
unaware of this: when he recognized the inconsistency, straightaway it
vanished.

Why? It is in the spirit of (4S) from the previous section to think of

one’s own characterization of conscious life as presenting the world
­transparently and coherently. But the world as Lewis pictures it is not

5
Thanks to Dominic Alford-Duguid and Michael Arsenault for making this point especially
sharply.
 Reconciliatory Views
256   

coherent. And while the second-person perspective permits a sort of

scattered hopping around among three consistent fragments—but that
is a nontransparent perspective. So incoherence among fragmentary
pictures of the world inevitably carries with it error about one’s total
picture of the world.
So the picture is this: presentation of a certain sort of aspect of one’s
sensorimotor state brings with it conscious certainty that that aspect of
that sort exists. If other aspects of one’s conscious picture of the world
are incompatible with the existence of that aspect, one’s total conscious
picture of the world is incoherent—and so, against (B), presentation is
not irrelevant to what it is like. But because incoherence in one’s picture
of the world is unrecognizable, one makes some sort of mistake about
what it is like for one.

S. But why does one judge, in the bad case, that the presented figment
is (for example) purple rather than red? In the suprabad case, one
recognizes it to be neither, but simply possessed of a certain feature P
characteristic of some mental figments but not others. What is it about
feature P that combines with the mistaken belief that one is seeing which
misleads us into judging the figment to be purple (Speaks 2013; Logue
2013)? Internalists have a single feature to which they can appeal in
rationalizing both the good and bad case judgments—what about you?
U. The discussion of (6) shows that this is a problem for everyone; what
everyone should say is this. The incoherent bad-case subject has to make
some judgment, of course. But that subject is in the incoherent position of
accepting a counteressential content that this figment is a physical object;
and rational psychology is paralyzed in the presence of incoherence, and
therefore not up to the task of saying which thing it is. So the question
is misplaced. The best we can do is attempt to enter sympathetically
into the position of the other, shut off much of what we know, and think
from a position of self-imposed artificial ignorance what we would do
in the situation of the other. Turning a phrase from McDowell (1994) on
its head: we may have wanted justification, but we will have to settle for
exculpation.

10.4.2. THE SUBJECT’S CONTRIBUTION

If someone—Flip, for instance—is afflicted by spectral inversion, then what it is like

for him to see a red thing is the same as what it is like for an ordinary person—
Norma, for instance—to see a green thing; and vice versa; and so on around the
spectrum (Shoemaker 1991). Spectral inversion is a vivid example of the sort of
“subjective contribution” to what it is like that prompted Berkeley’s perplexity over
 Love in the Time of Cholera   257

a round tower that looks square from far off, over the large appearance to the mite
of what we find small. It sets in motion the following aporia:

Let Flip have normal color vision for a member of his species. It would
be hopelessly parochial to assert that Norma has but Flip lacks the cor-
rect view on which color this patch of grass is: the cosmopolitan recog-
nizes that each thinks it is green; more generally, that typically everyone
is right about the colors of objects.
But while Norma thinks green is a “cool” color, Flip thinks green is a
“warm” color.6 The higher-order features of “coolness” and “warmth”
are incompatible, cannot be both possessed by a single color; so at least
one of Norma and Flip is wrong. Which is it? It would be hopelessly
parochial to vote for Norma: the cosmopolitan recognizes that both of
them are wrong; more generally, that everyone is mistaken about the
natures of the colors.
But it is within presentation with colors that we arrive at our view on
the natures of the colors. Mistakes about nature are necessary false-
hoods: matters about which we can only be uncertain through gross
confusion. So here again we find a contra-essential error. Whichever
one of them is wrong is not just mistaken: their conception of green
is semantically defective. (Relativization—cool for us, warm for them—
would only push error to the third order: at the third order, warmth and
coolness are presented as “absolute.”)
So presentation is a source of confusion so gross as to engender seman-
tic defectiveness. But the central theoretical role of presentation is as
a source of infallible certainty about objects and their features. Once
burned, twice shy: our misadventures at the higher order should under-
mine our confidence at the lower order—indeed, we now see that they
infect what is meant at the lower order, rendering it unintelligible.
So nothing can fill the theoretical role of presentation: the phenomenon
does not exist. But, as discussed in section 3, without presentation, it
may be that there are no truths about consciousness. So if we reach this
stage, consciousness vanishes.

Perhaps the time is right to revive Kant’s approach to this aporia. The second leap
toward cosmopolitanism—recognition that the structure we find in the world,
we put in in the first place—is both compulsory and forbidden. Compulsory
for the theorist seeking the most objective possible viewpoint. But forbidden
because from that viewpoint, conscious life itself vanishes—and with it, the
theorist.

6
“Cool” and “warm” are a pedagogically-convenient stand-in for whatever higher-order features
we in fact find to distinguish green as we see it from red as we see it.
 Reconciliatory Views
258   

The second leap toward cosmopolitanism may be one we only perform some-
times, in safe circumstances quarantined from others where it would be genuinely
damaging. Why think either universal cosmopolitanism or universal parochialism
is required of us? A third option is the cynical adventitious cosmopolitanism of
the savvy politician: in the home province, affirming wholeheartedly local paro-
chial biases; in the capital, as easily abandoning them in a cosmopolitan spirit of
national compromise. Diachronically inconsistent; less than wholeheartedly sin-
cere at any moment; quite possibly distasteful if not vicious. And yet life goes on.
This is the First Critique’s discomfiting alternation between transcendental
idealism and empirical realism. Our reasoning tells us that transcendental ide-
alism must be correct. But because what transcendental idealism means can only
be grasped from a viewpoint that is unattainable if ordinary life is to continue, our
desire to leap beyond empirical realism will be forever frustrated.

10.5. Appendix: Semantics for Deep Nonobjectivity

Think of (nonextremal) propositions as representing answers to questions which

can be reasonably asked and universally answered. “Are your shoes tied” cannot be
universally answered (some rightly say yes, others no). Still, in any given circum-
stance, it corresponds to the question whether the addressee’s shoes are tied at the
moment, a question over which disagreement requires mistake. “Is 2 + 2 = 17 ”
cannot be reasonably asked, in the sense that if I know the right answer for you,
I can’t make full sense of what things would be like for you were you to give the
wrong answer; and I therefore can’t make full sense of what you are uncertain be-
tween if you don’t answer the question. But “are your shoes tied” can be reasonably
asked: even framed, against a given context, so that it can be universally answered,
I can make sense of how someone in a given context might get the wrong answer.
And “is 2 + 2 = 17 ” can be universally answered: the correct answer for everyone
is in the negative.
The full field of propositions includes all nonextremal propositions as well
as two limit cases: the trivial proposition, representing the correct answer to any
question which can’t be reasonably asked; and the absurd proposition, represent-
ing its incorrect answer. The notion of “the objective world” used in the main body
of the paper, then, corresponds to the totality of correct answers to questions rea-
sonably asked and universally answered; alternatively, it is the totality of correct
propositions.
A question that cannot be universally answered is in one respect nonobjec-
tive: it is situated, with potentially varying answers for me and for you. A question
that cannot be reasonably asked is in another respect nonobjective: it is superfi-
cial, with an answer at no distance from conscious life; whatever it concerns, its
condition is not ultimately potentially elusive to reason. A question that can be
reasonably asked and universally answered is deeply objective: its subject-matter is
 Love in the Time of Cholera   259

the structure of the brutely nonconscious, considered in isolation from any pecu-
liarities of consciousness. Are any questions deeply nonobjective—on the one hand
situated; but on the other hand superficial?
Yes. One such question is “is it like this:?” The question is situated. For Fred,
who falsely believes that goats eat cans, the answer is affirmative. For me, it is
negative. Before I learned that goats do not eat cans, the answer was also negative
(but so was the answer to “is it like this:?”, a question I now answer in the affirma-
tive). But the question is also superficial: because it is like that for Fred, I can’t
make sense of what things are like for him if he answers otherwise; and therefore
I can’t make sense of what things are like for him if he takes both answers seriously.
A sentence ϕ entails a sentence ψ just if whenever a context c affirms ϕ,
c affirms ψ . A proposition P is a subset of W, “modal space,” the set of possi-
ble worlds. In a context-sensitive propositional semantics, a declarative sentence
ϕ receives a proposition ϕ  as its semantic value against a context c. In a
c

truth-conditional semantics, affirmation is verification: c verifies ϕ (c |= ϕ) just if

wc ∈ϕ  , for wc ∈W the “world of the context.” In a mindset semantics, affirma-
c

tion is support: c supports ϕ (c  ϕ) just if ic ⊆ ϕ  , for ic ⊆ W the “information

c

state of the context.”

A rigidifying operator returns an extremal proposition against c corresponding
to the affirmation-value of its prejacent at c. In truth-functional semantics, the char-
acteristic rigidifier is A, the familiar “actuality” operator, with  Aϕ  = {w : c ⊧ ϕ}; in
c

mindset semantics, the characteristic rigidifier is ∇, the “Veltman rigidifier” (Veltman

1996; Yalcin 2007), with ∇ϕ  = {w : c  ϕ}—with an intuitive reading along the
c

lines of “certainly.”
These rigidifiers exhibit inferential properties corresponding to their associated
notions of affirmation. Verification is “self-dual”: c ⊭ ϕ just if c |= ¬ϕ ; but support
is “non-self-dual”: sometimes both c  ϕ and c  ¬ϕ . As a result, while A inter-
acts with the classical connectives only classically, ∇ exhibits surprising interac-
tions with the classical connectives. In particular, both dilemma and reductio are
mindset-invalid: although ∇φ ⊣⊢ φ, ¬∇ϕ  ¬ϕ and ϕ ∨ ψ  ∇ϕ ∨ ∇ψ .
This makes ∇ distinctively useful in representing “transparent certainty”
(Hellie 2011). The equivalence of ϕ to ∇ϕ makes for transparency. And yet while
truth is bivalent, certainty is trivalent: whenever whether-ϕ can be reasonably
asked, it is coherent to accept any of ∇ϕ , ∇¬ϕ , and ¬∇ϕ ∧ ¬∇¬ϕ .
Because its being like such-and-such is a kind of transparent certainty, this
in turn makes ∇ϕ distinctively useful for representing “it’s like this: ϕ” For ϕ and
“it’s like this: ϕ” are equivalent—accepting one is accepting the other. And yet,
while accepting ¬ϕ requires accepting “it’s not like this: ϕ,” the converse is not
so: when consciousness is “grey” regarding whether ϕ, one accepts “it’s not like
this: ϕ” but does not accept ¬ϕ . If the context against which sentences are evalu-
ated represents what it is like for the subject under consideration, we can then say
that c ⊩ ∇ϕ just if, for that subject, it is like this: ϕ.
 Reconciliatory Views
260   

The overall picture advanced by the semantics is one on which we do not

think of distinctions in consciousness as distinctions in the world (Hellie 2013).
The field of propositions is one thing, the context entertaining them another: we
entertain the world by entertaining propositions from a context; our entertain-
ment of consciousness, by contrast, is nothing other than the context itself. That
is all to the good, because injecting distinctions in consciousness into the world
leads immediately to “Kaplan’s paradox”: with n worlds, there are 2n propositions;
with 2n propositions, there are 2n states of consciousness; with 2n states of con-
n
sciousness, if there are propositions about consciousness, there are 22 proposi-
2n n
tions about consciousness; but because 2 > 2 , that makes for more propositions
about consciousness than propositions—contradiction.

References

Anscombe, Gertrude Elizabeth Margaret (1963). Intention. Cambridge, MA: Harvard

University Press.
Block, Ned (1995). On a confusion about the function of consciousness. Behavioral and
Brain Sciences, 18, 227–247.
Block, Ned (2011). Perceptual consciousness overflows cognitive access. Trends in Cognitive
Sciences, 15, 567–575.
Bratman, M. E. (2000). Ruling passions: A theory of practical reasoning. Philosophical
Review, 109, 586–589.
Burge, Tyler (2005). Disjunctivism and perceptual psychology. Philosophical Topics, 33, 1–78.
Campbell, John (2002). Reference and Consciousness. Oxford: Oxford University Press.
Chalmers, David J. (2004). The representational character of experience. In Brian Leiter
(Ed.), The Future for Philosophy (pp. 153–81). Oxford: Oxford University Press.
Chalmers, David J. (2010). The Character of Consciousness. Oxford: Oxford University Press.
Ducasse, C. J. (1942). Moore’s refutation of idealism. In Paul Arthur Schilpp (Ed.), The
Philosophy of G. E. Moore (pp. 223–53). LaSalle: Open Court.
Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Boston: Houghton Mifflin.
Grice, Herbert Paul (1962). Some remarks about the senses. In R. J. Butler (Ed.), Analytical
Philosophy, first series (pp. 133–53). Oxford: Blackwell.
Harman, Gilbert (1990a). Immanent and transcendent approaches to the theory of
meaning. In Roger Gibson & Robert B. Barrett (Eds.), Perspectives on Quine (pp. 144–
57). Oxford: Blackwell.
Harman, Gilbert (1990b). The intrinsic quality of experience. In James Tomberlin (Ed.),
Philosophical Perspectives, vol. 4: Action Theory and the Philosophy of Mind (pp. 31–52).
Atascadero, CA: Ridgeview.
Heal, Jane. 2003. Mind, Reason, and Imagination. Cambridge: Cambridge University Press.
Hellie, Benj. (2010). An externalist’s guide to inner experience. In Bence Nanay (Ed.),
Perceiving the World (pp. 97–145). Oxford: Oxford University Press.
Hellie, Benj. (2011). There it is. Philosophical Issues, 21, 110–164.
Hellie, Benj. (2013). Against egalitarianism. Analysis, 73, 304–320.
Lewis, David. (1973). Why conditionalize? In his Papers in Metaphysics and Epistemology
(pp. 403–7). Cambridge: Cambridge University Press.
 Love in the Time of Cholera   261

Lewis, David. (1979). Attitudes de dicto and de se. The Philosophical Review, 88, 513–543.
Lewis, David. (1980). Veridical hallucination and prosthetic vision. Australasian Journal of
Philosophy, 58, 239–249.
Lewis, David. (1982). Logic for equivocators. Noûs, 16, 431–441.
Lewis, David. (1994). Reduction of mind. In Samuel Guttenplan (Ed.), Companion to the
Philosophy of Mind (pp. 412–31). Oxford: Blackwell.
Logue, Heather (2013). But where is a hallucinator's perceptual justification? In Richard
Brown (Ed.), Consciousness Inside and Out: Phenomenology, Neuroscience, and the
Nature of Experience (pp. 155–62). Dordrecht: Springer.
Martin, M. G. F. (2000). Beyond dispute: Sense-data, intentionality, and the mind-body
problem. In Tim Crane & Sarah Patterson (Eds.), History of the Mind-Body Problem
(pp. 195–231). London: Routledge.
Martin, M. G. F. (2004). The limits of self-awareness: Disjunctivism and indiscriminability.
Philosophical Studies, 120, 37–89.
Matthen, Mohan P. (2005). Seeing, Doing, and Knowing: A Philosophical Theory of Sense
Perception. Oxford: Oxford University Press.
McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
Moore, George Edward. (1903). The refutation of idealism. Mind, 12, 433–453.
Noë, Alva (2006). Experience without the head. In Tamar Szabó Gendler & John Hawthorne
(Eds.), Perceptual Experience (pp. 411–33). Oxford: Oxford University Press.
Price, Henry Habberly (1950). Perception. London: Methuen.
Quine, Willard Van Orman (1960). Word and Object. Cambridge, MA: MIT Press.
Rosenthal, David M. (2005). Consciousness and Mind. Oxford: Oxford University Press.
Russell, Bertrand (1910–11). Knowledge by acquaintance and knowledge by description.
Proceedings of the Aristotelian Society, 11, 108–128.
Shoemaker, Sydney (1991). Qualia and consciousness. Mind, 100, 507–524.
Snowdon, Paul (1980–81). Perception, vision, and causation. Proceedings of the Aristotelian
Society, 81, 175–192.
Snowdon, Paul (1992). How to interpret ‘direct perception’. In Tim Crane (Ed.), The Contents
of Experience (pp. 48–78). Cambridge: Cambridge University Press.
Speaks, Jeffrey (2013). Some thoughts about hallucination, self-representation, and
‘There it is’. In Richard Brown (Ed.), Consciousness Inside and Out: Phenomenology,
Neuroscience, and the Nature of Experience (pp. 147–54). Dordrecht: Springer.
Stalnaker, Robert C. (1984). Inquiry. Cambridge, MA: MIT Press.
Thompson, Evan (2007). Mind in Life: Biology, Phenomenology, and the Sciences of Mind.
Cambridge, MA: Harvard University Press.
Valberg, J. J. (1992). The Puzzle of Experience. Oxford: Clarendon Press.
Veltman, Frank (1996). Defaults in update semantics. Journal of Philosophical Logic, 25,
221–261.
Yalcin, Seth (2007). Epistemic modals. Mind, 116, 983–1026.
PART FOUR

Imagistic and Possible-Word

Content
11

Image Content
Mohan Matthen

It is a lamentable but inescapable fact that the world is not always as it appears to our
senses. On the brighter side, one can normally rely on any given appearance being
true. In short, the senses are reliable, but not perfectly so.
These propositions were, throughout most of history, taken for granted by most
philosophers, who, depending on their philosophical bent, emphasized one of them
over the other. In ancient Greece, Plato led the pessimists. He maintained that the
senses are unreliable and inchoate; what they tell us must be clarified, if not wholly
supplanted, by reason. The ancient sceptics overshot Plato: according to them, the
senses carry no credibility; nothing they tell us is more probable than the opposite.
Aristotle led the optimists: the senses provide a platform for systematic knowledge, he
says (Post. An. II 22). The Stoics went even further—according to them, some kinds
of perception are guaranteed to be true. Philosophers in ancient and medieval India
pursued the same lines of thought. They acknowledged sensory illusion. The central
problem of epistemology, according to them, was whether sensory illusion betokens
a veil of ideas between the perceiving subject and the external world. Philosophy
in the modern period continues these debates, albeit with increasing precision and
refinement.
All of these approaches to perception imply (on very natural assumptions) that
the perceiving subject is presented with a proposition—the proposition that the world
is a certain way. This proposition is often entitled the content of a perceptual state.
A subject’s perceptual experience is true or false according to whether the proposition
it presents—that is, the proposition that constitutes its content—is true or false. When
something looks green, the perceiver is being presented with the proposition that it is
green; if the thing is green, the propositional content of this presentation is true, and
hence the appearance itself is also true.
As the present volume demonstrates, claims such as these have come under
fire from many directions. Even in the ancient world, Epicurus denied that sensory
appearance could be false. His argument is especially interesting to my project in this
chapter because it explicitly rests on something like the idea that perception is an
image. “The portrait-like resemblance of impressions . . . would not exist if the things
265
 Imagistic and Possible-Word Content
266   

which we come into contact were not themselves something,” he wrote.1 The idea is
that external things leave their mark on the senses; these marks are the consequence
of a process of transmission from external object to the senses. Since this process
preserves a certain resemblance, sensory impressions are “portraits.”
If Epicurus is right, the “portrait-like” sensory appearance is merely a causal
trace. As such, it cannot have propositional content. The image cast on a screen
by a slide-projector is a copy of the slide, but it would be strange to call it “true”
for this reason. Epicurus thought that sensory appearance is like this projected
image. We can use it to figure out the condition of external things by inference
from nature of effect to nature of cause. Thus, he holds that “falsehood and error
are always located in the opinion which we add.” In effect, this position conflicts
with the idea that sensory appearance can be true: for if falsehood is “located in the
opinion which we add,” so also is truth. In him, therefore, we find one form of a
“no-content” view of the senses. (The impressed-image theory is inconsistent with
what we now know of sensory processing, as I will indicate in section 11.9, and an
untenable basis for a no-content view.)
These ancient positions, as well as their more recent descendants, are interesting
not just for what they say, but also for what they omit. Sensory content is imag-
istic, as Epicurus explicitly acknowledges. Opinion and belief, by contrast, are (in
standard cases, at least) linguistically expressed. Belief is not normally directed at
an image—my opinion that the sky will be blue today is directed toward a sentence,
not an image of the sky. One can, of course, believe an image, as I will later indicate.
But there is still a serious omission in contemporary theory: there is no discussion
of how an opinion can interact with an image. Since the former is linguistic in the
manner it conveys content, and the second is not, this is a problem—images cannot
bear on sentential attitudes directly; only a sentential description of an image can
do so.
In this chapter, I seek to clarify the nature of image content and its expressive
limitations. In particular, I shall inquire into how the characteristics of image con-
tent bear on the issues of representational content indicated above.

11.1. Image and Perception

Perceptual experience2 has image content. To wit:

(a) Perceptual experience presents the subject with a spatiotemporally

ordered array of sensory qualities.

1
Letter to Herodotus, § 51 (= Long and Sedley 15A11). Note: ‘portrait’ = eikon.
2
X perceives p only if p: for ‘perceives’ is what philosophers call “factive”. In my usage, “perceptual
experience” is experience as in perception, i.e., experience involving sensory qualia with the feeling
of here-and-now as per (b). Hallucinatory experience may be perceptual in this usage, though
hallucination is not perception, on account of its failing the factivity condition.
 Image Content   267

But this is a characteristic that perceptual experience shares with iconic memory
and sensory imaging. It is different from these other imagistic states because:

(b) Perceptual experience presents the subject with a state of affairs as

occurring here (in the sense that everything is experienced as located
relative to the perceiver) and now.

Memory and sensory imaging do not present states of affairs as occurring here and
now. This is how they differ from perceptual experience.3
It is hard (as we shall see) to understand how both (a) and (b) could be true.
Experiences of the sort described by (a) are imagistic—some of the key character-
istics of imagistically expressed content are laid out in sections 11.3‒11.4. But, as I’ll
argue in sections 11.6‒11.8, images are incapable of fixing places and times, except
relative to one another. That is, an image can present one thing as further away
and over to the left of another, but not as in a particular place (e.g., London) or at a
particular time (e.g., five minutes ago). So how does perceptual experience present
something as here (i.e., in this particular place) and now (at this time)? How can it?
This is just one question that arises from taking image content seriously.
Before we go on, a clarificatory remark is in order. Often, image content
is contrasted with propositional content, on the grounds that images cannot be
assessed, as propositions are, in terms of truth and falsity and related semantic
evaluations. In my view, this is a mistake.
Here is why. On any non-deflationary account (see section 11.9), an image is an
image of a state of affairs—an image of a way the world can be. (Put aside impossible
images, which require a more complicated account.) For example, an image of a face
presents a state of affairs in which a face exists in a certain place relative to a point
of view—a face that is such-and-such color, surmounted by thinning hair, with a
long nose, . . . and so on. The image may, further, present the face as singing a song,
having soft skin, and possessing other non-visual features. (My notion of an image is
multisensory; the “sensory qualities” of (a) above may belong to any modality.) We
can, therefore, assess the relationship between an image and any possible world. Is
the image an image of that possible world? Or: how true is the image of that world?
And this is just like the relationship between a sentence and a possible world.
An image is an image of a type of situation. It is not an image of a particu-
lar situation—rather, it is an image of any situation of the type that it depicts. In
order for it to be true—that is, true non-relationally, not merely true of a speci-
fied situation—it has to be applied to, or asserted of, a particular situation or ex-
istentially quantified over a range of situations. Perceptual experience applies its

3
Hume distinguished perceptual experience (“impressions”) from memory and imagination by
the “force and liveliness” of the former. He would have said that a state of affairs is experienced as “here
and now” because the idea of the state of affairs is livelier, and is hence an impression (not merely an
idea). It is noteworthy in view of the Shared Content Argument in section 11.6, that Hume defines
perceptual experience in a way that is independent of content.
 Imagistic and Possible-Word Content
268   

image-content to the perceiver’s here-and-now. It conveys this import simply in

virtue of the kind of experience it is. In other words, when a subject has a percep-
tual experience, its import to her is that the image she experiences is an image of
the situation here and now. It is true if the present situation is as presented.4
By contrast with perceptual experience, which presents a particular situation,
namely the here-and-now, and perceptual imaging, which presents a type of situ-
ation, recollection is quantified: it presents its image-content as being true of some
situation or other in the past that the subject was herself a part of, and occupied
a certain place relative to the point of view. And anticipation presents its image
content as true of a future situation, the time of which may be specified by an
accompanying thought. These asserted images are not applied to particular places
and times; rather they are quantified over place and time ranges. They are accurate
or inaccurate depending on whether there was or will be a situation of the type
represented. Asserted images possess truth-value and verisimilitude. On the other
hand, a merely imagined image is not asserted of any situation or range of situa-
tions. It entertains a situation-type, and does not present this as an image of any
particular situation. As such, it is neither true nor false.5
The contrast between image content and propositional content is a mis-
take, therefore. Image content expresses a proposition when it is asserted (and a
situation-type when it is not). The correct contrast is between sentential vehicles of
meaning and imagistic vehicles. Both sentences and images express propositions,
but they do so in different ways. They are, one might say, different formats for
expressing propositions. My aim in this chapter is to understand the peculiarities
of image content—that is, of imagistic vehicles of perceptual meaning.

11.2. From Image Content to Belief

Many philosophers assume that there is a non-problematic way to express sensory

content by means of sentences. This is natural enough in many cases. A sentential
report of perceptual content such as:
S. That looks like five men running toward me.

4
Tim Crane (2009, pp. 467–468) argues that pictures cannot be asserted except by using
non-pictorial means (or “symbols,” as he says). Taking a pass on ‘symbol,’ I agree. Crane thinks that this
thesis gives him reason to deny that perception is a propositional attitude. My claim is that perceptual
experience asserts an image of the here-and-now simply by virtue of being the kind of experience it is. It
is for this reason that perceptual experience can be true or false. This is why it is a propositional attitude.
5
In an otherwise excellent discussion of images from which I have learned much, Charles Siewert
(1998) suggests that a merely imagined scene could turn out to be true—for instance, when I imagine a
unicorn and it turns out that there is a unicorn that exactly matches my image. I think this is a mistake.
There was nothing in my imagining that conveyed to me that my image was an image of this or any
other actual creature.
 Image Content   269

is perfectly adequate for many purposes. In theoretical contexts, however, it can

be dangerous to use sentential specifications of sensory content without clarify-
ing how they stand to sensory images. Consider, for example, the very natural
claim that perception justifies belief (which occupies a central role in such works
as Pollock 1974 and McDowell 1994). It is obvious what this claim means when
perceptual content is specified sententially, as above. Unfortunately, though, the
relations that images bear to sentences are not well understood. And it is conse-
quently theoretically under-specified how perceptual images stand to sentences,
and consequently exactly what sorts of images support a belief like S.
The logical empiricists addressed the question of image-sentence relations
head-on. According to them, perceptual images (or what Hume called an “idea”) un-
problematically support certain simple sentences, which they called “protocol sen-
tences.” On the standard reading, protocol sentences are restricted to the specifying
the locations of sensory qualities—for example, “There is an instance of red over to
the left”.6 Now, it is not unanimously agreed exactly what features are given in sense
perception. Some, like Hume, take a very restrictive view (“if we see it, it is a color”).
For Hume, protocol sentences for visual content would be restricted to “There is
an instance of C in L,” where C stands for a color and L for a location. But others
(the Gestalt psychologists and, more recently, Spelke 1990, Xu 1997 and 2007, and
Matthen 2005, ch 12) hold that material objects and their shapes are directly seen;
this would allow a larger class of protocol sentences.7 To circumvent this disagree-
ment, let us just stipulate that there is a class of qualities directly given by sense per-
ception. The class of protocol sentences is defined relative to this class. I would add
that these qualities are given as inhering in objects located in shared public space.
Let us concede to the logical empiricists that the relationship between sensory
images and protocol sentences is unproblematic. There remains the question of
how we reason from perceptual image content to non-protocol beliefs. Consider
non-protocol perceptual beliefs that depend on past learning as well as perception.
For example, consider: “That is an Airbus A380 taxiing out for take-off,” or “That
man is interested in that woman”. We arrive at such beliefs on the basis of percep-
tion—how? According to the logical empiricists, such beliefs are derived from per-
ception in a two-step process. First, we move from perception to a set of protocol
sentences. Then we move to beliefs such as the above by the use of “constructions”
on sets of protocol sentences.

6
Christopher Peacocke’s (1983, 1992) notion of “scenario content” is restricted to such
feature-location content, though it is less conservative than the empiricist tradition, inasmuch as it
allows for perceptual representations of three-dimensional space. Austen Clark’s (2000) idea of located
features is another example.
7
It was, moreover, a matter of debate among members of the Vienna Circle whether protocol
sentences were descriptions of the subject’s own phenomenal state (and hence private) or of
feature-instances in shared real space. Carnap seems to have started with the phenomenal interpretation
and moved, under the influence of Neurath, toward the public space interpretation (Uebel 1992). I am
taking protocol sentences in the second way.
 Imagistic and Possible-Word Content
270   

For reasons I won’t discuss here, this is a discredited program. From my point
of view, the main objection is that there is no clear way to capture what an A380
looks like, given that we see such objects from many different points of view in
diverse conditions of illumination (etc.). Even more obviously, we cannot define
how sexual interest appears to the senses, in all of its many different contexts. In
short, we cannot define or “construct” such objects or features in terms of sensory
qualities alone.8 The interesting question arises, therefore, whether one can go di-
rectly from perceptual images to “rich” perceptual beliefs by-passing the interme-
diary stage of protocol perceptual beliefs. The problem is that this would require
an indirect and problematic image-sentence transition.
In sum, the obstacle to a satisfactory account of perceptual belief-formation
is that, putting protocol sentences aside, we don’t have a good general account of
transitions from image-content to sentence-content. On the other hand, we do
have good general accounts of sentence-sentence inference. The logical empiri-
cists solve the problem of belief-formation by relying on the special case where
we do have a clear transition. They assumed that they could then use the method
of construction to carry the inference train forward to rich perceptual beliefs.
Given the failure of their program, it seems that we would want to explore learned
image-sentence transitions. And this takes us back to square one: we have at best
a poor understanding of how we reason with images, as opposed to sentences.
Philosophers who discuss perceptual belief rarely see the need for an account of
this.9 John McDowell (1994) is an exception: his assertion of conceptual content
in perception is presumably meant to serve just this role. However, conceptual
content is just a placeholder, since we have no good idea of how the conceptual
content of perceptual images interact with concepts as they are found in sentences
and language (including the language of thought).
Though I will make a tentative suggestion about the transition from images
to rich perceptual beliefs, my purpose in this chaper is mainly negative. I shall
sketch some important characteristics of sensory images, with a view to showing
how these lead to certain expressive limitations. This rules out certain kinds of per-
ceptual content. More specifically, I’ll argue that images cannot express particular
times and places (though they can express relations of locations and times, such as
‘x is to the left of/earlier than y’). As well, they cannot refer to particular individu-
als. So if perceptual experience and iconic memory refer to particular individuals,
or express fixed spatiotemporal locations, they must do so in virtue of something
outside of the image they present. I’ll make a suggestion as to what this might be.

8
Others object on the grounds that the program is foundationalist. But first, this might be
historically false (Uebel 1993), and secondly, there is nothing wrong with foundationalism.
9
See, however, Elisabeth Camp (2007) for an insightful discussion of “Thinking with Maps.”
Camp uses “icons” with conventional meaning to carry some of the weight of the informational
content of maps; she is concerned with displays in which imagistic content carries only some of the
representational burden. In this chapter, I am concerned with pure image content—no captions,
shading, and other such symbols with conventional meaning.
 Image Content   271

Finally, I’ll argue that on natural assumptions about sensory images, defla-
tionary theories of content are on shaky ground: it’s hard to understand images as
not expressing content. On the other hand, it is probably true that images do not
express rich sentential content.

11.3. Elements of Image Content

Perceptual experience has image content, but so also do sensory imaging and epi-
sodic memory. In this section and the next, I will lay out some general characteris-
tics of sensory image content as it is found in these types of experience.

11.3.1. PREDICATIVE FEATURE-PLACING

i. Sensory content is about properties or features. In perception, these usually

appear as predicatively belonging to individual sensory objects.10

Traditionally, sensory content was taken as specifying only an array of sense

properties (or features as I will call them) in a space-like arrangement—the sen-
sory “field,” as it is called. But the Gestalt psychologists established that the scene
is broken up into objects. In vision, these are primarily material objects,11 but
there are other types as well—shadows, volumes of darkness, holes, gaps, light
sources, patches of light, rays of light, clouds, and perhaps many more. (Spelke
1990 is an influential and informative review of the development of object per-
ception in infants.) These objects “take their properties with them”: when they
move, their features (or at least some of these features, depending on the kind
of object) change location to match. (A shadow can move and grow and change
shape; i.e., the same shadow can be seen as occupying different places and as
having different shapes and sizes.) Movement and qualitative change are per-
ceived as properties of objects, and objects must therefore be included in percep-
tual content, over and above features and locations. (See section 11.4 below and
Matthen 2005, ch. 12.)
As will become clear in what follows, sensory images are not of particular
objects, i.e., of individuals. On the other hand, perceptual, and possibly recollective,
experience always is of particular objects. As I have indicated, one problem is how.

10
See Matthen (2004) and (2005, ch. 12) for a fuller defense of the predication claim than is
possible here, as well as an argument against Austen Clark’s (2000) thesis that features are merely placed
in visual field-places rather than predicated of individuals such as material objects. A brief argument is
given at the start of section 11.3 below.
11
Xu (1997) argues that material object is the only sortal at work in visual perception. Her
argument is meant to exclude finer sortals, such as man, but she doesn’t consider shadows and the like
for which the rules of overlap and interpenetration are different.
 Imagistic and Possible-Word Content
272   

I shall argue that this particularity comes from outside the image. Sensory images
mark places (identified, not absolutely, but relative to a point of viewing) as occupied
by some object of a certain (sortal) type.12 (I’ll argue in support of this in section 11.5.)
To sum this up, the content of a sensory image can be expressed by a set of
what I call “(predicative) feature-placing structures” of the form <sortal S, feature
F, location L>. Each such structure expresses a situation-type in which an object
of sortal type S instantiates feature F occupies location L (identified relative to a
point of view).13 As mentioned earlier, an experience acquires truth-value when
such a feature-placing structure is asserted of a particular situation, or quantified
over a range of situations.
In sensory images, locations are given from some point of view. Some say that
the point of view is that of the subject, but this is not always true. In “observer
perspective” memory images, the subject images herself from a removed perspec-
tive. For example, you might recall yourself going to kindergarten, viewing yourself
from above or from behind. (It is noteworthy here that the self is not identified
by recognition; rather, the self comes “tagged” as such.) As well, one can imagine
how a scene would look from a particular point of view without imagining oneself
placed at that point of view. The decisive examples are those in which you yourself
appear in the scene. For instance, you can imagine what you look like from behind
if you were to cut your hair differently. Such examples demonstrate the possibility
of imaging a scene, even one that does not have oneself as a part, without imagining
that one is oneself the viewer. Of course, this is not true of perceptual experience.
The “here and now” of such experience depends on the point of view being that
of the perceiver. But one should not generalize from this to other types of sensory
experience.

ii. Image content is logically simple. It expresses no more than a series of

<S, F, L> structures, as above. In particular, it cannot express negation,
disjunction, or quantification. There is no image, for instance, of all tigers
being striped.
iii. The senses represent some places as unoccupied. This could be
understood as a feature-placing structure of a different form <Ø, Ø, L>,
meaning “There is nothing, and no instance of any feature, at L.

Representing a place as unoccupied is perceptually (as well as logically) different

from giving no information about a place. Audition gives you no information
about certain places; you may just fail to hear anything in the space in between

12
My position is that sensory image content is “abstract” in the sense of Tye (2000, p. 62). That
is, it does not involve particular objects. However, I hold that perceptual (and possibly recollective)
experience is object-involving.
13
The complete set of feature-placing structures available to a perceiver at a time would amount
to something like Peacocke’s (1982) scenario content, adding sortals to identify the subjects of feature
attribution, and subtracting the assumption that the point of view is that of the subject.
 Image Content   273

yourself and a source of loud music, without “hearing” silence there. But you can
also hear a room go silent, or silence in a particular place. Similarly, vision gives
you no information about occluded places or about the backs of objects you see
(though it represents these objects as having backs)—but it can also reveal that a
particular place is not occupied.
I’ll return to the Predicative Feature-Placing condition in section 11.4, where
I discuss the objects of each sense modality.

11.3.2. SPATIOTEMPORAL CONNECTEDNESS

i. A sensory image places objects and their features in a unified

spatiotemporal matrix. A unified matrix is a set of spatiotemporal
locations such that every member is spatially and temporally related to
every other.
ii. Perceptual experience presents the subject with a single unified matrix.

The Spatiotemporal Connectedness condition is closely connected to one of Kant’s

main claims in the Transcendental Aesthetic, namely that things perceived as
existing objectively occupy “one space”—i.e., that there is only one spatial matrix
for such things. (See Matthen, forthcoming, for discussion.)
What about free-floating visual features: after-images, “stars” that are seen
when one is struck on the head, floaters, phosphenes, and so on? Similarly, what
about the ringing in the ears of tinnitus? It seems that these figments are not
taken as possessing location in the external world, relative to material objects
and other external things. They are not, moreover, asserted of the here-and-now
situation; that is, a floater is not experienced as part of a situation-type that the
here-and-now instantiates. A floater may be experienced as over to the left of the
visual field, but not as having real spatial location relative to a chair or table over
to the left. While floaters and the like may obscure or occlude seen objects, they
nonetheless don’t seem to occupy the same position any external object, or even
a determinate position in front or to the left of any external object. We show no
tendency to reach for, or around, floaters or after-images. The same holds for
ringing in the ear; it doesn’t seem to come from anywhere, or even to surround us.
Figments of this kind are perceived as self-generated, and they do not seem
to occupy the space occupied by external objects. They do not have the feel of
something perceived, as opposed to merely experienced; a ringing in the ears is
not, for example, experienced as existing here where I am, or now. If Kant is right
(and I think he is), everything one perceives as objectively existing is perceived as
existing in a single space; further, he adds, one can only represent one space. It
seems to follow that such “figments” are not experienced as spatially located, prop-
erly speaking, even though they have some kind of spatiality.
Spatial connectedness plays itself out somewhat differently in different
modalities.
 Imagistic and Possible-Word Content
274   

a. The Spatial Connectedness condition is fairly obvious for vision. Visual

experience is of objects arrayed in a “visual field”. (Opinions about the
nature of this field vary,14 but nobody denies that there is one.) Since the
field is a unified spatial matrix with no gaps, everything in it has location
with respect to everything else. In visual perception, each place in the field
is located relative to the subject.
However, this need not be so for visual imagination. For one can
visualize two objects without imagining how they are related to one
another. For instance, one might image the Eiffel Tower and the Tokyo
Tower side by side in order to compare them, with the Eiffel Tower on
the left. Yet, one might not imagine a specific distance between the two
landmarks. In such a case, I will stipulate that one is entertaining more
than one image at the same time. Each of these images presents a unified
matrix.15
b. Auditory objects are spatially arrayed in normal auditory perception,
in which there is awareness of direction and of distance. Moreover,
everything heard is locatable relative to everything seen, touched, etc. It
is often noted that auditory location is not as precise as visual location;
nevertheless, everything heard is located relative to the subject, if only as
surrounding her, as a voice does in an echo chamber, or residing inside
her head, as does an auditory image generated by stereo headphones. By
contrast, when you hear a voice or a melody “in your head”—i.e., when
you imagine it or remember it—it may not seem to have extension or
location. For even if an imaged piece of music has many voices, they
may not be heard as in any particular direction or at any particular
distance from you, or from each other. But when you hear it live or play
it on the stereo, all of the voices have sensed locations.16 So a melody in
your head is, in a sense, an incomplete image that lacks certain spatial
information.
c. Touch is complicated: when something touches you—for instance, when
you are leaning back in your chair—then if you are immobile, you simply
feel pressure on your back. Of course, parts of the body are spatially
arrayed, so the pressure on your back is positioned in space. However, it

14
Traditionally, the visual field was taken to be two-dimensional. For powerful arguments in
favour of a three-dimensional field, see Austen Clark (1996). See also Matthen (2005, ch. 12).
15
Colin McGinn (2004, p. 23) suggests that imagined spatial matrices may have gaps. His point
seems to be that the two towers exist in a single matrix, which has a gap in it. I am not sure what
gaps are in this context, or why they do not have determinate size. This is why I prefer to say that the
matrices are separate. This is not a big point.
16
Interestingly, music perception seems to present a kind of non-spatial motion (Charles
Nussbaum, forthcoming). That is, it is very natural to think of harmonic progressions, rhythmic lines,
etc. as moving. Most likely, this is a cross-modal effect, as is the perception of higher frequency notes
as spatially higher.
 Image Content   275

is anchored to the coordinate system of your body, not that of external

space. Thus, even if you are aware that you are moving—e.g., when you
are walking or driving in your car—pressure on your back is not sensed
as moving. By contrast, when you haptically explore an external object
by stroking or feeling it, the object you are touching is externalized,
and tactile qualities—hardness, coldness, flatness, etc.—are sensed as
belonging to it.

As J. J. Gibson writes:
In active touching and looking the observer reports experiences [that] corre-
spond to the environment instead of to the events at the sensory surface. The
experiences noted with passive stimulation can scarcely be noticed, if at all.
(1962, pp. 489–490)
So touch seems simultaneously to deploy two coordinate systems, though (as the
transition between passive and exploratory touch shows) the two systems repre-
sent the same spatial matrix.

d. Interestingly, haptic exploration has multimodal impact: flavors get

attached to objects in the mouth by just the above-mentioned process of
haptic exploration using the tongue and mouth. It is the cracker in one’s
mouth that seems to be salty, though presumably the salt is distributed
throughout the mouth by being dissolved in saliva.17
e. Finally, smell: given a stationary subject, the objects of olfactory
perception have undifferentiated spatial location (“Here!”) or (according
to some) no location at all (because it is held that to a stationary subject,
they are merely sensations: see Christopher Peacocke 198318 and Lycan
2000). However, given movement, manipulation, and sniffing—that
is, by exploratory or active smelling—odors can be precisely located as
emanating from particular external objects,19 which are in turn located
with the aid of vision and touch. Like touch, smell has an active mode.
This active mode gives awareness of a different field of locations than the
passive mode.

17
Flavor is a very complex case, since its components come from a variety of receptors, some in
the tongue, some in the nose, some in the face. All of these are synthesized into a single experience and
“referred” to the mouth, or to the object in the mouth. (After tastes are in the mouth, but obviously
not in any object.)
18
Peacocke: “A visual perceptual experience enjoyed by someone sitting at a desk may represent
various writing implements and items of furniture as having particular spatial relations to one another
and to the experiencer . . . A sensation of small (sic), by contrast, may have no representational content
of any sort, though of course the sensation will be of a distinctive kind” (5).
19
Batty (2010) distinguishes between smell, which is an olfactory property of the perceiver’s
surroundings, and odor, which is a chemical substance in the air.
 Imagistic and Possible-Word Content
276   

11.3.3. PROPERTY CODING

Sensory appearances are representations of properties, for instance, color, pitch,

loudness, and so on. Colors have characteristic looks; notes sound a particular
way; and so on. Sensory qualities are recognized by how they appear.
Here are some important constraints on the “vocabulary” of property presen-
tation in sensory images:

i. Unique coding Each modality presents each determinate sensory property

P in only one way. (Some features come in bundles. For example, there
is a look not of a square simply, but rather of a square as viewed from a
particular angle.20) There are no “synonyms” within a modality: the visual
presentation of each color is the same; the visual presentation of a square
from a given oblique angle is the same. And so on.
ii. Non-Ambiguity Sensory experiences of determinate features (or feature
bundles) are unambiguous. Thus, the phenomenal character of the visual
experience of one shade of blue cannot be the same as that of another.21
iii. Availability A subject cannot experience a sensory property without
knowing what sensory property she is experiencing.

Availability is somewhat reminiscent of Mark Johnston’s (1992) Revelation prin-

ciple. Johnston says that the intrinsic nature of a sensory feature is fully revealed
when one experiences it. I claim only that one cannot experience a sensory prop-
erty without conclusively identifying that property, not that one thereby knows
anything about the property. Finally:

iv. Similarity coding Sensory presentations enable subjects immediately to

recognize similarity within each sensory property-type such as color or
pitch (i.e., each determinable).22
Similarity Coding is stated by Paul Boghossian and David Velleman (1989): “One
can tell on the basis of visual experience alone whether two objects appear simi-
larly coloured.” Color is a sensory feature, and the same holds, mutatis mutandis,
of other sensory features.

20
It is common among philosophers to collapse these into the two-dimensional projection of
a shape—they say that a square viewed at an angle looks like a trapezoid—but see Matthen (2010b).
21
Compare Boghossian’s (1994) condition of Transparency: (a) If two of a thinker’s token thoughts
possess the same content, then the thinker must be able to know a priori that they do; and (b) If two of
a thinker’s token thoughts possess distinct contents, then the thinker must be able to know a priori that
they do. I do not endorse this condition in its full generality.
22
The Property Coding condition excludes the kinds of devices suggested by Camp 2007 for
map-like representations: for example, “A black (or other fully-saturated) icon [c]‌ould represent
certainty that the relevant object/property is at that location” (163). This would entail that black would
represent both blackness and certainty.
 Image Content   277

11.3.4. PART-WHOLE COMPOSITION

The senses present their objects as spatiotemporally extended only by present-

ing the spatiotemporal parts of these objects located with respect to each other as
they are in the object (or sometimes as a two-dimension projection of that array).
For example, they represent a triangle by representing the sides and angles of the
triangle; they represent the sides by representing the vertices and the points in
between. (When a part of such a line is occluded, it is represented “amodally.”)
Susan Carey (2009, p. 458) puts the above condition in the following much
simpler way: “parts of the representation correspond to the parts of the entities
represented.” This formulation works fine with pictures or maps. Carey’s example
is a picture of a tiger: “The head in the picture represents the head of the tiger.” The
intended contrast here is with the word ‘tiger’, no part of which represents a part
of any tiger.
It is less clear what counts as the representer or vehicle—the analogue of the
picture—when we see a tiger. Perhaps an activation pattern of neurons in the
brain.23 It is also unclear whether the parts of such a neural activation pattern
show the same part-whole composition. That is, if N is a state of the nervous
system that imagistically represents a tiger, then must there be a contiguous part
of N that represents the head of a tiger? And must this part be spatially related
to the part of N that represents the tail in the way that tiger heads and tails are
spatially related?
We can, however, say this: if N is any kind of image of T (mental or graphic),
then N must represent not only T but at least some of the spatial parts of T—i.e.,
those that can be sensed from the perspective that the subject occupies. In this
respect, the mental image of a tiger will still contrast with the word ‘tiger’.

11.4. More about Predicative Structure in Image Content

The feature-object structure of perceptual content is, again, relatively obvious for
vision. Here is a simple pair of visualizations that demonstrates it.
First, imagine a row of transparent objects. Imagine that starting from the left
and continuing towards the right, each successively turns blue and then back to
transparent, perhaps by an internal light being turned on and then off. Imagine
that this is done slowly enough that the display does not fuse, and is not seen as
a single moving object. Call this the Shifting Blue phenomenon or SB.
Now imagine just one of these objects in its blue state moving from left to
right. Call this the Moving Object phenomenon or MO.

23
Carey: “I assume that representations are states of the nervous system that have content that
refer to concrete or abstract entities (or even fictional entities), properties, and events” (2009, p. 5).
 Imagistic and Possible-Word Content
278   

Of course, there is more than one difference between SB and MO: for instance,
there are spatial gaps in SB and a smooth trajectory in MO. However, the phe-
nomenological difference between SB and MO is not limited to the sequence of
blue-occurrences in various places. It includes that a single moving blue object
is presented in MO, but not in SB. (Actually, if SB is speeded up, it will look just
like MO: the gaps will disappear.) This difference between the two presentations
is not captured by all the feature-place pairs in each. For since the idea of move-
ment is one of a single object taking its properties with it, this difference must be
accounted for by positing that blue is predicated of one object in the phenome-
nology of MO, and of many in SB. (Objects need not be material objects: colors
may appear to belong to fringes, shafts of light, shadows, vapors, auras, etc.)
Motion perception (and also perception of change) demonstrates that visual
content represents features as belonging to movable objects, mainly material
objects. What about the other senses?
i. In audition, the smallest objects are sounds. Sounds are events that
cause the waveform vibrations in the air that excites the ears (Casati
and Dokic 1994; Pasnau 1999; O’Callaghan 2009): events such as a bow
being drawn across a string or the vibration of somebody’s vocal chords.
(More specifically, they are the last cause of such waveform vibrations
that are not themselves waveform vibrations: Matthen 2010a, pp. 82–83.)
Auditory qualities—loud, soft, high, low, etc.—are transmitted by waves
in the acoustic medium, and correlate with properties of those waves.
Nevertheless, what we hear is a property of the event that causes the
waveform. Audition attributes features to located events: for instance,
a man is playing the drum loudly there. This attributes a property to the
event that consists of the man beating a drum; it does not attribute a
property to the sound waves that the beaten drum emits.
Audition also identifies composite sounds: melodies, phonemes,
speech streams. These are composed of sounds, but the auditory system
recognizes them as single entities (Matthen 2010a). These objects have
properties that are detected by audition: a falling tone, a diminuendo, a
harmonic resolution, etc. are such properties.
ii. In active touch, the primary objects are surfaces of objects, but when the
subject can either see the object he is touching, or can actively touch it
by stroking, feeling, and palpating, surfaces appear voluminous (Gibson
1962). When vision is active alongside touch, the objects to which tactile
qualities are attributed are often those defined by vision, both with respect
to their location and with respect to their exact contours. (In passive
touch, as discussed earlier, the objects are part of the body.)
iii. The objects of olfaction are odors, distributions of volatile chemicals in
the atmosphere, but they are not, of course, presented as such. Odors
have extended location, though it takes movement to detect this. Odors
 Image Content   279

are the bearers of smells, which are olfactory qualities such as sweet
and burnt.24 One can also smell material objects, but this takes active
olfaction: moving around, picking things up, and sniffing. In such cases,
again, the bearers of olfactory qualities are defined by other senses—
vision and touch, in particular.
iv. Finally, flavors are typically attributed to objects in the mouth, which are
located by touch. Analogously to vision, it is possible to have free-floating
flavors, such as after-tastes, located in the mouth, but with flavor too these
are normally not mistaken for flavors that reside in a sensed object.

11.5. Limitations of Image Content

Though images and sentences both express propositions, there are propositions
that cannot be expressed by images. For example, as I said earlier, negation and
universally quantified propositions cannot be expressed by a sensory image.
Now, some sententially expressed accounts of sensory content seem to violate
the above conditions on image content. For example, Susanna Siegel (2006) claims
that when a subject S is looking at an object o that appears to be external to her, S’s
visual experience has the following content:
(PC) If S substantially changes her perspective on o, her visual phenome-
nology will change as a result of this change. (2006, p. 358)
PC violates every condition on images laid down in the preceding section. For
instance, PC seems to imply that one of the things visual experience presents as
content is its own “visual phenomenology”: that is, according to PC, visual expe-
rience does not merely possess phenomenology, but reveals something about this
phenomenology. But this violates the Property Coding condition because there is
nothing that visual phenomenology looks like. (Rather, visual phenomenology is
how objects and features look.) Again, it is unclear how a visual experience could
represent the “will change if ” conditional in PC. Without wishing to question
that some conditionals are implied by visual states—for example, the conditional
proposition that a moving object will arrive at a particular place if it continues
to move in the direction it is now moving—it is unclear how conditionals can be
visually presented, in particular the conditional stated in PC. And where is this
conditional? Perspective-change has spatiotemporal parts—but how are they rep-
resented? For these reasons, it is hard to see how an image could represent PC.
Elsewhere Siegel says, more simply, that

24
I take the distinction between odors and smells from Clare Batty (2010); however, she argues
that the object of olfaction is a “know-not-what”.
 Imagistic and Possible-Word Content
280   

in the typical experiences of object-seeing, objects are presented as being den-

izens of the external world rather than as mind-dependent entities of some
sort. (2006, p. 374)
This better respects the conditions on images articulated above. Externality is, one
could say, a feature that attaches to a located object. Somebody could doubt that
externality looks like anything—“What is the characteristic look of externality?”
one might ask, “What color is it?” (See Hume’s argument in section 11.6, below.)
But I don’t see this as much of an issue: clearly some things look as if they are in the
external world, and others do not. After-images, for example, do not. (This is what
Siegel relies on.) The contrast between PC and the above formulation illustrates
how some things can and some things cannot be expressed by images.
Much the same criticism applies against John Searle’s famous account of
visual content:
I have a visual experience (that there is a yellow station wagon there and that
there is a yellow station wagon there is causing this visual experience).” (1983,
p. 48)
This requires that visual experiences be represented by visual experiences. But visual
experiences are not visibilia; they are nowhere in the visual field. It requires, more-
over, that a causal interaction that has no location in the visual field be represented.

11.6. Shared Image Content

Now, consider these cognitive states:

P (Perception). You are standing in the rain: you see, hear, and feel it.
M (Imaging). You imagine standing in the rain: you imagine how it
would be to see, hear, and feel it.
R (Iconic recollection). You recall standing in the rain: you recall how it
looked, sounded, and felt.
P, M, and R are states of a perceiver that involve similar sensory images. They are
about similar scenes; as such, they share content. And in principle, anything one
can perceive or imagine or recall can constitute the content of one of the other
kinds of act. This gives us the following:
Shared Content Principle Any perceived feature-placing structure could figure
in the image content of perception or recollection or imagination.
There are important differences among the above states too, as we have noted. P
is here-and-now directed relative to the perceiver, and gives her an unmediated
reason to believe that its content holds of the here and now. (It may be relevant
that P is steadier than M or R in the sense that it is more stable, and requires no
effort to maintain in a steady state—however, this is not part of the image content
 Image Content   281

of the P-experience.) R is past-directed—it gives the perceiver reason to believe its

content, but in the past tense. It may well be indeterminate as to the exact time of
occurrence, and place-directed only by captioning or recognition. (See 11.1.2.iii‒v
above.) M does not have any indexical reference to place or time; it does not give
the subject reason to believe anything about the world external to her own mind.25
One might ask: “How can the import of P, M, and R be different, given that
their imagistic content is the same?” Must there not be some importantly different
element of content that accounts for the differences noted earlier? If this is so, does
it show that there is a non-imagistic kind of content that the senses deliver, con-
tent that distinguishes perception, imaging, and memory? This runs counter to
most conceptions of how the senses operate, and this raises problems. I’ll return to
these questions later (though it is not the purpose of this chapter to answer them
authoritatively). But first let me say why I think it is difficult to account for these
as differences in image content.
Here is an argument using the Shared Content Principle to reach the conclu-
sion that the here-and-now character of perception is not a part of image content.
Shared Content Argument I
By the Shared Content Principle, if F can be imagistically attributed to an
object at an absolutely identified location in perceptual experience, then it can
be so attributed in imagination.
But imagination does not represent absolute locations. (Condition
I.2.iii‒iv).
Therefore the definite location of feature F is not imagistically repre-
sented in perception.
It follows that the here and now character of perceptual experience and the
past-regarding aspect of episodic memory are not imagistically represented.
I am not, of course, denying that perceptual states reveal the absolute locations
of objects in the subject’s vicinity. In fact, I mean to assert that they do. My claim
is rather that they do not do so by means of their image content, for if they did,
then the image content would continue to do so when reproduced in imagination.

11.7. Hume on the Expressive Limitations of Images

The position I have articulated on the expressive limitations of images has inter-
esting historical counterparts. Consider Hume’s argument that existence is not
imagistically represented:
I do not think there are any two distinct impressions, which are insep-
arably conjoin’d. Tho’ certain sensations may at one time be united, we

25
See McGinn 2004 for other such differences between perceptual and non-perceptual sensory
experiences.
 Imagistic and Possible-Word Content
282   

quickly find they admit of a separation, and may be presented apart.

And thus, tho’ every impression and idea we remember be consider’d as
existent, the idea of existence is not deriv’d from any particular impres-
sion. (Treatise I.II.vi)
Now, as is well known, Hume is committed to some extreme positions concerning
content. In particular, he is committed to:
Berkeley’s Principle
If my being in a sensory state S is compatible with the falsity of p, then p is not
the content of S.
Berkeley uses this principle to argue, for example, that visual states never rep-
resent distance. Hume uses it to argue that there is no idea of a substance, or of
causation (except as conjunction), and so on. However, the above argument is
not committed to any such premise. It rests only on certain assumptions about
image content, and these are pretty much as those articulated in section 11.1,
above.
In the above passage, Hume’s premise is that image content (ideas and
impressions) can express two distinct properties only by two separable ideas.
Now suppose that I perceive something—for instance, a person sitting across
from me—as existent. It is possible to imagine that that very person does not
exist. In order to do this, however, I am not obliged to imagine the person as
perceptually different in appearance. Thus, it is possible to subtract existence
without subtracting any perceptually derived idea. This shows that the idea of
existence is “not deriv’d from any particular impression.” At this point, Hume’s
empiricism kicks in. Since there is no separable idea of existence in what we
perceptually experience, Hume concludes that existence is not a distinct fea-
ture. “The idea of existence, then, is the very same with the idea of what we
conceive to be existent,” he says. In other words, we conceive of everything as
existent.
It is not clear to me that Hume is right in this last conclusion. It could well
be that in order to imagine that the person sitting across from me is not real, one
does have to subtract the perceptually derived idea of externality. (This is presup-
posed by Kant’s notion that when things are seen as existing in space, they are
seen as existing independently; it was also my gloss on Susanna Siegel’s argument
in section 11.5 above.) However that might be, I agree with Hume’s insistence
that images, i.e., ideas in a spatial matrix, are limited in expressive power. My
method for detecting these limitations is somewhat similar to his. My method is
to look for shared content. If two sensory acts are directed to the same sensory
image, but differ with respect to some commitment, p, then p cannot be part
of the image content of these sensory acts. And if expressing p would require
contravening the characterization of image content in section 11.1 above, then p
cannot be expressed by any image.
 Image Content   283

11.8. Outside the Image

Let us return now to perceptual experience. Here is a second Shared Content

Argument about experience, again quite Humean in spirit.
Shared Content Argument II
Suppose that perceptual experience P imagistically expresses scene S.
P conveys to the subject that the scene that presently surrounds her is of
type S.
A memory experience and an imaginative experience that express S
would not convey to the subject that the scene that presently surrounds her
is of type S.
Therefore, the imagistic content of P is not what conveys to the subject
that the scene that presently surrounds her is of type S.
The above argument shows that the sense that perceptual experience expresses
something about the present occurs “outside the image”.
I have argued elsewhere (Matthen 2005, ch 13; 2012) that vision is about
individual objects because the visual system sub-personally furnishes the per-
ceiver with egocentric coordinates for these objects, and that these coordinates
enable the perceiver to attend to and interact with these objects. Since attention to
an object is required for the formation of beliefs about that object, these egocentric
coordinates serve the epistemic role of vision. These coordinates are not a part of
the visual image, but are rather furnished to a system that controls gaze and atten-
tion independently of the conscious visual image. They are, moreover, absent from
non-perceptual vision—memory and imagination.
Here, I want to offer a simpler model. This simpler model has the disadvan-
tage that it relies on a substantial amount of approximation and idealization. Its
main virtue is that it generalizes to all of the senses, and that it fits well with the
here-and-now characterization of perceptual experience offered in this chapter.
By the definition given earlier, scenes are spatiotemporal distributions of fea-
tures in sensory objects. The spatial relations of these sensory objects are defined
by the distances between them. For it is a fact of geometry that a set of elements
with fixed pair-wise distances forms a rigid structure the shape and size of which
remains fixed under rotation and movement from one place to another. It follows
that as long as you estimate the pair-wise distances for a set of objects correctly,
you know how they are arrayed, regardless of your perspective. Let’s call this a
scene.
With this in mind, imagine that the perceiver is herself one of the objects
in the scene S that she senses. She fixes a position in space indexically: here. The
scene that she experiences is a rigid structure with place-holders that can be laid
on top of objects in the three-dimensional space around her. Every object in the
scene S has an absolute location fixed by the perceiver’s position. A given element
 Imagistic and Possible-Word Content
284   

of her perceptual experience, Fxi is true of an object a, if a occupies L in her scene

S and Fa.
An important point to note is that though the locations in the scene are ego-
centrically located, the model that the perceiver builds up is objective, or allocentric,
in the following way: since the distances between objects are given, and since these
distances specify a rigid solid, the perceiver is able to distinguish between changes
of location in the scene and changes of her own position. When she moves from
one place to another, the orientations and distances of things change relative to her
own position. On the other hand, the distances between objects stay the same, and
thus the objective disposition of objects remains constant. If she stays in the same
place, but objects around her change their positions, then, once again, the orienta-
tions and distances of things may change relative to her own position; however, the
inter-object disposition changes, and this marks change outside her own world.
The content of a perceptual experience is determined by two factors. The first
is the scene, which is a specification of how objects and their features are located
relative to each other. This specification may be incomplete; not all the qualities
of every object may be included in the perceptual scene. The second factor is an
egocentric location. Objects in the perceived scene are identified as individuals by
their location relative to the perceiver. (The perceiver may, of course, misperceive
these locations at any given moment, but multimodal processes of active sensory
investigation can correct such errors: see Matthen, forthcoming.)
I said earlier that this model is somewhat idealized, since it assumes that
the pair-wise distances between sensed objects is presented to the perceiver. The
information that she possesses may be incomplete in this regard. She will need to
build it up by moving around and actively exploring the scene. Moreover, a matrix
of pair-wise distances will not differentiate among Kant’s incongruent counter-
parts (e.g., mirror images). The point that I am trying to make clear here is simply
that absolute locations can only be presented in some such way from outside the
image content of a sensory state.

11.9. Deflationary Views of Content

It is hard to deny some form of image content for perception. It is certainly

possible to contest some of the details given in earlier sections, but at the very
least it cannot be denied that the senses non-inferentially afford us informa-
tion regarding sensory property-instances arranged in a perceptual field. Many
of the details of conditions 2–4 of section 11.2 can be contested; however, it is
difficult, with any degree of plausibility, to reject these conditions out of hand.
It seems uncontestable that sensory awareness involves the presentation of
feature-placing structures—that is of objects and feature instances distributed
in space. The protocol statement view seems to capture at least a minimal spec-
ification of content.
 Image Content   285

Deflationary views of content hold that perceptual experience is incapable of

representing propositions. According to what I shall call the No Content View,
perceptual states express nothing at all. (Epicurus had a view like this, but see now
Travis 2004, Gupta 2006.) According to Naïve Realism, perceptual states do not
represent objects: rather, they are relations to mind-independent objects (Martin
2002, 2006). In conclusion, I want to look briefly at these two views, in order to
examine how they might be modified to accommodate image content.
Before we continue, let me summarily dismiss the Epicurean view that
perceptual experience is merely an impressed image, merely a causal trace of
external objects that the perceiving subject may utilize to draw conclusions about
the external world, but in no way a state that can be assessed as true or false.
Of course, perceptual experience is the causal trace of an external object. What
Epicurus (understandably) failed to recognize, however, is that perceptual expe-
rience is not merely an impressed image. It is rather the result of sensory systems
(rather than the person of the perceiving subject) extracting data from the stimu-
lation of sensory receptors. (See Matthen 2005 for an extended discussion.) Thus,
for example, the impression of distance and three-dimensional shape is extracted
from the retinal image using a variety of cues, including binocular displacement,
brightness gradients, texture, and other such characteristics of the retinal images.
Perceptual experience is how sensory systems record and make known the results
of content-extraction from these impressed images.
This said, one should acknowledge a core truth in the Epicurean position.
Sensory systems extract content from impressed sensory images in a relatively
inflexible way. The perceiving subject needs to take account not only of percep-
tual experience, but also of other evidence, some of which has to be weighed up
in ways that are not pre-determined. Suppose that some surface in a department
store looks electric blue, but that you have read about the amazing perceptual illu-
sions that this store uses in its interior design. By how much will you discount
the appearance of blue? This, unlike sensory processing, is indeterminate and a
matter of free choice. Nevertheless, perceptual experience should be construed
as conveying a message, minimally a message about “protocol” qualities and their
predication of located objects. This is the best way to conceptualize the interaction
between experience and epistemic evaluation.
Given how well established sensory processing is, we should treat contempo-
rary deflationary views as fully informed about at least the outlines. Anil Gupta’s
(2006) No-Content view of experience, though reminiscent of Epicurus, is based
on an epistemic distinction between perceptions and judgments. In assessing his
views, it is charitable to grant him full knowledge of recent cognitive science of
perception. In that spirit, let us look at what he says.

The given in my experience of, say, looking at a ripe tomato does not contain
judgments such as “That is a tomato,” “That tomato is red,” and “I am seeing a
tomato.” It is plain on reflection that my visual experience, when considered
 Imagistic and Possible-Word Content
286   

in isolation, does not entitle me to the judgement that the object before me is
a tomato.. .
The second consideration is that experience is passive, and it is always a
good policy not to assign fault to the passive. . . . When I have what is called
a “misleading” experience, experience has done nothing to mislead me. The
fault, if any, lies with me and my beliefs—beliefs for which I am responsible.
When on a foggy day, I take a pillar to be a man, it is not my visual experience
that tells me that there is a man before me; the experience is ill-equipped to do
such a thing. I form the belief that there is a man. (2006, pp. 185–186)

It is difficult to make philosophical sense of these passages. Indeed, it is hard to

resist the impression that they are intended as pseudo-Wittgensteinian goads that
lead one, in irritation, to glimpse some deeper truth—and not as a philosoph-
ical argument at all. However this may be, they do not bear scrutiny on a literal
construal.
Gupta says that perception does not entitle the subject to a judgment.
(Presumably he means that it does not so entitle the subject all by itself.) This is
true enough—that a tomato looks red is often not sufficient reason for judging that
it is red. This may be a reason for saying that experience does not contain a judg-
ment. But how does it show that there is perceptual experience does not contain,
or represent, any proposition? Again, it is true that “I form the belief.” But this does
not gainsay the fact that perceptual experience gives me reason to support specific
beliefs and actions and undermine others. One could hold that it is the perceiver’s
epistemic responsibility to take into account the reasons that might tell against
a given perceptual experience. Thus, it is unclear why the claim that perceptual
experience does not contain judgment carries No-Content force.
What is one to make of the injunction that one ought not to assign “fault” to
the passive? (Is “passive” a remnant of the impressed image theory?) The fact that
judgment is active tells me nothing about the veracity of perception, or the lack of
it. Finally, even if perceptual experience did tell you something, how it still would
not follow that you are entitled to believe it: it could still be your fault if you did.
Now, it could be that Gupta is making the mistake, tackled in section 11.3,
of denying that an image expresses a proposition. More charitably, he is perhaps
pointing to a gap between perception, which is imagistically presented, and judg-
ment, which has sententially expressed content. And, as we have noticed, there are
genuine puzzles about this gap.
If this is the problem, then one promising approach is to adopt a view
like that of M. G. F. Martin (2010). Martin considers statements such as “That
seems red,” and “That looks like an A380.” His claim is that these are “compara-
tive,”—they compare the perceptual experiences that occasion these sentences to
appearances of the states of affairs mentioned. I would parse Martin’s suggestion
as follows: one experiences a certain image and recognizes a certain similarity
between it and images of situations in which the contained sentences are true.
 Image Content   287

Extending Martin’s suggestion, sentential content can be explicated as follows:

Sentence S (partially) expresses the content of a sensory experience E if rele-
vant worlds in which S is true resemble worlds in which the image content of
E is satisfied by the relevant situation.
For example:
“That is an A380 taxiing out for take-off ” partially expresses the content of
my perceptual experience if worlds in which there is an A380 taxiing out for
take-off resemble the image content of my experience when these worlds are
looked at from my point of view.
Martin’s proposal helps address the problem of the relationship between sensory
experiences, which are imagistic in nature, and linguistic reports of these expe-
riences. The reports capture what the subject wishes to convey about the imag-
istic representation by means of a comparison. It is plausible to hold that reports
like “That looks like an A380” and the underlying sensory image is that the latter
resembles an A380 situation. It is reasonable to think that this similarity is ap-
parent to the perceiving subject only as a result of learning. And it follows that a
report like the above cannot be produced without the intrusion of a background
theory. But this does not gainsay the fact that the image content on which they
report is (a) propositional in character, and (b) theory-independent. Gupta’s posi-
tion wrongly takes the theory-ladenness of sentential reports of image content for
the theory-ladenness of image content.26

X. Concluding Remark about Naïve Realism

In this chapter, I began by laying out certain characteristics of image content.

These characteristics force certain expressive limitations on vehicles of image con-
tent. Since perceptual states have such content, but are not expressively limited,
I inferred that they must have meaning outside the image. This raises difficulties
for some deflationary views of content.
Let me close with a remark concerning Naïve Realism (Snowdon 1981,
1990; Martin 2006), the position that experiences are simply relations to
mind-independent objects. The usual objection put to naïve realists is the argu-
ment from hallucination, which goes like this:
A perceptual experience of (a mind-independent object) O is phenomenally
identical to some hallucinatory experience of O.

26
Gupta’s view is similar to Charles Travis’ (2004) view. Travis’ argument is considerably more
complex and nuanced, but in the end, I think it too treats sentential expressions of perceptual content
as if they were independent of underlying image content.
 Imagistic and Possible-Word Content
288   

No hallucinatory experience of O is a relation to O.

Therefore, a perceptual experience of O is phenomenally identical to some
experience that is not a relation to O.
Assuming that phenomenal character individuates experiences, it follows that it is
possible to remove the object of a perceptual experience while leaving the experi-
ence unchanged. This is precisely what naïve realists deny.
I want to offer a modified version of the above argument.
Shared Content Principle
Any perceived feature-placing structure could figure in the image content of
perception or recollection or imagination.
States of recollection and imagination are not relations to their objects; it
is possible to remove their objects without altering their image content.
Therefore, a feature-placing structure can figure in the image content of
a perceptual experience even when the object of that experience is removed.
The basic idea here is that perceptual image content does not depend on actually
existent objects and is not therefore relational. It seems to me that this conclusion
is unfriendly to Naïve Realism.27

References

Batty, Clare (2010). A representational account of olfactory experience. Canadian Journal

of Philosophy, 40, 511–538.
Boghossian, Paul (1994). The transparency of mental content. Philosophical Perspectives, 8,
33–50.
Boghossian, Paul A., & Velleman, J. David (1989). Colour as a secondary quality. Mind, 98,
81–103.
Camp, Elisabeth (2007). Thinking with maps. Philosophical Perspectives, 21, 145–182.
Carey, Susan (2009). The Origin of Concepts. New York: Oxford University Press.
Casati, Roberto, & Dokic, Jerome (1994). La Philosophie du son Chapter 2: Distinction entre
les sens. Nîmes: Chambon.
Clark, Austen (1996). Three varieties of visual field. Philosophical Psychology, 9, 477–495.
Clark, Austen (2000). A Theory of Sentience. Oxford: Clarendon Press.
Crane, Tim (2009). Is perception a propositional attitude? Philosophical Quarterly, 59,
452–469.
Gendler, Tamar S., & Hawthorne, John (Eds.) (2006). Perceptual Experience.
Oxford: Clarendon Press.
Gibson, J. J. (1962). Observations on active touch. Psychological Review, 69, 477–491.

27
I am very grateful to Chris Gauker for detailed comments after a careful reading of the whole
manuscript.
 Image Content   289

Gupta, Anil (2006). Experience and knowledge. In T. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 181–204). Oxford: Clarendon Press.
Johnston, Mark (1992). How to speak of the colors. Philosophical Studies, 68, 221–263.
Lycan, W. (2000). The slighting of smell (with a brief note on the slighting of chemistry).
In N. Bhushan & S. Rosenfeld (Eds.), Of Minds and Molecules: New Philosophical
Perspectives on Chemistry (pp. 273–290). New York: Oxford University Press.
Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17, 376–425.
Martin, M. G. F. (2006). On being alienated. In T. Gendler & J. Hawthorne (Eds.), Perceptual
Experience (pp. 354–410). Oxford: Clarendon Press.
Martin, M. G. F. (2010). What’s in a look? In B. Nanay (Ed.), Perceiving the World (pp.
160–225). New York: Oxford University Press.
Matthen, Mohan (2004). Features, objects, and places: Reflections on Austen Clark’s Theory
of Sentience. Philosophical Psychology, 17, 497–518.
Matthen, Mohan (2005). Seeing, Doing, and Knowing: A Philosophical Theory of Sense
Perception. Oxford: Clarendon Press.
Matthen, Mohan (2010a). On the diversity of auditory objects. Review of Philosophy and
Psychology, 1, 63–89.
Matthen, Mohan (2010b). How things look. In B. Nanay (Ed.), Perception and the World
(pp. 226–252). New York: Oxford University Press.
Matthen, Mohan (2012). Visual demonstratives. In A. Raftopoulos & P. Machamer (Eds.),
Perception, Realism, and the Problem of Reference (pp. 43–67). Cambridge: Cambridge
University Press.
Matthen, Mohan (forthcoming). Active perception and the representation of space. In D.
Stokes, M. Matthen, & S. Biggs (Eds.), Perception and Its Modalities. New York: Oxford
University Press.
McDowell, John (1994). Mind and World. Cambridge, MA: Harvard University Press.
McGinn, Colin (2004). Mindsight: Image, Dream, Meaning. Cambridge, MA: Harvard
University Press.
Nussbaum, Charles (forthcoming). Musical perception. In M. Matthen (Ed.), The Oxford
Handbook of the Philosophy of Perception. Oxford: Clarendon Press.
O’Callaghan, Casey (2009). Sounds. Oxford: Clarendon Press.
Pasnau, Robert (1999). What is sound? Philosophical Quarterly, 49, 309–324.
Peacocke, C. (1983). Sense and Content: Experience, Thought and their Relations.
New York: Oxford University Press.
Peacocke, Christopher (1992). A Study of Concepts. Cambridge, MA: MIT Press.
Pollock, John (1974). Knowledge and Justification. Princeton: Princeton University Press.
Siegel, Susanna (2006). Subject and object in the contents of visual experience. Philosophical
Review, 115, 355–388.
Siewert, Charles P. (1998). The Significance of Consciousness. Princeton: Princeton University
Press.
Snowdon, Paul (1981). Perception, vision and causation. Proceedings of the Aristotelian
Society, New Series, 81, 175–192.
Snowdon, Paul (1990). The objects of perceptual experience. Proceedings of the Aristotelian
Society, Supplementary Volume 64, 121–166.
Spelke, Elizabeth S. (1990). Principles of object perception. Cognitive Science, 14, 29–56.
Travis, Charles (2004). The silence of the senses. Mind, 113, 57–94.
 Imagistic and Possible-Word Content
290   

Uebel, Thomas E. (1992). Rational reconstruction as elucidation? Carnap in the early proto-
col sentence debate. Synthese, 93, 107–140.
Uebel, Thomas E. (1993). Neurath’s protocol statements: A naturalistic theory of data and
pragmatic theory of theory acceptance. Philosophy of Science, 60, 587–607.
Xu, Fei (1997). From Lot’s wife to a pillar of salt: Evidence that physical object is a sortal
concept. Mind and Language, 12, 365–392.
Xu, Fei (2007). Sortal concepts, object individuation, and language. Trends in Cognitive
Science, 11, 400–406.
12

What is the Content of a Hallucinatory

Experience?
Michael Tye

Sometimes our senses mislead us. Sometimes things are as they appear. One
natural way to take these anodyne remarks is as indicating that our perceptual
experiences are sometimes inaccurate, sometimes accurate. Perceptual experi-
ences, then, have accuracy conditions. Where there are accuracy conditions,
there is representational content. Historically, the idea that perceptual expe-
riences have representational content was not the usual one. The most pop-
ular view was that content resides further downstream in the beliefs that are
formed on the basis of perceptual experience. Perceptual experiences them-
selves are ‘blind’. Sometimes, this view is cashed out further by holding that
one who undergoes a perceptual experience is presented with an appearance
(or sense-datum); at other times, it is held that undergoing a perceptual expe-
rience is a matter of being appeared to in a certain way. On both these views,
there is no accuracy or inaccuracy at the level of perceptual experience. Thus,
there really are no purely perceptual illusions or hallucinations. When one sees
a blue object, say, that looks green, it is not that one’s experience misrepresents
the object as green, but rather that one is presented with an appearance that is
green. This appearance is the ‘look’ of the object—and the object genuinely has
that look, so there is no inaccuracy (and no need to admit that the experience
has a representational content). If one forms the belief that the object is green,
then mistake arises there: the object is not as the belief represents it to be. But
there is no mistake in the experience itself.
This chapter assumes that such nonrepresentational approaches to percep-
tual experience are mistaken. Views that countenance appearances (or ‘looks’)
as real entities have to face a range of very puzzling questions; and adverbialist
approaches need to construct highly complex and contrived adverbs to account
for the full range of perceptual scenarios. What has motivated these views at least
for some is the thought that there cannot be content without concepts, and so
content only really arises at the level of thought or belief. The present chapter also
assumes that this is a mistake. There is no difficulty in making sense of the idea 291
 Imagistic and Possible-Word Content
292   

that perceptual experiences are nonconceptual representations, representations

that do not involve the exercise of concepts.
So, sometimes our senses themselves really do mislead us. Sometimes, our
perceptual experiences really are inaccurate, just as sometimes our beliefs are. This
happens when we hallucinate. Here is an example. Keith has just taken a halluci-
nogenic drug (see http://stylefrizz.com/200803/keith-richards-for-louis-vuitton-
ad-campaign-2008-by-annie-leibovitz/). A few minutes earlier, he was occupied
with the beginning of H. H. Price’s well-known book on perception (shown on the
suitcase in the picture, we may suppose).1 The combined effect of these activities
is that Keith is now hallucinating a ripe tomato. This is not a de re hallucination.
There is no particular tomato located elsewhere out of Keith’s vision such that he
is hallucinating that tomato as being before him. Keith is hallucinating a tomato
without there being any particular tomato that he is hallucinating. In this case,
Keith’s experience is inaccurate. The world is not as it seems in his experience.
How can this be? If perceptual experiences have representational content, just
what is the content of Keith’s experience in this case?2
Consider a case of veridical perception. Suppose that Keith is seeing a ripe
tomato. The view of naïve realism is that Keith sees the tomato directly. Keith is in
direct contact with the perceived object. There is no tomato-like sense-impression
that stands as an intermediary between the tomato and him. Nor is he related to
the tomato as I am to a pig when I see its footprint in the mud. He does not expe-
rience the tomato by experiencing something else over and above the tomato and
its facing surface. Keith sees the facing surface of the tomato directly.
Some disjunctivists have suggested that to do proper justice to the above
thought, we need to suppose that the objects we perceive are components of the
contents of our perceptual experiences in veridical cases. This supposition is sup-
ported further by the simple observation that if I see an object, it must look some
way to me, and if an object looks some way to me, then intuitively it is experienced
as being some way. On pain of losing direct contact with the object, that again

1
In particular, the passage that goes as follows: “When I see a tomato there is much that I can
doubt. I can doubt that it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can
doubt whether there is any material thing there at all. . . . One thing, however, I cannot doubt: that there
is exist a red patch of a round and somewhat bulgy shape” (Price 1932, p. 3).
2
I shall ignore the case of de re hallucinations. These may be regarded as illusions of a special sort.
As to how I am using the expression ‘representational content’ in sections 12.1 and 12.2 below, I take
it that a visual experience v, in having the representational content that p, is related to an entity picked
out by ‘that p’ that is v’s representational content. I further take it that v has accuracy conditions and
that v’s having the accuracy conditions it does follows from its having the representational content it
does. I intend my question, “What is the content of Keith’s experience?” to be asking for an elucidation
of the nature of that content. I should add that I do not hold, as some do, that the content itself has
accuracy conditions, though I cannot pursue this point here. On my view, representational contents
generally do not have accuracy or truth conditions. Accuracy (or truth) and inaccuracy (or falsehood)
are properties of the vehicles having content. Finally, I should note that in section 12.3 below, I present
an alternative notion of content.
 What is the Content of a Hallucinatory Experience?   293

suggests that the object itself figures in the content of the experience, assuming
that experience is representational at all.
In cases of illusion, the perceived object appears other than it is. In such cases,
intuitively, as already noted, the perceptual experience is inaccurate.3 And it is so
precisely because the object is not as it appears to be. The simplest explanation
of this, in my view, is that, where there is a perceived object, a perceptual experi-
ence has a content into which the perceived object enters along with its apparent
properties.4
Once it is acknowledged that in veridical and illusory cases the seen object
is a component of the content of the experience and thus that the content itself is
singular, a puzzle arises.5 In standard hallucinatory cases there is no object with
which the subject is in perceptual contact and correlatively no singular content.6
So, again, what is the content of Keith’s experience?
One possible answer is that the content is existential in the hallucinatory case.
Keith’s visual experience represents that there is something red, round and bulgy
before him, and since there is no such thing, his experience is inaccurate. The com-
bination of views that results is unlovely and implausible; for it is forced to postulate
a displeasing and radical asymmetry in cases that pre-theoretically seem alike.
Another possible answer is that Keith’s experience has no content at all in the
hallucinatory case. Keith is simply sensing a red, round, bulgy sense-datum or he
is sensing redly, roundly, and bulgily. Again, the resulting combination of views for
the veridical and hallucinatory cases is unlovely and implausible. And, as already
noted, there are other difficulties for views that introduce sense-data or go adver-
bial. This is so even if these views are restricted to hallucinations.
These reflections suggest that in the hallucinatory case, we should say that
there is content of the very same sort as in the veridical and illusory cases—con-
tent that is just like singular content but with a gap or hole in it where the object
is supposed to go. And this is what I have said in recent work (2009).7 However,
I have come to think that there are at least two better alternatives. The purpose
of this chapter is to explain what now seems to me problematic with the gappy
content proposal, to present the alternative views, and to draw out the conse-
quences of these reflections for the thesis of representationalism about phenom-
enal character.

3
Not everyone accepts this claim. One notable exception is Travis 2004. See also Brewer 2008.
4
On this view, the content is a structured entity. I should add that not all disjunctivists grant
that in cases of illusion, perceptual experiences have contents of the same sort as veridical perceptual
experiences. See, e.g., Martin 2006. Obviously, those disjunctivists who take this view cannot use the
present consideration to motivate their view.
5
McGinn (1982) and Davies (1992) deny that the content is singular for any perceptual
experiences. For criticisms of this position, see my 2009, ch. 4.
6
Assuming that the term ‘singular content’ is used in the usual way. For an opposing usage, see
Sainsbury 2006.
7
See also Bach (1997), Loar (2003), Burge (1991).
 Imagistic and Possible-Word Content
294   

12.1. The Trouble with Gappy Content

Consider the singular content that object, O, is red. On the Russellian view, this con-
tent is complex, having object, O, and redness as its components.8 A visual experience
(or other mental state) having that content is accurate if and only if O is red. One way
to think of the Russellian content here—the content that O is red—is as an ordered
pair having O as its first member and redness as its second. Another way to think of
the content is as a structured, possible state of affairs built out of O and redness.
It was suggested above that where a visual experience is hallucinatory, the
content is just like a singular content but with a gap or a hole in it where an object
should go. But does this really make sense?9 On the ordered pair conception of
singular content, there must be two items to form the pair. Since a gap or a hole is
not an item, or so it seems, there is no first member of the ordered pair and so no
ordered pair at all. On the possible state of affairs conception, the relevant complex
is structured out of O and redness in the singular case. But in the gappy case, there
is no object O. So, how is there a complex entity structured out of its components?
A possible reply is to say that the missing item in both cases is the empty set.
Where one hallucinates, the content is a complex entity built out of the empty
set and various properties. But intuitively this is a bizarre proposal indeed. If the
empty set is the gap filler, then the hallucinatory experience is about the empty set.
So, in a hallucination, one experiences the empty set just as, in the veridical and
illusory cases, one sees object, O. Furthermore, if the hallucinatory experience is
about the empty set then it is experienced as being some way or other, for example,
red. So then the empty set looks red, just as the seen object, O, looks red when
one sees O and experiences it as red! This is more than a little hard to swallow.
Furthermore, the proposal effectively reconfigures hallucinations as illusions of a
special sort. Instead of there being no object perceived, there is now an object of
a special sort that is perceived, or if not perceived, at least experienced and this
object is not as it is experienced to be.
The empty set proposal is a desperate attempt to save the gappy content view.
It confuses the truth that in hallucination one does experience something—for
example, a ripe tomato—with the falsehood that there exists some thing one
experiences, the empty set being proposed as the relevant thing since no ordinary
object is available to do the job.
Another possible reply is that the missing item is an absence. This needs a
little explanation. Roberto Casati and Achille Varzi (1994) make an interesting
case for the view that holes exist. If, for example, I say truly that there is a hole in
the cheese, on their view, what makes my remark true is the existence of a hole

8
Obviously, there are other views of singular content. For ease of exposition, I shall stick with the
Russellian conception in what follows.
9
This question is a pressing one for gappy proposition theorists (e.g., Braun 2005) generally.
 What is the Content of a Hallucinatory Experience?   295

surrounded by the relevant piece of cheese. What Casati and Varzi show (at a min-
imum) is that it is not at all easy to paraphrase away apparent commitment to holes
in our ordinary talk. Analogously, it might be suggested, a case can be made for the
view that there exist absences. Even if such a case were successful, however, there
is a pressing question for anyone who appeals to an absence in the context of a hal-
lucinatory experience, namely which absence is relevant?
For Keith, since he is hallucinating, there is no ripe tomato that he is seeing.
So, on the absence view, the relevant absence is that of any ripe tomato. That is
what he experiences as red. The content is an ordered pair of an absence of any ripe
tomato and redness, or a possible state of affairs structured out of these items,10
where the relevant absence presumably is a concrete entity existing wherever ripe
tomatoes do not and thus varying in its location with variations in the locations of
ripe tomatoes. But then how can Keith, sitting in his room, experience that entity?
How can that entity look any way to Keith?
Perhaps it will be replied that Keith experiences that entity by experiencing
part of it just as Keith sees a ripe tomato by seeing part of it (namely its facing
surface). If so, just which part of the absence does Keith experience? Again, at a
minimum, the proposal is very unpromising.
A third possible proposal is that the gap filler in the ordered pair is a
spatio-temporal region. On this proposal, in hallucinating a red object, one is
experiencing a particular spatial region as red. This seems plainly misguided,
however. Spatial regions do not look red, so one cannot experience them as red.
Perhaps it will be suggested that, in hallucination, one experiences a specific spa-
tial region as filled by an object having so-and-so properties, for example, redness.
The trouble now is that if one says this for the hallucinatory case, one should say
it for the veridical case too, at least if one wants to avoid the earlier mentioned
implausible and unlovely asymmetry. The price paid is that the seen object in the
veridical case is no longer in the content. In its place is the spatial region occupied
by the object. But that is not what one sees.
Another difficulty for the spatial region proposal is that it is very unclear
which region is to serve to fill the gap in the content. Suppose, for example, you
are in a dark room and you hallucinate a sudden bright pinpoint of light. It could
be that your experience does not locate the light at any particular distance away
from you. So, which spatial region in the room is the one that you are experiencing
as being occupied by the light? There is no obvious answer.
It seems to me that the net effect of these reflections is to cast serious doubt on
the wisdom of the idea that the gap in a gappy content should be filled. Whatever
else gappy content is for hallucinations, it can’t be a species of ordered pair content.
What are the remaining alternatives?

10
Obviously I am over-simplifying here, since the content of visual experience is extremely rich.
But this makes no difference for present purposes and so I ignore it.
 Imagistic and Possible-Word Content
296   

Perhaps we should try to think of gappy content as something that aspires to

be an ordered pair of a certain sort, but fails to achieve that status. With this in
mind, let us consider next what I shall call “the mailbox proposal.” A mailbox has
a slot that is designed to take letters. The slot may be empty or it may have a letter
placed in it. Perhaps we should think of the content of a visual experience as being
like a mailbox. In the veridical and illusory cases, there is a structure (the mailbox)
containing an object (the letter placed in the slot). In the hallucinatory case, there
is the same structure but no object (letter in the slot).
The immediate trouble with this proposal is that, as it stands, it offers no illu-
mination as to truth or accuracy conditions and so one may reasonably ask what it
has to do with content at all. However, this worry can be overcome by complicat-
ing the mailbox model a little. Suppose that the mailbox is designed to take letters
for a certain range of zip codes and only for those zip codes. Then no mistake has
been made by the person posting a letter just in case the letter in the mailbox has
one of the right set of zip codes. Analogously, it may be suggested, nature deliv-
ers an accurate visual experience—more simply, a visual experience is accurate
(true) if and only if the seen object (the ‘letter’ in the ‘mailbox’) has the property
it appears to have (a ‘zip code’ within the allowed range). Now where there is no
‘letter’ in the ‘mailbox’ or a disallowed ‘zip code’, the condition on the right hand
side of the biconditional is not met and the experience is inaccurate or false.
Leaving aside the point that the model proposed here is grossly oversimpli-
fied, given the richness of perceptual content, a general worry remains. Just what
is gappy content here? It isn’t the mailbox. As already noted, that isn’t content at
all. Nor is it an ordered pair consisting of the mailbox and a range of zip codes.
For that is present in veridical cases (cases in which there is a letter in the mailbox
with a permissible zip code for delivery from that mailbox). Clearly, something
further needs to be added. But what? Perhaps we should say that gappy content
is an ordered pair of the empty slot in the mailbox and a range of zip codes. The
trouble now is that we are back to a variant of one or other of the earlier ordered
pair proposals: in place of the empty set or an absence or a spatio-temporal region,
we now have an empty slot. And the person who is hallucinating experiences (and
thus is conscious of) that slot! No genuine progress has been made.
It might be suggested that what the mailbox model is really gesturing towards
is a view of gappy content as schematic content of a certain sort. Consider an open
sentence of the form ‘Fx’, where ‘x’ is a variable and ‘F’ a predicate for a specific
property (or cluster of properties). The initial thought is that in the hallucinatory
case, the visual experience is like the open sentence, ‘Fx’, whereas in the veridical
and illusory cases, it is like the closed sentence ‘Fa’, where ‘a’ refers to the object
seen. Since in the hallucinatory case, there is no value for the variable—no seen
object—the visual experience is either false or neither true nor false, depending
on the semantics devised for open sentences. If, for example, it is held that ‘Fx’
is true iff the value of ‘x’ is F, then where there is no value, the right hand side
of the bi-conditional is false, so the left hand side is false too, in which case, on a
two-valued semantics, ‘Fx’ is false.
 What is the Content of a Hallucinatory Experience?   297

These remarks do not yet provide us with a clear account of what the gappy
content is supposed to be in the hallucinatory case. And here there remains a
puzzle. To appreciate this, note that in the open sentence ‘Fx’, there are two com-
ponent parts: the predicate ‘F’ and the variable ‘x’. The counterpart to ‘F’ in the
gappy content of a visual experience is the property of being F, as it is in the sin-
gular content case. What is the counterpart to the variable ‘x’? It looks as if we
have to say, as before, that it is something like a slot in a mailbox. Gappy content,
then, presumably is an ordered pair consisting of the empty slot and the relevant
property or complex of properties. But now we are back with a version of the or-
dered pair view and all of the obscurity that goes with it.11
Furthermore, if we do make this proposal about gappy content then, to pre-
serve uniformity, we ought to say that singular content is an ordered pair consisting
of the filled slot and the relevant property. Unfortunately, this is too general. Not
just any old filling of the slot will do in the singular case. We need one particular
filling—that provided by the seen object—so that the singular content is now more
complex than previously supposed, having something like the following structure
(where P is the relevant property and CONJ the truth function for conjunction):
< CONJ << the relation of filling, < a, the slot >>, < the property P, a >>>
Modifying the account of gappy content correspondingly, we have:
< AND << the relation of filling, < --, the slot>>, < the property P, -- >>>
where—is the absence of any seen object or some other dubious item. Obviously,
we are getting nowhere.
Perhaps we should think of the content of a hallucinatory experience as being
like a tree structure. The model in this case is the tree structure linguists take
sentences to have. On this proposal, in veridical and illusory cases, when one
experiences an object, O, as red, the structure has two branches, each of which
has an entity at the end:
Propositional Content
/ \
Object O  being red
The left hand branch is reserved for objects and the right hand one for 1-place
properties. In the hallucinatory case, there are the same two branches but the left
hand one is empty:
Propositional Content
/ \
being red

11
Barwise and Perry (1981) call the propositional analogs of variables “indeterminates”. As applied
to the case of the content of hallucinatory experience, this does not help with the earlier worries.
 Imagistic and Possible-Word Content
298   

It is important to appreciate that in linguistics, the tree structure for a sentence

is not a part of the sentence. Rather when linguists talk of a sentence as having a
certain tree structure, they are saying that the parts of the sentence are related in
a certain tree-like way.
On this view, in the veridical and illusory cases, the content is to be under-
stood as follows:
< Quasi-branching relation Q, < O, the property of being red>>.
Here Q is the worldly counterpart of the linguistic branching relation connecting
the constituents of the sentence “O is red.” Accuracy conditions are now straight-
forward. The experience is accurate just in case O bears quasi-branching relation,
Q, to redness. Given that the same two-term relation, Q, is part of the content in
the hallucinatory case, it seems that the gappy content must be this:
< Quasi-branching relation Q, < --, the property of being red>>.
But this isn’t coherent unless ‘--’ picks out an entity, in which case we have made
no progress.
There is also a lack of clarity in the appeal to quasi-branching relation Q.
Perhaps Q is supposed to be what some have called “the exemplification tie”
linking objects and properties. If so, then, in the veridical and illusory cases, the
visual experience must be counted as accurate just in case O exemplifies redness,
in which case the ordered pair of O and redness would serve just as well as the
content and we really are back to square one.
There is a more complicated alternative in the general area of the last proposal
worth considering. Suppose, following Jeff King (2007), we take the propositional
content that O is red to be a worldly fact (or obtaining, actual state of affairs), spe-
cifically, the fact of O’s standing in relation R to the property of being red, where
this is to be unpacked as follows:
The propositional content that O is red = the fact of there being lexical items
a and b of some language L such that O is the semantic value of a, where a
occurs at the left terminal node of the sentential relation S that in L encodes
the instantiation function, and b occurs a S’s right terminal node and has as
its semantic value the property of being red.12
Then it might be proposed that we can extend King’s view to handle the case of
hallucinatory content in the following way. Where the person hallucinating seems
to see O and it seems to her that O is red, the content of her experience is the fact
of there being lexical items a and b in a language L such that a has no semantic
value and a occurs at the left hand terminal node of the sentential relation S that

12
This oversimplifies King’s proposal minimally. For King’s motivations for such a view, see his
2007. King takes his account to explain the unity of a proposition as well as the constitutive link (as he
sees it) between propositions and language.
 What is the Content of a Hallucinatory Experience?   299

in L encodes the instantiation function and b occurs at S’s right terminal node and
has the property of being red as its semantic value.
This does seem to me coherent. But it is complicated and it requires that
facts about language be part of the content of visual experience. This seems to me
counter-intuitive, especially if we accept (as I do) that experiences themselves are
more like maps or pictures than linguistic representations (see below). Perhaps
the proposal could be made more viable by allowing the relevant scheme of rep-
resentation to be internal and nonlinguistic in character; but obviously this would
necessitate further revisions since there would be no relevant sentential relation S.
The upshot, in my view, is that the prospects for understanding the putative
gappy content of visual experience are very gloomy. It behooves us to look for a
better alternative.

12.2. An Alternative View of the Content of Visual Experience

Visual experiences are not like conscious thoughts. One can think that there is a
ball on a box without one’s thought representing anything further about the ball
and the box; but one cannot have a visual experience of a ball on a box without
one’s experience representing such things as the color of the ball, the color of the
box, the relative size of the two, the shape of the box, the view-relative locations
of the ball and the box, and so on. Visual experiences, in representing one thing,
represent many. In this way, visual experiences are like maps or pictures whereas
thoughts are like sentences. Visual experiences are representationally rich. It is
tempting to infer from this that the content of visual experience must be corre-
spondingly rich. But this is a mistake. It falsely assumes that a property of the
vehicle of representation (the experience) must be a property of its content.
Structure in a representation need not be mirrored in structure in its content.
This point applies not just to experiences but to thoughts as well, even though
they lack the representational richness of experiences. Consider the thought that
Vulcan is a planet. The thought is complex, being composed of the concept Vulcan
and the concept planet, combined in a certain way (rather as the sentence “Vulcan
is a planet” is composed of the words ‘Vulcan” and ‘planet’ in a certain order).
However, the content of the thought need not be complex. The content can be just
a set of possible worlds—the set of worlds in which the referent of the concept
Vulcan has the property referred to by the concept planet (the property attributed
by the thought).13 Since the concept Vulcan has no referent either in the actual
world in any other possible world, the relevant set of possible worlds is empty.
Correspondingly, experiences are complex. They have representational parts.
Some parts represent objects seen (if there are any); others represent properties of

13
This is not to suggest that thoughts are to be individuated by their contents, so conceived. For
more on the nature of thought, see Sainsbury and Tye 2012.
 Imagistic and Possible-Word Content
300   

which the subjects of the experience are conscious. The latter representational parts
are arguably more like features (as is the case with real pictures), and so in one sense
they are not really parts at all but rather features of parts. But however this is devel-
oped further, a complexity in representational structure for experiences need not be
reflected in a corresponding complexity in representational content.
Once this point is appreciated, given the difficulties already encountered in
trying to understand gappy content, a natural suggestion is that the content of
a visual experience is simply a set of possible worlds, namely the set of worlds at
which the experience is accurate. On this view, the content of a visual experience
is unstructured in the sense that it has no component parts.
This suggestion preserves uniformity in content for all experiences. Experiences,
whether they are veridical, illusory, or hallucinatory, have associated with them an
appropriate set of possible worlds. An experience, thus, is accurate, if and only if the
actual world belongs to the appropriate set of possible worlds. Which is the appro-
priate set? Answer: the set of worlds at which the objects picked out by representa-
tional parts of the experience have the properties the experience aims to attribute to
those objects (however this is further cashed out). The objects thus picked out are
the objects (if any) that are seen. Where there are no seen objects, as in a hallucina-
tion, there are no possible worlds at which the objects picked out by the representa-
tional parts of the experience have the experienced properties. So, the set of worlds
associated with a hallucinatory experience is the empty set.14
What, then, of the reasons given at the beginning of the chapter for the view
that seen objects are components of the contents of visual experiences? Consider
first this reasoning from earlier on:
(1) If I see an object, it looks some way to me.
(2) If an object looks some way to me, then it is experienced as being some
way.
(3) If an object is experienced as being some way, then it is a component part of
the content of the experience (assuming that the experience has a content).
So,
(4) If I see an object, the seen object itself is a component part of the content
of the experience, assuming that experience is representational at all.
The premise I reject here is (3). If a given object, O, is experienced as being
some way then the experience represents O as being some way (e.g., red). But this
is now cashed out further in terms of the experience having as its representational

14
On this view, (rather obviously) experiences are not to be individuated simply by their
phenomenal character. The experience of seeing something, O, and experiencing it as red has a
different content (set of possible worlds associated with it) than does the experience of hallucinating
something red, even though the experiences have (or may have) the same phenomenal character. The
former experience has the set of possible worlds at which O is red as its content (and there are many
such worlds); the latter experience has the empty set as its content.
 What is the Content of a Hallucinatory Experience?   301

content the set of possible worlds at which O is that way (red). And O is not part
of the content of the experience.
Of course, the set of possible worlds at which an experience that is about O
is accurate has members, namely possible worlds, each of which has O as a com-
ponent part (whether one supposes that worlds are maximal states of affairs or
(implausibly) concrete configurations of objects). But O is not a component part of
that set, any more than my heart is a component part of the set whose members are
me and my mother.15 So, O is not a part of the content of the experience.16
Even though, on my view, O is not a part of the content of the experience, the
experience remains crucially dependent on O. This is because the content of the
experience is specified by reference to O as the set of worlds at which O is red. Thus,
given that in actuality there is such an object as O, had things been different and O
not existed, neither would the experience occasioned in actuality by P’s seeing O.
The same points can be made in the case of thought on an austere view of
thought content. The thought that Cicero is an orator is a thought about Cicero.
It is about Cicero in virtue of its having an atomic nominative concept that refers
to Cicero. The thought attributes the property of being an orator to the person so
referred to. The content of the thought is the set of possible worlds at which Cicero
is an orator and the thought is true since the actual world is a member of that set.
That content does not contain Cicero as a component part even though the thought
is about Cicero. And that thought would not have existed had Cicero not existed.
The second piece of reasoning offered earlier for the view that the seen object
is in the content of the experience went as follows:
(5) In cases of illusion, the seen object appears other than it is.
(6) If the seen object appears other than it is, then the visual experience is
inaccurate.
The best explanation of such inaccuracy is
(7) Where there is a seen object, a visual experience has a content into which
the seen object enters along with its apparent properties so that the expe-
rience is accurate if and only if the object has those apparent properties.
I deny that (7) really is the best explanation of the relevant inaccuracy. If the
content of a given visual experience is a set of possible worlds at which the seen
object, O, has the properties, P1‒Pn, attributed to it by the experience then we have

15
Set membership should not be confused with the part-whole relation. The former is irreflexive,
asymmetric and intransitive; the latter reflexive, asymmetric and transitive.
16
Earlier, I assumed that the members of an ordered pair are parts of the ordered pair. This
assumption might be challenged on the grounds that ordered pairs can be defined set-theoretically
(e.g., on Wiener’s definition, <a,b>:= (((a), the empty set), ((b))), in which case they have sets as their
parts. The trouble here is that there are too many equally good definitions and thus too many equally
good candidates. We should accept all or none of them. We can’t accept all of them because they are in
conflict. So, we should accept none.
 Imagistic and Possible-Word Content
302   

an equally good explanation of the inaccuracy for the case in which O appears
other than it is. In that case, the actual world is not a member of the set of worlds
at which O has the properties, P1‒Pn and O is not a component part of the content.

12.3. A Second Alternative View of the Content of Visual Experience

I begin with some general remarks about indexicals and Kaplan’s theory of indexi-
cals (1989). Then I show its relevance to the case of visual experience. Indexicals
are terms that change their reference from utterance to utterance. Examples are ‘I’,
‘here’, ‘she’, ‘that’, ‘today’ and ‘here’.
Consider the following two utterances:
Tim: “I am hot”
Tom: “I am hot”.

Intuitively, these two utterances have the same linguistic meaning, but what Tim
says is different from what Tom says. Tim, who is cold (let us suppose) says some-
thing false; but Tom (who is hot) says something true. So, the content of Tim’s
remark is different from the content of Tom’s.
On Kaplan’s theory (1989), indexicals have contents with respect to contexts.
For example, the content of ‘I’ with respect to a given context C is the subject or
agent of C; the content of ‘that’ with respect to C is the object demonstrated in C;
the content of ‘here’ with respect to C is the location of C. The content of a sen-
tence containing an indexical is a structured proposition having as its constituents
the content of the indexical (the agent, place, object demonstrated, etc.) and the
contents of the other terms, where these contents are taken to be worldly enti-
ties: particulars, properties and relations. Thus, in the case of Tim’s utterance of
the sentence “I am hot,” the content of Tim’s remark is a structured proposition
containing Tim himself (the subject in this context) and the property of being hot
(the content of the predicate ‘is hot’17). The sentence is false in the context, given
that Tim is cold.
On Kaplan’s theory, the linguistic meaning of an indexical term is a function that
maps contexts onto contents, where the latter are those contents the term has at each
context. Kaplan calls this function the term’s character. Thus, consider the term ‘here’.
Its character is a function from contexts whose value at each context is the location
of that context. Similarly, the character of the term ‘that’ is a function from contexts
to the objects demonstrated in those contexts. In the case of sentences containing
indexicals, their characters are functions from contexts to the structured propositions
that are the contents of the sentences in those contexts.

17
We may say that it is the character of ‘is hot’ to refer to or express the property of being hot,
regardless of the context. The character here is a function that yields the property of being hot at every
context.
 What is the Content of a Hallucinatory Experience?   303

It seems to me not unnatural to think of Kaplanian characters as contents of a

special sort. Intuitively, the sentence “I am hot”, in having the meaning it does, has a
certain representational content: it represents that the speaker is hot. This content is
not specified by giving the truth-conditions of the sentence in any particular context.
So, it is not content of the sort we have been concerned with so far. Rather character is
something that determines truth-conditions relative to contexts. For a given context,
character has, as its value, content, as we have understood it in previous sections. To
avoid confusion, I shall call content of this sort “content*”. I turn next to the relevance
of content* to visual experience. To bring this out, it is useful to reflect upon the case
of demonstratives used in failed demonstrations.
Suppose I mistakenly think that I have demonstrated something and that I have
used the term ‘that’ to refer to it. In reality, there is nothing to be demonstrated, no
referent for my utterance of ‘that’. The content of the term ‘that’ in any given context
of utterance, on Kaplan’s theory, is the object demonstrated in that context. Where the
context lacks a demonstrated object, ‘that’ lacks a content with respect to that context.
The term ‘that’ does, however, have a linguistic meaning and thus a content*. This is a
function that maps contexts onto the objects demonstrated in those contexts (where
there are such objects).
Consider now the case of visual experience and hallucinatory experience in par-
ticular. The natural extension of the above account is that hallucinations are like cases
of failed demonstration. Just as the word ‘this’ has a content* but it lacks a content,
when uttered in a failed demonstration, so each experience (individuated solely phe-
nomenally) has a content*, but it lacks a content, when tokened in a hallucination.
This is a reflection of the fact that what visual experiences fundamentally aim to do
is to put us in contact with objects around us. Where there is no object, as in the case
of hallucination, there is no contact and so, as we might say, the experience is a failed
experience.
On this proposal, visual experiences do not have gappy contents. There are no
such things. Rather each visual experience has a content*. That is to say, associated
with each visual experience (individuated as a phenomenal type) is a function that,
for each context in which it is tokened and in which there is a seen object, O, takes
as its value a singular content (into which O and the properties O is experienced as
having enter).18 The singular content is the content the experience has with respect to
the relevant context.19 Where the context is such that there is no seen object, the visual
experience lacks a content with respect to that context.
It may be wondered whether the views I have sketched in this section and the
previous one of the content of visual experience have consequences for the thesis
of representationalism with respect to the phenomenal character of visual experi-
ence. It is to an examination of this issue that I turn next.

18
For ease of exposition, I assume here a single object seen. Obviously, this is an over-simplification.
19
So, seen objects do indeed enter into the content of visual experience in such cases.
 Imagistic and Possible-Word Content
304   

12.4. Representationalism

Strong representationalism (or intentionalism) is the view that phenomenal char-

acter is one and the same as a certain sort of representational content. Weak repre-
sentationalism, unlike strong representationalism, is not an identity thesis. It does
not purport to identify phenomenal character with representational content. It is
rather a supervenience thesis. It asserts that necessarily experiences with the same
representational content have the same phenomenal character. This more modest
thesis offers no real illumination about the nature of phenomenal character.
If perceptual content is understood in the way that I elaborated in section
12.2 (as a set of possible worlds) then counter-examples can easily be constructed
to weak (and so also to strong) representationalism. Simply take any two phe-
nomenally different hallucinatory experiences. These experiences have the same
content, namely the empty set. But by hypothesis, their phenomenal character is
different. So, on the set of possible worlds view, both weak and strong representa-
tionalism are false. What about on the Kaplanian view presented in section 12.3?
Consider two phenomenally different hallucinatory experiences. Here there
are two different phenomenal types, T and Tʹ. Neither type, within the context of
the hallucination in which it is tokened, has a content. But each type has a con-
tent*. Moreover, these content*s are different. To see this, suppose that type T is
tokened in a hallucination of a yellow flash and type Tʹ is tokened in a hallucina-
tion of a blue flash. T could have been tokened in the context of seeing something,
S. In that context, the value of the function which is T ’s content* is a singular
content consisting of S and the property of being a yellow flash. Similarly, Tʹ could
have been tokened in the context of seeing S. In that context, the value of the func-
tion which is Tʹ’s content* would have been a different singular content composed
of S and the property of being a blue flash. So, the functions themselves—that
is, the content*s—are different. On the content* view, then, there is no threat to
either weak or strong representationalism (at least if representationalism is taken
to include content*s in the contents that fix phenomenal character).
I want finally to consider whether, for the philosopher who wishes to think
of the content of visual experience in the unstructured set of possible worlds way,
there is any other broadly representationalist view in the neighborhood that is not
refuted by the case of phenomenally different hallucinatory experiences.
Representationalists generally accept that persons, in undergoing visual expe-
riences, are conscious of a range of properties or features that may or may not be
instantiated in the surrounding environment.20 These properties or features are

20
Along with (most) other representationalists, I am happy to say that, in the hallucinatory
case, the perceiver is conscious of an un-instantiated property. This seems to me to be part of naïve
commonsense. Suppose that you had never seen any red things and then, one day, you hallucinated a
red car. Did you not then encounter redness in your experience? Did you not then “get a good look” at
redness (Hawthorne and Kovakovitch 2006), one that enabled you then and there to know what it is
 What is the Content of a Hallucinatory Experience?   305

predicatively represented by the visual experiences (via the experiences having

parts (or features of parts21) that represent them). In veridical cases, the seen things
have the properties so represented. The following thesis, then, is one that represen-
tationalists should endorse:
(R) Necessarily, visual experiences that predicatively represent the same prop-
erty complex have the same phenomenal character.22
By avoiding mention of content, (R) is not troubled by the case of phenomenally
different hallucinatory experiences. Such experiences predicatively represent dif-
ferent properties.23
To summarize: on the alternatives presented in this chapter, we have the fol-
lowing packages of views:

Package 1: Each visual experience has a representational content. Such

content is unstructured: a set of possible worlds. Each visual experi-
ence also has a phenomenal character. Such character is (or supervenes
on) an external property complex. The complex is predicatively repre-
sented in visual perception whether the experience is veridical, illusory
or hallucinatory. Within this package of views, as on classical accounts
of perception, there is a common factor, indeed there are two, neither of
which determines the other.24 And this is the case even though within
the package there is no room (or need) for what have traditionally been
called ‘qualia’.

like to experience red? Of course, some further account needs to be given of what it is for an experience
to represent or be about a property that does not require that the property be instantiated. But there are
several such accounts in the literature. Take, for example, the Normal tracking account of basic sensory
representation (in first approximation, a sensory state is about a property, P, just in case the state is of
a type that is Normally tokened if and only if P is tokened and because P is tokened). This relationship
between the state and the property can obtain even if in some Abnormal situations the property is
not instantiated. The same is true if we think of property representation by experiences in terms of
indicator function (see Dretske 1995).
21
If experiences are like pictures, they have parts, features of which play a representational role
analogous to the role played by predicates in sentences; for present taxonomic purposes I am assuming
that this counts as a kind of predicative representation.
22
The thesis that visual experiences predicatively represent property complexes is also endorsed
by Mark Johnston (2004) and Colin McGinn (1999). Johnston writes of “sensible profiles” and McGinn
of “clusters of properties.” In my view, visual experiences are nonconceptual representations. (This is
another way in which visual experiences differ from thoughts.) So, where properties are conceptually
represented in perceptual acts, they are not represented by perceptual experiences proper but rather by
associated judgments or beliefs (within, for example, hybrid mental acts of seeing-that).
23
Likewise the visual experiences undergone in perceptually different waterfall illusion scenarios
do not present a problem for (R). Although the content of these experiences is the same—the empty
set—the property complex predicatively represented is different.
24
Given package 1, experiences that are exactly alike phenomenally can differ with respect to
their representational content (take a veridical experience and a phenomenally identical hallucinatory
one) and experiences that are exactly alike with respect to their representational content can differ
phenomenally (take phenomenally different hallucinatory experiences).
 Imagistic and Possible-Word Content
306   

Package 2: Each visual experience (individuated phenomenally) has a rep-

resentational content*. Such content* is a function from contexts in
which an object is seen to singular contents made up of the seen ob-
ject and the property complex predicatively represented by the visual
experience in that context.25 Within this package of views, as on clas-
sical accounts of perception, there is a common factor, indeed, as in
package 1, there are two: phenomenal character and content*, where
the latter determines or fixes the former. Phenomenal character is (or
supervenes on) the external property complex in the experience’s con-
tent*. Again, there is no room (or need) for what have traditionally been
called ‘qualia’.26

Appendix

How we should think of the property complex that is predicatively represented in

visual experience? The scene before one’s eyes is usually made up of a number of
different objects arranged in various spatial relationships. Suppose that the scene
one is viewing consists of a red, triangular thing to the left of a green, round thing.
Evidently, what it is like for one to view this scene is different from what it is like
to view a second scene containing the same things but with the green, round item
on the left. What is the property complex with the phenomenal character of one’s
experience viewing the first scene is to be identified or upon which it supervenes
on the representationalist view? It cannot just be the complex:
Being red and being triangular and being green and being round.
For this complex does not distinguish one’s experience phenomenally from one’s
experience of the second scene. Nor is the following complex any better:
Being red and being triangular and being to the left of and being green and
being round.
For the constituents of this complex, like those of the first one, can be commuted
and so again no distinction has been drawn between the phenomenal characters
of the two experiences.27

25
I ignore here more complicated cases in which there are multiple things seen. For some
discussion of these, see the appendix.
26
Of these two packages, the one I prefer is package 1. In part this is because of considerations of
systematic unity and fit with belief content (see here Sainsbury and Tye 2012).
27
Of course, these property complexes are much too coarse-grained to capture phenomenal
character anyway, since the experiences predicatively represent determinate shades of color,
determinate shapes, location and many other details. But this makes no difference to the point I am
currently making and so I ignore it.
 What is the Content of a Hallucinatory Experience?   307

Suppose we complicate things by saying that the relevant property complex

is as follows:
Being red and being triangular and being to the left of a green round thing and
being green and being round and being to the right of a red triangular thing.
In the case of the second experience, the property complex is different with respect
to the left/right relations, so this proposal at least has the virtue of keeping the phe-
nomenal characters of the two experiences distinct. But it seems ad hoc. Why pick
the above property for the first experience? Why not pick this one instead, say?
Being red and being triangular and being to the left of a green round thing
that is to the right of a red thing and being green and being round and being
to the right of a red triangular thing that is to the left of a green thing.
And there are indefinitely many other equally good candidates.
So, what is the right account of the relevant property complex? Let us call the
two objects in the scene ‘O1’ and ‘O2’. One’s first experience is not about an ordered
pair composed of O1 and O2. Intuitively, it is simply about O1 and O2. What one’s
experience does is to attribute a complex property to O1 and O2. Since one’s first
experience is accurate if and only if O1 is red and triangular and O2 is green and
round and O1 is to the left of O2, the property so attributed is as follows (using
‘%xFx’ as a singular term abbreviating an expression of the form ‘the property of
being an x such that x is F’):
(P1) %x%y(x is red & x is triangular & y is green and y is round and x is to
the left of y).
That complex property is not a property of O1 in the veridical case nor is it a prop-
erty of O2. It is a property of objects, O1 and O2.28 It is also not the property one’s
experience of the second scene attributes to objects, O1 and O2. The property in
that case is this:
(P2) %x%y(x is green & x is round & y is red and y is triangular and x is to
the left of y).
So, the two phenomenal characters are in no danger of being conflated.29

References

Bach, K. (1997). Searle against the world: How can experiences find their objects?
Manuscript. http://userwww.sfsu.edu/~kbach/Searle.html.

28
O1 and O2 jointly instantiate (P1). It is an interesting further question in metaphysics as to how
to understand what is involved in O1 and O2 having a property that is not possessed either by O1 alone
or by O2 alone or by the ordered pair <O1, O2>. Here are some further examples of joint instantiations:
John and Jane together lifted the piano; the children stood in a circle.
29
Thanks to Mark Sainsbury for written comments and Brian Cutter for discussion.
 Imagistic and Possible-Word Content
308   

Barwise, J., & Perry, J. (1981). Situations and Attitudes. Berlin: Walter de Gruyter.
Brewer, B. (2008). How to account for illusion. In A. Haddock & F. Macpherson (Eds.),
Disjunctivism: Perception, Action, Knowledge (pp. 168–180). Oxford: Oxford University
Press.
Burge, T. (1991). Vision and intentional content. In E. LePore and R. Van Gulick (Eds.), John
Searle and his Critics (pp. 195–214). Oxford: Blackwell.
Casati, R, & Varzi, A. C. (1994). Holes and Other Superficialities, Bradford Books. Cambridge,
MA: MIT Press.
Davies, M. (1992). Perceptual content and local supervenience. Proceedings of the Aristotelian
Society, 92, 21–45.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Hawthorne, J., & Kovakovitch, K. (2006). Disjunctivism. Proceedings of the Aristotelian
Society, Supplementary Volume, 80, 145–183.
Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 103, 113–183.
Kaplan, D. (1989). ‘Demonstratives’ and ‘Afterthoughts’. In J. Almog, J. Perry, & H. Wettstein
(Eds.), Themes from Kaplan (pp. 481–565). New York: Oxford University Press.
King, J. (2007). The Nature and Structure of Content. Oxford: Oxford University Press.
Loar, B. (2003). Phenomenal intentionality as the basis of mental content. In M. Hahn &
B. Ramberg (Eds.), Reflections and Replies: Essays on the Philosophy of Tyler Burge (pp.
229–258). Cambridge, MA: MIT Press.
Martin, M. G. F. (2006). On being alienated. In T. S. Gendler & J. Hawthorne (Eds.),
Perceptual Experience (pp. 354–410). Oxford: Oxford University Press.
McGinn, C. (1982). The Character of Mind. Oxford: Oxford University Press.
Price, H. H. (1932). Perception. London: Metheun & Co Ltd.
Sainsbury, M. (2006). Reference without Referents. Oxford: Oxford University Press.
Sainsbury, M., & Tye, M. (2012). Seven Puzzles of Thought (and How to Solve Them).
Oxford: Oxford University Press.
Tye, M. (2009). Consciousness Revisited: Materialism without Phenomenal Concepts.
Cambridge, MA: MIT Press.
PART FIVE

The Constituents of Perceptual

Content and the Role of
Perception
13

What Does Vision Represent?

William G. Lycan

This chapter addresses each of two interlocking issues. If the two did not interlock,
the world would be a better place. But they do: What does vision represent? And,
the Representational theory of sensory qualities. I shall begin with the title question.
But, first: Does perception represent, at all? It seems to me and to many hard
to deny that it does. But that is because we tend to think mainly of vision, a very
one-sided diet of cases as Wittgenstein would have called it; that smell, taste and
touch represent is not at all obvious—though I have argued at some length that
they do (Lycan 1996, ch. 7). Moreover, there is a recent upsurge of Wittgensteinian
and/or Gibsonian and/or Naïve Realist and/or disjunctivist skepticism about
even vision’s being representational: Campbell (2002), Travis (2004), Noë (2005),
Brewer (2006), Fish (2009). I am unconvinced by those skeptics’ arguments,
but I cannot go into them here.1 I would agree with majority opinion that if any
sense modality represents, vision does. And I shall just assume that vision does
represent.
Of course, if vision and the other senses do not represent, the Representational
theory is a non-starter; and that would be very bad for us materialists.

13.1. Conservative Views of What Vision Represents

In the spirit of traditional sense-datum theory, one might hold that vision repre-
sents only sense-datum-type properties, principally colors and shapes; compare
Marr’s “primal sketch.” Or one might allow that vision represents depth also, as

I am grateful to a number of people for help with this paper: to Nico Orlandi, Bill Fish, and
Susanna Siegel for (respective) conversations in 2006‒07 that inspired it; to Austen Clark, who gave
me very substantive help on it before commenting expertly on an early version presented to the
“Naturalized Philosophy of Mind and Language” conference in honor of Ruth Garrett Millikan,
University of Connecticut (October 2008); and to audiences at that conference and at the ANU, the
University of Otago, North Carolina State University, and the Tufts University Center for Cognitive
Science.
1
For effective critique, see Siegel (2010a) and Schellenberg (2011). 311
 The Constituents of Perceptual Content and the Role of Perception
312   

in Marr’s 2.5-D sketch, and volumetric shapes and distances as in Marr’s 3-D
sketch.
That is still a very spare inventory.2 Surely [but watch that “surely” operator!]
vision represents everyday objects, not just volumetric shapes and distances. And
object-recognition is obviously [!]‌one of vision’s functions. In The Modularity of
Mind (1983), Jerry Fodor defended the 3-D sketch view, but added that vision fur-
ther applies Rosch’s “basic categories,” such as “dog,” “shoe,” “chair,” “red,” “lady.”
The addition is incongruous with Fodor’s own strong modularity doctrine, which
requires pretty strict informational encapsulation. Perhaps he thinks, modularity
be damned, it is just common sense. But it is not Moorean common sense; it lies
somewhere between empirical psychology and the philosophy of representation.
Fodor does say that module outputs must be phenomenologically accessible, and
that the mere 3-D sketch does not satisfy that requirement. I am not sure why he
thinks the 3-D sketch is not phenomenologically accessible, but he defends the
requirement in a long footnote (1983, p. 136 n. 31).
One might add motion and change, as such.
There are further objections to such conservative views. (1) We simply see
and recognize individual things such as people (and their faces), not just prop-
erties or types of object. (Ruth Millikan has argued that “the ability to reidentify
things that are objectively the same when we encounter them in perception is
the most central cognitive ability that we possess” (2000, p. 109). (2) We sim-
ply see and recognize things as socially characterized—dollar bills, post offices,
square dances, without anything that could fairly be described as person-level
inference. Some liberals will go to the extreme, and maintain that we can
perceive (as such) electrons, social class, surges of monetary inflation, global
warming, and the like. (3) We perceive causal relations (Siegel 2006a, 2010b).
(4) Lyons (2005) defends a moderately and discriminatingly liberal view, based
on a highly original notion of “perceptual kinds” in the world. (5) There is evi-
dence that when we hear utterances or read text, we directly perceive meanings;
that is, our language modules are built to deliver sentence meanings and even
implicatures without any inference or other calculation from more primitive
percepts on the part of the whole subject (Pettit 2002). (6) Millikan argues for
an even broader field of direct perceivables: e.g., we can directly hear rain just
in virtue of someone saying, “It’s raining” (2004, p. 122); indeed, she says, “It’s
raining” is the sound of the rain from where we sit! (7) We see absences, as such
(Sorensen 2008; Farrenikova 2012).3

2
Clark (2000, 2004) suggests a conservative view, arguing that what vision primarily represents
are (a) features placed at locations and (b) “proto-objects” of the sort posited by Pylyshyn (2001). But
he does not rule out the notion that perceptual representation is layered (see below) and that vision
represents more sophisticated items at a higher level of organization.
3
Farrenikova points out that there is a difference between rich or nonconservative visual content
and conceptually high-level content. On her view, nonhuman animals can see absences.
 What Does Vision Represent?   313

13.2. Intractability of the Dispute

How might we adjudicate? That is, what should be our test for whether such-and-such
a thing or property is specifically represented in and by vision, without benefit of
either inference or some other contribution from general cognition?
We must be careful about this use of “vision.” Does it mean, (i) visual experience
as some writers put the question, (ii) seeing, whether “experienced” (i.e., consciously)
or not,4 or (iii) the visual system as investigated by cognitive and neuroscientists? I shall
try to mean (ii). (iii) would be an entirely empirical matter, though obviously scientific
results regarding the visual system bear fairly directly on our own issue. Moreover, I do
not doubt that the visual system subpersonally represents properties that are not seen
by the whole person whose visual system it is. (i) seems needlessly restrictive, since
(according to me) the difference between a type of mental state occurring noncon-
sciously and that same type of state occurring consciously is superficial, a matter of
whether the state is itself represented by a higher-order state (Lycan 1996, 1998, 2004);
it would not normally affect the state’s own representational content.
So my question more precisely is: what sorts of properties or things are spe-
cifically represented in and/or by person-level seeing? But at this point the issue
deepens horribly.
As Susanna Siegel has pointed out (2010b, ch. 2), negative arguments in this
area are ineffectual. But, I maintain, so are the most obvious positive arguments:

1. Introspection settles nothing. It could not have convinced a Russellian

that vision represents anything but colors and proximal shapes (of course
Russell himself did not believe that vision represents at all).5 Introspection
would not convince the author of objection (2) above that vision represents
only colors and shapes, because s/he introspects that there just visually
seem to her to be dollar bills, post offices and square dances. S/he is not
(normally) aware of any cognitive construction or inference.—But then, we
are not aware of very many of our own mental states or cognitive processes.
And introspection is just not fine-grained enough to adjudicate
objection (3). We just cannot tell introspectively whether everyday
implicatures are calculated à la Grice or simply seen/heard.
2. Uses of the word “looks” do not settle anything either. They are libertine
going into the debate. A car looks expensive; a building looks pretentious;
a piece of music (going by its score) looks very chromatic, nearly atonal.
Those are perfectly proper uses of “looks.” But not for that reason
should anyone think that vision unaided represents properties such as
expensiveness, pretentiousness or atonality.

4
Some writers use “experience” liberally, to include all cases of seeing (or whatever sensing)
whether conscious or not. I prefer to reserve the term for sensing consciously, i.e., for sensings of which
their subjects are aware.
5
But neither could the Russellian riposte by appealing to the Argument from Illusion, even
though Russell thought the argument showed that belief in anything but sense-data required inference
 The Constituents of Perceptual Content and the Role of Perception
314   

3. You cannot appeal to any psychosemantics. For a psychosemantics

cannot itself be assessed without prior assumptions about what things or
properties are represented by the modality in question.6
4. One might look to misperception. If a person forms a false perceptual
belief, we can ask whether that person misperceived, or rather perceived
veridically but miscognized. The answer would reveal whether the relevant
property was represented in vision or merely computed from visual output.
But any such argument will beg the question. A conservative will allow
that the subject (strictly) misperceived only when the subject mistook
a color or shape etc.; a liberal will allow a much wider range of uses of
“misperceive,” and insist that they are perfectly literal and strict.
5. One might invoke a prior notion of “direct” perception, and argue that all
and only what is directly seen is visually represented, or one of the two
component generalizations. But, first, “direct” is notoriously ambiguous or
context-relative: It means, not mediated; but not mediated by what kind
of thing? (Classically, it means not mediated by perceiving something else
first, but that is a very weak sense; the corresponding notion of “indirect”
perception is dramatically demanding.7)
Second, there is no obvious conceptual link between directness in any
sense and what vision itself represents (except, trivially and unhelpfully,
where “direct” means, not mediated by any but visual representations).
Even if Millikan is wrong about the rain, objections (2) and (3) above
suggest that social kinds and meanings may be directly perceived in some
perfectly good sense of “direct”; but it would not follow that they were
represented by vision alone.8

from sense-data; the facts of illusion could equally be just misrepresentation by the visual system. For
some resistance to my phenomenological claim, recall Firth (1965).
6
Siegel (2006, sec. 2) goes into this matter more thoroughly, considering more than one way in
which psychosemantics might be thought to help answer our title question. But she argues convincingly
against each.
7
And it still lingers, at least among some older epistemologists: Hearing any apparent departure
from Naïve or at least very Direct Realism about perception, such a philosopher will reflexively accuse
the speaker of holding that we perceive sense-data, or representations, or retinal stimulation, etc.
8
Millikan’s discussion of directness suggests each of two senses. In one of them, “direct” means,
not mediated by inference, where “inference” is read fairly strongly as person-level conceptual activity.
Millikan believes there are many psychological processes in which intentional representations produce
further representations in a regular and indispensable computational way, but what is required for
actual inference is that the representations are the vehicles of the subject’s beliefs (2000, pp. 118–119).
Beliefs figure in socially recognized forms of inference, deduction, induction, abduction, not merely in
algorithmic internal processes; and beliefs are catholic and all-purpose, not dedicated. None has any
specific job to do, and none is isolated from other beliefs in any way. All that seems quite right to me,
as a good thing to mean by “noninferential,” and noninferential is one good thing to mean by “direct.”
In the other, much weaker sense of “direct” (2004, p. 162), it means, produced by a process
essentially involving no other intentional representations as opposed to merely natural signs in the
sense of Grice and Dretske. That too is a perfectly fine thing to mean by “direct,” though I do not think
Millikan went on to make much use of the notion.
 What Does Vision Represent?   315

6. Siegel (2006a, 2006b, 2007, 2010b) has offered a new style of argument
for particular representational claims: the method of phenomenal contrast.
And, unlike the five preceding sorts of appeal, her style of argument
affords progress, though I shall argue that it too is ultimately inconclusive.
However, I will postpone discussion of Siegel’s method in order to address
a prior question.

13.3. Denying the Presupposition

Important qualification: My discussion so far has tacitly assumed at least a weak

version of Fodor’s (1983) modularity thesis: that there is a definite distinction be-
tween what is represented by vision alone and what is represented only when the
visual system’s output has been operated on by computational or inferential pro-
cesses that draw on background information that the visual system itself does not
have. But (i) that picture is hostage to neuroscience, and (ii) it also has strong phil-
osophical opponents in the tradition of Kuhn and Hanson (and Ernst Gombrich),
most notably Churchland (1988). On the latter view, vision is penetrated by back-
ground knowledge, assumptions and expectations practically from the retina; it is,
as the saying went in the 1960s, thoroughly theory-laden.
I subscribe to the weak modularity thesis (though I do not endorse all of Fodor’s
stronger modularity claims). But if it should prove to be as false as Churchland
maintains, our question is just a bad one: particular individuals, dollar bills and
whatnot are represented neither by “vision” alone nor by “cognition” resulting from
the output of vision. There is no such distinction. (But for resistance to this radical
view based on findings in perceptual neuroscience, see Gilman 1992, 1994.)
An intermediate position is that there are grades or stages in the transition from
very specifically visual activity to perceptual belief. The idea would be that although
there is an informationally encapsulated core module, it is smaller than Fodor sup-
poses, and is not surrounded by the sheer marshmallow but rather by a series of at
least a few outer layers, each more permeable by background information than the
previous one. Perhaps background information begins only gradually to penetrate
early visual processing, but does so more and more as the processing proceeds.
That would make sense of several facts, at least these: (i) That although
Fodor’s famous point about optical illusions is obviously right—the illusory look
firmly persists even when we know very well that (as it might be) the two lines are
equal in length—there is at least a little cultural relativity to optical illusions (Segall
et al.1968). (ii) More generally, there do seem to be grades of actual cognitive pen-
etration by culture. At the lowest such grade, contra Kuhn, we do not live in at all
different worlds; we expect anyone on earth to see certain distinctively shaped
objects, whether or not they know what they are. We expect nearly anyone to see
a fruit or a vegetable. (I do not mean as such, i.e., as “fruit” or “vegetable” in the
 The Constituents of Perceptual Content and the Role of Perception
316   

rich technical senses those words have in our language, but neither am I speaking
purely de re as opposed to under some low-level common representation involv-
ing potential food.) Perhaps a bit less universal is a piece of meat. Then there is
gradually increasing cultural relativity. Likewise, (iii) even within a culture such
as that of the United States in 2012, abilities to see grow with background knowl-
edge, though not in the entirely fluid, freewheeling way alleged by Churchland.
Everyone sees food, as before; everyone sees clothing as such even if under no
very specific subclassification; ordinary landscape features; ordinary artifacts. But
a child may not see a tool as such, and even a mature person may not see a particu-
lar tool as the tool it is; at least some background knowledge is required. Likewise
for stars. And then come dollar bills, post offices, and square dances, and then
arcane chemical stains. . . .
If this gradualist view is correct, what of our question? It is not simply invali-
dated by false presupposition, as Churchland would have it. Rather, it is relativized
to processing stage, and is then empirical: Is property P represented in the small
core? If not, is P represented at stage Core+1? And so on. This approach takes the
issue out of the purview of introspection and of common sense, which, if my ear-
lier pessimistic discussion assessment was right, is a good thing.
I am going to assume that there are limits that stop short of complete
theory-ladenness and the most libertine uses of “looks.” I mentioned cars’
looking expensive and buildings’ looking pretentious. Maybe those are border-
line cases. But consider a house looking uninhabited; a person looking as if s/
he is suffering from Sartrean vertigo; a southern New Zealand peak looking as if
global warming has advanced; a bank manager looking as though the worldwide
recession has deepened; and local space looking Riemannian (to a physicist who
has truly internalized the General Theory of Relativity). I do not see how “vision,”
per se, could represent those. If we can broadly speaking see such things, it would
have to be because either (a) the freewheeling Kuhn et al. thesis is correct and our
question was misconceived, there being not even a relativized or gradualist dis-
tinction between “vision” and cognition, or (b) vision represents such things but
only derivatively, in virtue of representing simpler properties (see below). I shall
continue to reject the freewheeling thesis.

13.4. Siegel’s Method

Siegel (2010b) appeals to the phenomenal change that occurs when vision is chun-
ked, when we substitute a more sophisticated recognitional capacity for what was
at first more painstaking. Her leading example is that of a novice forester who is
assigned a task to do with pine trees. At first the novice has to examine a tree to
see whether it is a pine tree, by comparing it to at least a short checklist of features.
But over time, the novice acquires a more direct recognitional ability and simply
spots pine trees as such. That is new phenomenology, or at least it contrasts with
 What Does Vision Represent?   317

the original checklist phenomenology. It is tempting to say that the novice can
now see, hence visually represents, a kind, “pine tree,” and to conclude that vision
sometimes represents particular natural kinds.
If that were the argument, it would get nowhere, for familiar reasons. That
the novice can now pick out pine trees more readily and it feels different to her
to do so shows (without question-begging) nothing about her specifically visual
representata. A Marr-inspired conservative could grant this datum; for that
matter, so could a Russellian sense-datum theorist. But the foregoing is not Siegel’s
argument. Rather, Siegel (2006a, 2010b) proceeds by closing loopholes. Her master
argument is this (2010b, p. 101): Let E1 and E2 be, respectively, the before and after
types of visual experience exemplified by our novice’s original and better-trained
perceptions of pine trees, and let “K-properties” be a type of property denied by
the conservative—in our example, the property of being a pine tree. The overall
experience of which E1 is a part is the “contrasting” experience, and the overall
experience of which E2 is a part is the “target” experience.
(0) The target experience differs in its phenomenology from the contrasting
experience.
(1) If the target experience differs in its phenomenology from the contrast-
ing experience, then there is a phenomenological difference between E1
and E2.
(2) If there is a phenomenological difference between E1 and E2, then E1 and
E2 differ in content.
(3) If there is a difference in content between E1 and E2, it is a difference with
respect to K-properties represented in E1 and E2.

________________________

∴(K) In some visual experiences, some K-properties are represented.

(0) Siegel takes to be obvious, and so do I (though I daresay it will be denied by
some hardheads about perception who reflexively reject virtually any appeal to
phenomenology). It is the conditional premises that can individually be resisted.
Well aware of that, Siegel defends each, seriatim, by essentially an inference to the
best explanation: the consequents explain the respective antecedents, and in each
case Siegel spends time rebutting competing explanations.
Anent (1), the competitors will be forms of nonsensory phenomenology. (The
idea would be that although E1 and E2 differ in overall phenomenology, their cores
of specifically sensory phenomenology are identical, and the difference is made
by phenomenal character of another sort.) Siegel (pp. 102–108) surveys several
possible types of nonsensory phenomenology, first cognitive phenomenology
and then background conditions such as drunkenness or depression. Subdividing
cases, she argues that none is plausibly what makes the (entire) difference between
E1 and E2. I think her case here is fine so far as it goes.
 The Constituents of Perceptual Content and the Role of Perception
318   

Anent (2), the competing explanations of the antecedent are changes that
are sensory but not representational. Siegel protests (pp. 109–110) that a “raw
feeling” occasioned by seeing the pine tree, such as one of familiarity, would
be extremely confusing were it not to represent something, a collection of
sense-data at least, as being familiar. (Siegel does not consider a “raw feel” of any
other type. I note that there are plenty of sensory feels or feel-components that
do not themselves represent, though according to me those must be identified
with functional properties; e.g., consider the motivational aspect of a pain expe-
rience (Lycan 1998).
Anent (3), the competing posit would be a visual representatum other than
a K-property. Siegel considers the example of a pine-tree-type shape Gestalt, the
idea being that what E2 represents that E1 does not is, not a tree, but a complex of
shapes that is generic enough to cover those of most pine trees but specific enough
to be shared by few non-pines. However, Siegel argues that although this Gestalt
move may work for pine trees, it does not extrapolate to other cases, such as a
human face’s property of expressing doubt (the faces of different sorts of people
may not express doubt in ways that can be captured in a shape Gestalt), or more
trenchantly, a line of Cyrillic text’s meaning.
At last we have an argument that does not beg the question and does a good
job of distributing the burden(s) of proof. And let us get rid of an apparent
objection: Siegel’s method does seem, short of question-begging, to frontload
liberalism, perhaps extreme liberalism. For we all have very high-level rec-
ognitional capacities that seem sensory and not to require inference. There is
obviously a phenomenological difference between seeing a car in economic
ignorance and having the very same car look expensive to you. And likewise for
ignorantly perceiving the Cyrillic text and seeing it with understanding, merely
seeing a face and seeing the same face as expressing doubt, etc. For that matter,
there is a phenomenological difference between practically anything and any-
thing. It should not be so easy to prove a priori that vision alone represents such
fancy properties.
Siegel (2007) addresses this, pointing out that the method of phenomenal con-
trast itself says nothing whatever about which properties are represented in vision;
the substantive work is done by premise (0), the phenomenology itself. It could
have been that Russell was firmly right, and there was no detectable phenomenal
difference between the novice’s experience of (what is in fact) the pine tree and
her/his learned experience of the tree as such; ditto for doubt in faces and mean-
ing in Cyrillic script. And pace premise (1), it could have been that there was no
sensory-phenomenal difference; etc. But the facts are otherwise. And the point of
Siegel’s master argument is to move us by discrete stages of reasoning from the
mere fact of phenomenal difference to the conclusion that a K-property is visually
represented.
Yet there are problems.
 What Does Vision Represent?   319

13.5. More Substantive Objections to Siegel’s Method

First, Siegel’s focus is on experience, i.e., conscious experience, phenomenology. In

my view, the latter requires at least some degree of internal attention, monitoring
or higher-order quasi-perception if you will, to the first-order visual state whose
content is in question (Lycan 1987, 1995, 1996, 2004). But (a) just as measure-
ment alters the quantity measured, attention alters the first-order state attended.
Moreover, (b) attending and monitoring are themselves intentional.9 They add
content to the overall experience, and introspection is not good at making fine
distinctions. So from the fact that conscious experiences differ phenomenally, it
does not anything like follow that the original underlying first-order perceptual
states themselves differed in content.
This point attacks Siegel’s premise (2). She does not consider the competing
explanations, that attention alters the first-order state, or that it adds content.
Second, attention/monitoring cannot settle what is purely visual vs. what is a
cognitive contribution, because as we have seen, the boundary is now unclear even
to sophisticated theorists. Of course Siegel does not claim otherwise, but again,
since she is talking about conscious experience, she is relying on tacit assumptions
about introspective awareness of our own first-order mental states. “Experience”
may involve the cognitive as well as the purely visual.
Third, Siegel’s examples almost certainly involve aspect-perception, “seeing
as.” The experienced forester sees the pine tree as such; we see the face as express-
ing doubt, and the aficionado sees the Cyrillic line as meaning what it does. But
the relation of “pure” vision and cognitive assumptions within aspect-seeing is
notoriously vexed. I shall go further into this below.10
But the issue of what vision represents gets worse; visual content seems to
have a more complicated structure than just, this type of property or that.

13.6. Layering Views

By a “layering” view I mean one according to which one and the same representa-
tion has multiple (at least double) contents, systematically related to each other by
some asymmetrical priority relation.

Millikan (1989)’s view is at least hospitable to layering. For her there is

no one standard relation that must ground any representation in the

9
Since writing the previous draft of this paper, I have learned from Wesley Sauret that monitoring
and attention are not as similar as they sound, and that majority opinion in neuroscience is that attention
is not in itself intentional. (More on this in a joint paper, in preparation.) It remains true that attention
alters the first-order state and the relevant sensory qualities; see Block (2010) and Lycan (forthcoming).
10
For further criticism of Siegel’s argument, see Lyons (2005).
 The Constituents of Perceptual Content and the Role of Perception
320   

environment, in order for the representation to have propositional con-

tent. That is because what a brain state represents depends on what that
state is going to be used for. What it is going to be used for depends on
which of the subject’s psychological agencies are going to “consume” it,
and for what, which depends in part on what their own functions are.11
So one and the same representation may well have multiple contents.12
Peacocke (1992): We represent indexical “scenario” content; low-level
properties (nonconceptual); and high-level properties. [I am not clear
whether it is the same vehicle that does both representings, but I gather
that that is what Peacocke intends.]
Lycan (1996): We represent high-level properties by representing scenario
content and low-level properties of (often unreal) external objects.
The model here is that of deferred ostension: Pointing at a chalk mark
on a blackboard, we refer to a numeral; thereby we refer to a number;
thereby we refer to an office in Emerson Hall; and thereby we refer to
its occupant, a person.
Noë (2004): We perceive high-level properties, though only as “present as
absent,” by actually-perceiving “perspectival properties” (= “appearance
properties”) of external objects.

11
Arguments for that crucial claim: (1) Any item is a representation only if it is used by some
cognizer as one; a representation represents something to someone or something. Since many of our
internal representations are subpersonal and do not represent anything to us ourselves, they must
represent to our subpersonal agencies. (2) On general grounds, Millikan is a teleofunctionalist and
she also—famously—gives a theory of teleology in terms of selection for effects; so representations as
such must be characterized in terms of their characteristic effects, and effects are relative to consuming
subsystems. (3) The consumer theory gives the best solution to the well-known indeterminacy problem,
most often exemplified by the question of whether the frog’s distinctive tongue-zap-stimulating
representation has black dots, small black things, flies, frog-food, edibles generally, or something else,
as its intentional object. (4) The theory handles all other data better than do competing views.
12
But I do not think Millikan herself believes in layering. She does maintain that higher-level,
more abstract representations are systematically and automatically produced by lower-level ones, a
process she terms “translation” and contrasts (as before) with inference. But the translation relation
holds between distinct, indeed successive representations: You represent some edges and shadings,
and (causally) respond to that by then representing a face-shape, and respond in turn to that in the
context by representing Johnny (2000, ch. 4 and p. 113). She does not use any such expression as “by”
or “in virtue of.”
She does have the excellent notion of “levels of distality”:
Contrary to much of the philosophical tradition, there is no single level of the outer world, such
as physical objects versus the mere surfaces of physical objects, or such as the presence of certain
phyical [sic] objects or of events versus mere sounds, of which the eyes or the ears are designed
exclusively to produce direct representations. Depending on the animal’s needs, various levels
of distality of direct perception may be mediated by the same sensory end organs. The affairs
naturally signified by retinal patterns, vibrating ear drums, stimulated odor sensors, and so forth,
are at various distances and are mediated in diverse ways. (2004, p. 162)
But that does not entail layering.
 What Does Vision Represent?   321

Schellenberg (2008): We perceive “situation-dependent” properties of ex-

ternal objects, and thereby the high-level properties of the same objects,
the perception of the latter depending epistemically on that of the
former.

Schellenberg’s view is superior to mine, in that it does not require any element of
illusion or seeing things that are not really there. In fact, I recant my 1996 view in
favor of hers.
(Of course all these views continue simply to assume that vision does represent.)
I have characterized layering theses by reference to a single representation’s
having multiple contents. There is a slightly less ambitious thesis available: that
an original representation having conservative or low-level content immediately
causes a second that has next-level, more sophisticated content, and the second
immediately causes a third, . . . , all too rapidly for introspection to detect that
the layered representations are numerically distinct. Some may prefer this ver-
sion. I continue to prefer the original, though I am not sure how we might decide
between the two.
Layering views offer some compromise with Siegel and other liberals. It seems
likely that although K-properties are not represented in the lowest layer, they may
be represented by something else’s having been represented in that layer or an
intermediate one.
But things get worse again. Things, because the original question of what
vision represents is about to be aggravated, and there are now more troublesome
apparent counterexamples to the Representational theory. I turn to matters of
aspect-perception.

13.7. Aspect-Perception

Contemporary literature on aspect-perception, of which there is surprisingly little

(but see Gilman (1992, 1994), Orlandi (2011a, 2011b, forthcoming)), starts with the
famous sec. xi of Wittgenstein’s Investigations (1953). One immediately thinks of
the duck-rabbit and other ambiguous figures. But it is important to realize that
there are different subspecies here, e.g.:

(i) Perceiving ordinary objects under aspects. A human head can be

perceived as a coconut perched on an inverted dinner plate (Booth
Tarkington’s Penrod and Sam, ch. XVII). And, contra Wittgenstein, a
human head can be perceived as a human head. A doorknob, normally
and correctly seen as convex, can at will be seen as concave.
(ii) Perceiving ordinary objects under very high-level aspects. A cloud can
be seen as the head of Thomas Eakins. (An ancient “Peanuts” reference.)
The shrunken snowcaps on the southern Alps of New Zealand can be
seen as global warming.
 The Constituents of Perceptual Content and the Role of Perception
322   

(iii) Pictorial seeing-as (the duck-rabbit).

(iv) Attentional phenomena (Peacocke’s (1983) dot arrays, Nickel’s (2007)
figure, Block’s (2010) Gabor patches).
(v) Using ordinary objects as aspect representations. In Patrick O’Brian’s
Aubrey and Maturin novels, naval officers sometimes while away
sociable drunken dinners in the gunroom by refighting old battles at the
table, using pieces of dinnerware such as plates, glasses and cutlery to
represent ships and their tactical movements. In such a case a wineglass
can be seen as H.M.S. Unspeakable.

Different subspecies may need different treatments. The obvious big difference is
between seeing an object as F and seeing a picture as a picture of something F.
The late Richard Wollheim (1996) made a valuable distinction between seeing-as
and “seeing in.” It is less that we see the duck-rabbit figure as a picture of a duck
(though of course we can do that) than that we see a duck and alternately a rabbit
in the picture. And it is fairly clear that there is a stricter sense in which (unless
visual conditions are very unusual) an ordinary wineglass cannot be seen as a ship
of the line.
Wittgenstein’s mystery was this: Seeing an aspect seems a voluntary matter
of interpretation; yet it is not merely interpretation, a cognitive construction
put on what is strictly seen, but itself a kind of seeing, a specifically visual
phenomenon that (as Wittgenstein would not have put it) is part of visual
phenomenology.
So—again—what does vision represent? Does it represent a coconut when
Penrod sees Georgie Bassett’s head as one? (In the story, the drowsy Penrod did
sincerely take Georgie’s head to be a coconut, but we should also consider the case
where Penrod is aspect-flipping at will and whimsically chooses to see it as one.)
I am reasonably sure that in the second case at least, Penrod’s visual system does
not represent the property of being a coconut, and I am even more sure that in
the Aubrey case the officers’ visual systems do not represent wineglasses as being
ships, but I could not confidently defend those views against the rhetorical ques-
tion: If seeing-as is a specifically visual and not merely cognitive phenomenon,
then how is seeing a head as a coconut not, at least in part, visually representing
the head as a coconut?

13.8. Sensory Qualities

“Sensory qualities” as I shall use the term are the distinctive introspectible qualita-
tive properties given in sensory experience—colors, smells, tastes, textures. A par-
adigm would be the color of an after-image, or that of a hallucinated object, to
emphasize that the qualities I am talking about are subjective, not or not necessarily
real features of real objects in the external world.
 What Does Vision Represent?   323

There is of course a metaphysical problem about subjective color. You hallucinate

a Granny Smith apple. It is green. But nothing in your brain is green, nor is there a real
green physical thing in front of you. Therefore it is a nonphysical thing.—Of course,
there have been philosophers who believed in Russellian sense-data and were quite
happy that they were nonphysical, but (a) that will not do if you are a materialist, and
(b) it should not be that easy to refute materialism even if materialism is in the end
false.
The Representational theory steps in: Sensory qualities are only intentional con-
tents or objects of mental states, represented properties of representata. The apparent
phenomenal bearers of sensory qualities (such as the apple) are intentional objects
of sensory states; some are real, some nonexistent. If you see a (real) Granny Smith
the greenness you see is that of the apple. If you hallucinate one, the greenness you
experience is still that of the apple you seem to see, even though the represented apple
and its represented greenness are not real. So much is, I hope, by now familiar (e.g.,
Dretske 1995, Tye 1995, Lycan 1996).
The Representational theory has an important implication: No sensory-quality
difference without a representational difference. Thus, a standard attack strategy is to
find counterexamples to that implicatum (e.g., Block 1996, 2003, 2010).
My layering view was originally a response to putative size-constancy counter-
examples to the Representational theory. (Notice I just assumed that vision represents
some high-level properties.) I extend the layering view to accommodate color con-
stancy also: We represent the greenness of the distant trees by representing blue.

13.9. Aspect-Perception and the Representational Theory

Fiona Macpherson (2006) has offered ambiguous figures as counterexamples to

Representationalism: Aspect-seeing of such a figure makes a phenomenal differ-
ence that is hard to cast as a content difference.13
A bit of care is needed here. Is the claim that there is a difference in sensory
quality as I am using the term? I am not sure, in part because the term itself is vague
at the edges. Lycan (2008) suggested that a property counts as a sensory quality14
just in case a subject becomes aware of it through the operation of something
having the function of feature detection. The phenomenology of aspect-switching

13
Two side issues: First, Macpherson’s official target is only a Representational theory, Tye’s
(1995), that restricts itself to nonconceptual content. She simply ignores layering views, not to mention
a Representationalism based on Fodor/Rosch.—But then, we do not know that vision does represent
ordinary objects.
Second: In the matter of psychosemantics, Macpherson again singles out Tye and (in passing)
uses his particular covariation psychosemantics against him. That is his problem, not one for
Representationalism generally.—But then, if there is any correct psychosemantics for vision, it is
probably comparably simple.
14
There mellifluously called a “Q-property.”
 The Constituents of Perceptual Content and the Role of Perception
324   

A B

Duck/Rabbit Picture Necker Cube Square/Regular

Diamond
FIGURE 13.1

does not seem to be that of feature detection, and if not, then I would argue that
aspect examples could not refute the Representational theory of sensory qualities.
But I shall ignore this and assess Macpherson’s cases on her own terms. Here is her
opening display.
Peacocke claims that his layering view handles the Mach square/diamond,
and without resorting to higher-level properties such as “diamond.” He appeals to
representation of the “symmetry-about” relation: when we see a square as such, vi-
sion represents the figure as symmetrical about the bisectors of its sides, but when
we see the same figure as a diamond, vision represents a different property, that of
symmetry about the bisectors of its angles. Macpherson makes several objections,
of which the most trenchant is that some ambiguous figures are not symmetrical
at all:

Distorted Square Kite

FIGURE 13.2

Following Ferrante, Gerbino, and Rock (1997), Macpherson considers the

suggestion that in aspect-seeing an ambiguous figure we subjectively impose
axes, which apply regardless of the figure’s actual orientation with respect to the
viewer, e.g.:
 What Does Vision Represent?   325

Up Up

Right

Left Right
Left

Down
Down
FIGURE 13.3

Thus, the visual experience may represent properties that parts of the figure
have in relation to the axes, such as “having an angle pointing directly up.”
Macpherson objects that this fails to account for the shape constancy of un-
ambiguous figures. Representationalists “must explain either why only one set
of axes can be imposed upon certain figures (the non-ambiguous ones) and why
more than one set can be imposed upon other figures (the ambiguous ones), or
they must explain why the experiences that represent different properties do not
give rise to experiences with different phenomenal characters in non-ambiguous
figures, but do so in the other ambiguous figure cases” (p. 108).
The former at least does need explaining. But I do not agree that the burden is
on the Representationalist. What first and foremost needs explaining is why some
figures are ambiguous and others not, and that is everyone’s problem. Once that
explanation is provided, the Representationalist can (probably) buy into it.
In any case I am inclined to take seeing-as to be an attentional phenomenon
(Chastain and Burnham 1975; Ricci and Blundo 1990; Kleinschmidt et al. 1998).
That at least reduces Wittgenstein’s mystery to a slightly more tractable puzzle about
attending: that visual attending in particular is a visual phenomenon, in that at least
it affects visual phenomenology; but in some sense what is seen does not change.
As noted above, Christopher Peacocke, Bernhard Nickel and Ned Block have
offered cases involving shifts of attention as counterexamples to the Representational
theory. I have addressed those elsewhere (Lycan 2000, forthcoming), and will not
repeat those discussions here. I offered a number of options for accommodating
the examples within the Representationalist framework, and I now suggest that
most of those options will apply back to cases of aspect perception.
But what of representational content in aspect perception? The obvious sug-
gestion is that when Penrod sees Georgie’s head as a coconut, his visual state
represents a coconut. Certainly somewhere in the experience—especially in
Tarkington’s original version as opposed to the voluntary-flipping version—the
 The Constituents of Perceptual Content and the Role of Perception
326   

concept “coconut” is automatically deployed. But as usual, we also have little

reason to think that the visual system knows anything about coconuts. Even if
somehow it does, physical objects can be seen as wedding gifts and as particle
accelerators too. Wittgenstein’s problem remains: Aspect perception seems thor-
oughly visual; interpretation happens; but the interpretation involves cognitive
penetration without being purely cognitive.
I have contended that the Representational theory of sensory qualities is
unrefuted by the aspect phenomena, but I have done nothing to limit the perti-
nent representational contents. If the Representationalist is to succeed in finding a
representational difference every time there is a phenomenal difference, s/he may
have to commit to some contentious representata, and thereby become hostage
to our unresolved issue of what vision does or can represent. That is a significant
liability.

13.10. Conclusions

I can draw hardly any. My main thesis remains, that it is hard to see what could
establish a claim as to what vision does or does not represent. Another thing that
remains is the possibility that the question is a bad one. All this, I fear, is grist to
the mill of the skeptics who deny that vision represents in the first place. I probably
should not have written this chapter.

References

Block, N. J. (1996). Mental paint and mental latex. In E. Villanueva (Ed.), Philosophical
Issues, 7: Perception (pp. 19–40). Atascadero, CA: Ridgeview Publishing.
Block, N. J. (2003). Mental paint. In M. Hahn & B. Ramberg (Eds.), Reflections and
Replies: Essays on the Philosophy of Tyler Burge (pp. 165–200). Cambridge, MA: MIT
Press.
Block, N. (2010). Attention and mental paint. In E. Sosa & E. Villanueva (Eds.), Philosophical
Issues, 20: Philosophy of Mind (pp. 23–63). Oxford: Wiley-Blackwell.
Brewer, B (2006). Perception and content. European Journal of Philosophy, 14, 165–181.
Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press.
Chastain, G., & Burnham, C. A. (1975). The first glimpse determines the perception of an
ambiguous figure. Perception and Psychophysics, 17, 221–224.
Churchland, P. (1988). Perceptual plasticity and theoretical neutrality: A reply to Jerry
Fodor. Philosophy of Science, 55, 167–187.
Clark, A. (2000). A Theory of Sentience. Oxford: Oxford University Press.
Clark, A. (2004). Feature-placing and proto-objects. Philosophical Psychology, 17, 451–477.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: Bradford Books / MIT Press.
Farrenikova, A. (2012). Seeing absence. Philosophical Studies, online, November.
Ferrante, D., Gerbino, I., & Rock, I. (1997). The right angle. In I. Rock (Ed.), Indirect
Perception (pp. 163–70). Cambridge, MA: MIT Press.
 What Does Vision Represent?   327

Firth, R. (1965). Sense-data and the percept theory. In R. J. Swartz (Ed.), Perceiving, Sensing,
and Knowing (pp. 204–70). Garden City, NY: Anchor Books.
Fish, W. (2009). Perception, Hallucination, and Illusion. Oxford: Oxford University Press.
Fodor, J. A. (1983). The Modularity of Mind. Cambridge, MA: Bradford Books / MIT Press.
Gilman, D. (1992). A new perspective on pictorial representation. Australasian Journal of
Philosophy, 70, 174–186.
Gilman, D. (1994). Pictures in cognition. Erkenntnis, 41, 87–102.
Kleinschmidt, A., Buchel, C., Zeki, S., Frackowiak, R. S. (1998). Human brain activity
during spontaneously reversing perception of ambiguous figures. Proceedings of the
Royal Society of London, Series B: Biological Sciences, 265, 2427–2433.
Lycan, W. G. (1996). Consciousness and Experience. Cambridge, MA: Bradford Books / MIT
Press.
Lycan, W. G. (1998). In defense of the representational theory of qualia (replies to Neander,
Rey and Tye). In J. Tomberlin (Ed.), Philosophical Perspectives, Vol. 12: Language, Mind
and Ontology (pp. 479–87). Atascadero, CA: Ridgeview Publishing.
Lycan, W. G. (2000). Representational theories of consciousness. In E. N. Zalta (Ed.),
The Stanford Encyclopedia of Philosophy. Summer 2000 edn.; extensively revised
and greatly expanded editions, 2004, 2006, 2012. http://plato.stanford.edu/entries/
consciousness-representational/.
Lycan, W. G. (2004). The superiority of HOP to HOT. In R. Gennaro (Ed.), Higher-Order
Theories of Consciousness (pp. 93–113). Amsterdam and Philadelphia: John Benjamins.
Lycan, W. G. (2008). Phenomenal intentionalities. American Philosophical Quarterly, 45,
233–252.
Lycan, W. G. (forthcoming). Block and the representation theory of sensory qualities. In A
Festschrift for Ned Block, ed. A. Pautz & D. Stoljar. Cambridge, MA: MIT Press.
Lyons, J. (2005). Clades, Capgras, and perceptual kinds. Philosophical Topics, 33, 185–206.
MacPherson, F. (2006). Ambiguous figures and the content of experience. Noûs, 40, 82–117.
Millikan, R. G. (1989). Biosemantics. Journal of Philosophy, 86, 281–297.
Millikan, R. G. (2000). On Clear and Confused Ideas. Cambridge: Cambridge University
Press.
Millikan, R. G. (2004). Varieties of Meaning. Cambridge, MA: Bradford Books / MIT Press.
Nickel, B. (2007). Against intentionalism. Philosophical Studies, 136, 279–304.
Noë, A. (2004). Action in Perception. Cambridge, MA: MIT Press.
Noë, A. (2005). Real presence. Philosophical Topics, 33, 235–264.
Orlandi, N. (2011a). Ambiguous figures and representationalism. Phenomenology and the
Cognitive Sciences, 10, 307–323.
Orlandi, N. (2011b). The innocent eye: Seeing-as without concepts. American Philosophical
Quarterly, 48, 17–32.
Orlandi, N. (forthcoming). Embedded seeing-as: Multi-stable visual perception without in-
terpretation. Philosophical Psychology.
Peacocke, C. (1983). Sense and Content. Oxford: Oxford University Press.
Peacocke, C. (1992). A Study of Concepts. Cambridge, MA: MIT Press.
Pettit, D. (2002). Why knowledge is unnecessary for understanding language. Mind, 111,
519–550.
Pylyshyn, Z. (2001). Visual indexes, preconceptual objects, and situated vision. Cognition,
80, 127–158.
 The Constituents of Perceptual Content and the Role of Perception
328   

Ricci, C., & Blundo, C. (1990). Perception of ambiguous figures after focal brain lesions.
Neuropsychologia, 28, 1163–1173.
Schellenberg, S. (2008). The situation-dependency of perception. Journal of Philosophy, 105,
55–85.
Schellenberg, S. (2011). Perceptual content defended. Noûs, 45, 1–37.
Segall, M. H., Campbell, D. T., & Herskovit, M. J. (1968). The influence of culture on visual
perception. In H. Toch and C. Smith (Eds.), Social Perception (pp. 139–44).
Siegel, S. (2006a). Which properties are represented in perception? In T. Gendler & J.
Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford University
Press.
Siegel, S. (2006b). Are Kind Properties Represented in Perception? Manuscript.
Siegel, S. (2007). How can we discover the contents of experience? Southern Journal of
Philosophy, 45, 127–142.
Siegel, S. (2010a). Do visual experiences have contents? In B. Nanay (Ed.), Perceiving the
World (pp. 333–68). Oxford: Oxford University Press.
Siegel, S. (2010b). The Contents of Visual Experience. Oxford: Oxford University Press.
Sorensen, R. (2008). Seeing Dark Things: The Philosophy of Shadows. Oxford: Oxford
University Press.
Travis, C. (2004). The silence of the senses. Mind, 113, 57–94.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: Bradford Books/MIT Press.
Wittgenstein, L. (1953). Philosophical Investigations. Oxford: Basil Blackwell.
Wollheim, R. (1996). Seeing-as, seeing-in, and pictorial representation. In Art and its
Objects (pp. 205–25). Cambridge: Cambridge University Press.
14

Phenomenal Intentionality and

Secondary Qualities
THE QUIXOTIC CASE OF COLOR
Terry Horgan

I maintain that the most fundamental kind of mental intentionally is phenomenal

intentionality. Phenomenal intentionality is inherent to phenomenal character,
and phenomenal character is itself an intrinsic and narrow feature of mentality;
hence, phenomenal intentionality itself is both intrinsic and narrow. Elsewhere,
often collaboratively with John Tienson or with Tienson and George Graham,
I have defended a specific version of the phenomenal intentionality thesis (e.g.,
Horgan and Tienson 2002; Horgan, Tienson, and Graham 2004; Graham,
Horgan, and Tienson 2007, 2009; Horgan and Graham 2009). I will call this the
Graham-Horgan-Tienson account (for short, the GTH account), in order to dis-
tinguish it from other accounts that also embrace the phenomenal intentionality
thesis (e.g., Strawson 1994, 2008; McGinn 1988; Siewert 1998; Loar 2002; Georgalis
2006; Pitt 2004; Farkas 2008; Kriegel 2011). The GTH account, as so far articu-
lated, makes a number of claims about the contents of perceptual experience.
However, Graham and Tienson and I have been somewhat noncommittal about
the secondary-quality content of perceptual experience, and about how such con-
tent connects to the metaphysics of color. In this chapter I will set forth and defend
a specific position concerning this matter, within the general framework of the
GTH account. I will focus specifically on visual experience, and in particular its
color-representing aspects, but I will maintain that the position is plausibly gen-
eralizable to all kinds of secondary-quality intentional content. I will also propose
a way to draw a general primary/secondary distinction that is plausible, theoret-
ically well motivated, and emerges naturally from my discussion of phenomenal
color-intentionality.
The plan of the chapter is as follows. In section 14.1, I will briefly describe the GTH
account, including its commitments concerning the nature of perceptual-experiential
intentional content. In section 14.2, I will describe one potential way of treating per-
ceptual color-content within the framework of the GTH account. This approach has
329
 The Constituents of Perceptual Content and the Role of Perception
330   

the consequence—given pertinent and uncontested scientific knowledge, and given

plausible constraints on a credible metaphysics of color—that there are no colors in
the world and that color-attributions in thought and language are always just false.
In section 14.3, I will advance a reason for seeking to eschew this hard-line posi-
tion, and to seek a version of the GTH approach that allows color-attributions to
be true. In section 14.4, I will set forth a version of the GTH approach that meets
these desiderata while still doing full justice to the color-representing aspects of
visual phenomenology. In section 14.5, I will explain why this proposed account
would be hard to implement within philosophical approaches to mental intention-
ality that do not acknowledge the reality of phenomenal intentionality and its status
as the fundamental form of mental intentionality. I will explain too why certain
approaches that acknowledge both the reality of phenomenal intentionality and its
fundamentality, while yet deviating from the GTH framework, also would have a
hard time implementing my proposed account. In section 14.6, I will elaborate the
proposal, by bringing it to bear on several instructive thought-experimental sce-
narios. In section 14.7, I will propose a way to distinguish between primary-quality
perceptual content and secondary-quality perceptual content, given the treatment
of color-content proposed in section 14.4; I will argue that the distinction is plau-
sible and well motivated, rather than being ad hoc. I will explain why a version of
the GTH account that incorporates this primary/secondary distinction entails that
the perceptual-experience-based beliefs of certain thought-experimental phenom-
enal duplicates of ordinary humans (e.g., brain-in-vat phenomenal duplicates) are
systematically false. In section 14.8, I will invoke some key ideas from the general
approach to truth that I have elsewhere advocated (sometimes collaboratively with
Matjaž Potrč or with Mark Timmons or with Robert Barnard) that I call contex-
tual semantics—the leading ideas being (a) that truth is semantic correctness under
contextually operative semantic-correctness standards, and (b) that such standards
are an implicit contextual parameter in judgment and discourse (Horgan 2001;
Horgan and Timmons 2002; Horgan and Potrč 2006, 2008; Barnard and Horgan
2006). I will explain how incorporation of these ideas allows certain claims that are
false in most contexts of judgment/discourse to be true in specific non-standard
contexts—e.g., (in a suitable philosophical or scientific context) the claim “Nothing
in the world is really colored,” or (in a different kind of context) the claim “The left/
right experiential invert veridically perceives his surrounding environment.” Finally,
in section 14.9, I will briefly explain how I propose to reconcile the whole discussion
with the metaphysical position, advocated by Potrč and myself that we call austere
realism (Horgan and Potrč 2006, 2008)—a position according to which the right
ontology does not include any “ordinary objects” such as tables, dogs, or rocks.
Before proceeding, let me briefly address two questions.1 First, what factors
are responsible for the question ‘Does perception have content?’ having recently

1
Both were posed by an anonymous referee, who asked that I address them.
 Phenomenal Intentionality and Secondary Qualities   331

become a “live” one? I think that one important factor is the following. Until late
in the twentieth century, a very dominant doctrine in the philosophy of mind
was what Graham and Tienson and I call “separatism”—the idea that the phe-
nomenal aspects of conscious mentality are non-intentional, whereas the inten-
tional aspects of conscious mentality lack proprietary phenomenal character.
(Here, ‘conscious’ essentially means not unconscious; fans of separatism typically
embrace Ned Block’s influential distinction (Block 1995) between being phenom-
enally conscious and being “access conscious.”) On an orthodox version of sepa-
ratism, sensory-perceptual experiences were regarded as lacking representational
content (despite having so-called “phenomenal character”), whereas the percep-
tual beliefs that arise from sensory-perceptual experience were regarded as lacking
proprietary phenomenal character. In recent years, however, separatism has been
called into question by two influential developments. One is the rise of so-called
“representationalism,” which (i) treats phenomenal character (including the phe-
nomenal character of sensory-perceptual experience) as being intentional, and (ii)
embraces one or another externalist, reductive, account of mental intentionality
itself.2 The other is the rise in popularity of what Uriah Kriegel (2013) has dubbed
“the phenomenal intentionality research program”; those within this camp, myself
included, contend that the phenomenal character of sensory-perceptual experi-
ence is inherently intentional, and that this intentionality is narrow and intrinsic
(rather than conforming to any externalist account).3
Second, what is at stake in this debate? In my view, quite a lot—but I will
briefly mention just a few issues. First is the nature of sensory-perceptual phe-
nomenal character itself. (I claim that it is intentional, and intrinsically so.)
Second is whether phenomenal character in general, and sensory-perceptual
phenomenal character in particular, can be “naturalized” via some kind of
externalist, reductive, psycho-semantics. (I claim that it cannot be naturalized
that way at least, and perhaps cannot be naturalized at all—although I remain
a “wannabe materialist.”) Third is whether, and if so how, sensory-perceptual
experience provides evidential justification for the perceptual beliefs to which
such experience gives rise. (I claim that typically it does, in part because
perceptual beliefs inherit their content fairly directly from the content of
sensory-perceptual phenomenal character.4)

2
See, for instance, Dretske (1995), Tye (1995), and Lycan (1995). In my view, when fans of this
approach appropriated the label ‘representationalism’, they were engaging (perhaps unwittingly) in
terminological tyranny.
3
This view is no less deserving of the label ‘representationalism’ than is so-called representationalism.
That’s why those who have appropriated this label are guilty of terminological tyranny.
4
Jack Lyons (2009) maintains that sensory-perceptual phenomenal character is either entirely
non-intentional or at best extremely thin in its intentional content, and hence that sensory-perceptual
experience cannot provide evidential justification for perceptual beliefs. He defends a non-evidentialist,
reliabilist account of the justificatory status of perceptual beliefs. In Horgan (2011) I argue that Lyon’s
approach presupposes an objectionable form of separatism.
 The Constituents of Perceptual Content and the Role of Perception
332   

14.1. The GTH Framework: Phenomenal Intentionality and

Externalistic Intentionality

Central to the GTH approach is the role of phenomenology or phenomenal con-

sciousness, by which we mean those aspects of one’s mental life such that there is
“something it is like” to undergo them. Briefly, the position goes as follows.
Phenomenology is narrow: it is not constitutively dependent upon anything
“outside the head” (or outside the brain) of the experiencing subject. Indeed, it is
not constitutively dependent upon anything outside of phenomenal conscious-
ness itself; in this sense, it is intrinsic. Your phenomenology, being narrow and
intrinsic, supervenes at least nomically upon physical events and processes within
your brain.
Phenomenology is also richly and pervasively intentional: there is a kind of
intentionality that is entirely constituted phenomenologically (viz., phenomenal
intentionality), and it pervades our mental lives. Among the different aspects of
phenomenal intentionality are the following. First, there is the phenomenology of
perceptual experience: the enormously rich and complex what-it’s-like of being
perceptually presented with a world of apparent objects, apparently instantiating
a rich range of properties and relations—including one’s own apparent body, ap-
parently interacting with other apparent objects which apparently occupy various
apparent spatial relations as apparently perceived from one’s own apparent-body
centered perceptual point of view. Second, there is the phenomenology of
agency: the what-it’s-like of apparently voluntarily controlling one’s apparent body
as it apparently moves around in, and apparently interacts with, apparent objects
in its apparent environment. (This is a central focus of the present chapter, of
course.) Third, there is conative and cognitive phenomenology: the what-it’s-like
of consciously (as opposed to unconsciously) undergoing various occurrent prop-
ositional attitudes, including conative attitudes like occurrent wishes and cogni-
tive attitudes like occurrent thoughts. There are phenomenologically discernible
aspects of conative and cognitive phenomenology, notably (i) the phenomenology
of attitude type and (ii) the phenomenology of content. The former is illustrated
by the phenomenological difference between, for instance, occurrently hoping
that Hillary Clinton will be elected U.S. President and occurrently wondering
whether she will be elected—where the attitude-content remains the same while
the attitude-type varies. The phenomenology of content is illustrated by the phe-
nomenological difference between occurrently thinking that Hillary Clinton will
be elected U.S. President and occurrently thinking that she will not be elected—
where the attitude-type remains the same while the content varies.
Since phenomenal intentionality is entirely constituted phenomenologically,
and since phenomenology is narrow, phenomenal intentionality is narrow too.
Hence, there is an exact match of phenomenal intentionality between yourself and
your brain-in-vat (BIV) physical duplicate. This exactly matching, narrow, inten-
tional content involves exactly matching, phenomenally constituted, narrow truth
 Phenomenal Intentionality and Secondary Qualities   333

conditions. But whereas the narrow truth conditions of your own beliefs are largely
satisfied, those of your BIV physical duplicate’s matching beliefs largely fail to be
satisfied; thus, the BIV’s belief system is systematically nonveridical.
On the other hand, an exact match in narrow content between your own
intentional mental states and the corresponding states in your BIV physical
duplicate does not require or involve an exact match in referents (if any) of all the
various matching, putatively referring, thought-constituents. For instance, some
of your own occurrent thoughts that you would express linguistically using par-
ticular proper names—say, the thought that Michele Bachmann is not a genius—
involve singular thought-constituents whose referents (if any) are determined
partly in virtue of certain external relations that obtain between you and those
referents. Thus, your occurrent thought that Michele Bachmann is not a genius
involves a singular thought-constituent that purports to refer to a particular spe-
cific person (viz., Michele Bachmann); its actually referring, and its referring to
the specific individual to whom it does refer, depends upon there being certain
suitable external relations linking you to a unique eligible referent (viz., Michele
Bachmann). A Twin-Earthly physical duplicate of yourself, in a Twin-Earthly
duplicate local environment, would refer to a different individual (viz., Twin-
Michele) via the corresponding singular thought-constituent of the corresponding
occurrent thought. And, in the case of your BIV physical duplicate, the matching
singular thought-constituent fails to refer at all, because the BIV does not bear
suitable externalistic relations to any suitably reference-eligible individual in its
own actual environment. (Parallel remarks apply to thought-constituents that pur-
port to refer to natural kinds, such as the thought-constituent that you yourself
would express linguistically with the word ‘water’.)
For mental states involving thought-constituents for which reference depends
upon externalistic factors, there are two kinds of intentionality, each involving its
own truth conditions. First is the kind of intentionality already mentioned above:
phenomenal intentionality, with truth conditions that are phenomenally consti-
tuted and narrow. Second is externalistic intentionality, with wide truth conditions
that incorporate the actual referents (if any) of the relevant thought-constituents.
Your own thought that Michele Bachmann is not a genius, and the corresponding
thoughts of your BIV physical duplicate and your Twin Earth physical duplicate,
have matching phenomenal intentionality, with matching truth conditions. (These
truth conditions are satisfied in your case and in the case of your Twin Earth du-
plicate, but not in the case of your BIV duplicate.) On the other hand, your own
thought that Michele Bachmann is not a genius and your Twin Earth duplicate’s
corresponding thought do not have matching externalistic intentionality, because
the externalistic truth conditions of these respective thoughts do not match: the
truth value of your own thought depends upon the intelligence level of Michele
Bachmann, whereas the truth value of your Twin Earth duplicate’s corresponding
thought depends upon the intelligence level of an entirely different individual, viz.,
Twin-Michele. (Each thought’s wide truth conditions are indeed satisfied.) As for
 The Constituents of Perceptual Content and the Role of Perception
334   

your BIV duplicate’s thought, it lacks externalistic intentionality and wide truth
conditions, because its singular thought-constituent which purports to refer to a
person called ‘Michele Bachmann’ does not actually refer at all.
Our account rests heavily and essentially upon two key contentions.
First, mental reference to many properties and relations—including var-
ious spatiotemporal-location properties, shape-properties, size-properties,
artifact-properties, and personhood-involving properties—is wholly constituted
by phenomenology alone. Even systematically nonveridical phenomenology, as in
the case of the BIV, provides reference-constituting experiential acquaintance with
such properties and relations. It makes no difference to such experiential acquaint-
ance with such properties—and hence it makes no difference to mental reference
to such properties—whether or not the properties with which one becomes expe-
rientially acquainted are ever actually instantiated in one’s ambient environment.
Second, in the case of thought constituents whose reference (if any) depends
constitutively upon certain externalistic elements, the mechanisms of reference-
fixation crucially involve phenomenally constituted grounding presuppositions
(as we call them). Thus, phenomenal intentionality is more basic than exter-
nalistic intentionality, since the latter depends in part upon the former (as well
as depending, in part, upon externalistic factors). Suppose, for example, that
you have an occurrent thought that you could express linguistically by saying
“That picture is hanging crooked,” where the singular thought-constituent
expressible linguistically by ‘that picture’ purports to refer to a picture on the
wall directly in front of you. This thought-content involves phenomenally consti-
tuted grounding presuppositions that must be satisfied in order for the singular
thought-constituent to refer: roughly, there must be an object at a certain location
relative to yourself (a location that you could designate linguistically by a specific
use of the place-indexical ‘there’), this object must be a picture, there must not be
any other picture at that location that is an equally eligible potential referent of
‘that picture’, and this object must be causing your current picture-experience. If
these grounding presuppositions are satisfied by some specific concrete particu-
lar in your ambient environment—some particular object that is a picture and is
uniquely suitably located—then your singular thought-constituent thereby refers
to that very object. Which object your thought-constituent refers to, if any, thus
depends jointly upon two factors, one phenomenally constituted and one external-
istic: on the one hand, the phenomenally constituted grounding presuppositions,
and on the other hand, the unique actual object in your ambient environment that
satisfies those presuppositions.

14.2. Colors and the GTH Framework I: The Hard-Line Approach

The GTH approach, as described in the preceding section, leaves open various
questions about phenomenal color-intentionality and about the metaphysics of
 Phenomenal Intentionality and Secondary Qualities   335

color. Are color-properties among those with which one is immediately acquainted
in experience in such a way that one can forthwith refer to these properties men-
tally even if one is an envatted brain? Or is it rather the case that mental reference
to color-properties incorporates a constitutive externalistic aspect, so that one’s
BIV’s thoughts purporting to refer to color-properties fail to refer at all? Are color-
properties ever actually instantiated in the world? If they are instantiated, does
color-experience represent their real nature?
Before taking up such questions, let me introduce a distinction that will prove
useful below, between two kinds of intentional content. On one hand is presenta-
tional content; it accrues directly to perceptual experience itself, whether or not
the experiencing subject forms a perceptual belief corresponding to that content.
(For example, when one is presented with an instance of the Müller-Lyer illusion
in a situation where one firmly believes that the two horizontal lines are the same
length, one of those lines nonetheless still looks longer than the other; thus, one’s
perceptual presentational content is as-of one horizontal line being longer than
the other, even though one does not form the corresponding perceptual belief.)
On the other hand is judgmental content: it accrues to full-fledged judgments,
including those that arise directly from perceptual experience.5
The distinction between presentational intentional content and judgmental
content is orthogonal to the distinction between phenomenal intentionality and
externalistic intentionality. Perceptual experience can have both phenomenal
presentational content (which is fully constituted by perceptual-experiential phe-
nomenology alone) and externalistic presentational content (which is constituted
in part by perceptual-experiential phenomenology and in part by suitable exter-
nalistic connections). An example of the latter would be a visual-presentational
experience as-of Sarah Palin—an experience which could occur even if one lacks
the corresponding belief. (Perhaps one is watching Saturday Night Live on televi-
sion, and one believes that one is really seeing Tina Fey.) Likewise, mental states
with judgmental content—e.g., occurrent beliefs—can have both phenomenal
judgmental content and externalistic judgmental content.
With the distinction between presentational and judgmental content at
hand, one potential way of extending the GTH approach to address phenomenal
color-intentionality and the metaphysics of color would be to embrace the fol-
lowing claims.
(1) 
Phenomenology directly acquaints the experiencing subject with
color-properties, in such a way that the presentational phenomenal

5
For more on the distinction between presentational content and judgmental content, see
Horgan (2007). I take this distinction to be orthogonal to the distinction between conceptual and
non-conceptual content. I understand the latter distinction in terms of whether or not the given aspect
of presentational content is something for which one has concepts. Judgmental content is conceptual;
but presentational content can be conceptual, or nonconceptual, or a mixture of both. (And judgmental
content can include indexical conceptual elements).
 The Constituents of Perceptual Content and the Role of Perception
336   

content of the subject’s experiences involves phenomenally constituted

reference to those properties.
(2) 
Color properties have certain characteristics that are directly given in
the presentational phenomenal character of visual experience, including
these: (i) when instantiated, they are instantiated on the surfaces of external
objects; (ii) when instantiated, they are intrinsic, objective, mind-independent,
non-dispositional properties of the objects that instantiate them; (iii) their
nature is manifest in the way they are experientially presented, and thus they
do not have non-manifest essences; (iv) they are sensuous (or phenomenal),
in the sense that that there is something they are like.6
(3) No such properties are ever actually instantiated.
Hence, from (1)‒(3),
(4) The color-attributing aspects of people’s visual experiences are systemat-
ically non-veridical.
(5) When a perceptual color-attributing judgment coincides directly with
the color-attributing presentational content of one’s visual experience,
the color-properties attributed by the judgment are identical to those
attributed by the visual experience itself.
Hence, from (4) and (5),
(6) Color-attributing judgments are systematically false.
This hard-line approach has two notable attractions. First, it acknowledges the real
phenomenological character of visual color experience. Colors are certainly not
presented in experience as being, say, mere dispositions of external objects to cause
certain subjective states of the experiencing perceiver. On the contrary, they are
presented as intrinsic, sensuous properties of external objects themselves. Nor are
colors presented in experience as being, say, certain light-reflectance properties; on
the contrary, perceptual experience does not traffic in such scientific-theoretical
modes of presentation (although thought about light-reflectance properties pre-
sumably does). Rather, they are presented as properties whose manifest nature is
just their intrinsic, sensuous, what-they-are-like-ness.
A second attraction is that the approach avoids scientifically implausible,
theoretically unparsimonious, ontological extravagance. It does so by denying
that the color-properties that are presented in visual-perceptual experience are
ever really instantiated in the world. Such putative properties are extraordinarily

6
Talk of what color experiences are like is derivative: color experiences present colors as
something-they-are-like properties—i.e., sensuous or phenomenal properties—of external objects. The
experientially manifest nature of these properties constitutes what they are like (as given in experience).
See Maund ([1997] 2006)—especially section 3, “The Natural Concept of Color.” What Maund calls the
“natural concept” of color picks out what, on my proposal, constitutes presentational color-properties.
 Phenomenal Intentionality and Secondary Qualities   337

queer-looking from the scientific perspective—a fact that was clear right at the
dawn of modern science, and has remained clear ever since. Science has no the-
oretical need for them and no natural way of incorporating them. The scientific
worldview has brought about a “fall from Eden” (as David Chalmers puts it, in
Chalmers 2006) as far as experientially presented color-properties are concerned.7

14.3. A Reason to Eschew the Hard Line

Despite these important attractions of the hard-line approach there is a powerful-

looking reason to avoid it if possible: viz., that people routinely and confidently
regard everyday color-judgments as quite literally true, and also as extremely well
warranted epistemically. Even philosophers typically hold such views, and often
do so even when doing philosophy—especially when the philosophy focuses on
issues other than the content of color-ascriptions or the metaphysics of color. (For
instance, it is a widespread view in epistemology that perceptual color-experiences
provide strong prima facie warrant for beliefs about the colors of objects in one’s
immediate environment, and that such beliefs are frequently true.) All else equal,
a philosophical position is better to the extent that it accommodates commonsen-
sical beliefs that are widely held and are widely regarded (including by epistemolo-
gists) as very strongly warranted.
One way to save the truth of color-ascriptions would be to deny one or both
of claims (1) and (2) above, concerning the presentational content of color-expe-
riences. But this approach is not credible, because claims (1) and (2) express facts
about the phenomenology of visual perception that are introspectively self-evi-
dent. (Or so I maintain, and will henceforth assume. I realize that many philoso-
phers are prepared to deny claims about the character of experience that common
sense takes to be self-evident.)
Another way to save the truth of color-ascriptions would be to embrace a
fairly radical revisionism: redefine color-words so that they refer to certain sci-
entifically respectable properties—perhaps, for instance, dispositions to cause
color-experiences, or perhaps certain light-reflectance properties. But this
approach is not credible either, insofar as it purports to save the truth of ordinary
color-ascriptions. For the approach really amounts to a proposal to change the
meaning of these color-ascriptions—and, indeed, to change it quite significantly.
The account is really a form of error theory concerning ordinary color-ascriptions,
even though it provides a semantic-revisionary proposal for giving new meanings
to color words.

7
As best I can tell from the philosophical and scientific literature about color that I am familiar
with, the extant position concerning colors and color-content that is closest to the one I will advocate
below is the position set forth in Chalmers (2006). Commenting on similarities and differences
between that position and mine is a task I leave for another occasion.
 The Constituents of Perceptual Content and the Role of Perception
338   

I will instead propose a position, formulable as an extension of the core

GTH account of phenomenal intentionality, with the following features: (i) it
embraces claims (1) and (2) concerning the presentational content of visual
color-experience, (ii) it retains the ordinary meaning of everyday color-judgments
and color-ascriptions, and (iii) it vindicates the common-sense belief that eve-
ryday color-judgments and color-ascriptions are very often both true and episte-
mically well-warranted.

14.4. Colors and the GTH Framework II: Colors as

Response-Dependent Dispositional Properties

My proposal concerns two kinds of phenomenal intentionality—specifically,

(1) the phenomenally intentional color-content of visual experiences, and (2) the
phenomenally intentional content of corresponding color-ascribing judgments.8
The first is a form of presentational content, and the second is the corresponding
form of judgmental content. The pertinent notion of correspondence is this: the
judgment pertains to the visually presented ambient situation, and inherits its con-
tent directly from the visual presentation of that situation.
A natural thought, with considerable prima facie plausibility, is this: such
“direct inheritance” of judgmental color-content from presentational color-
content is just the identity of color-content. Nonetheless, my proposal denies
this. Instead, I propose the following: The color-presenting constituents of visual
experience directly acquaint the experiencing subject with presentational color-
properties (as I will call them)—properties with the features described by claims
(1) and (2) above. This direct-acquaintance relation constitutes mental reference
to these presentational color-properties. Moreover, claim (3) above is true of
these properties: no presentational color-properties are ever instantiated. (To
adapt Chalmers’ terminology, our world is not an “Edenic” world.)
But judgmental color-content—including the judgmental content that
is “directly inherited” from presentational color-content—works differently.
Color-ascribing judgments deploy thought-constituents that are semantically
governed by (phenomenologically constituted) grounding presuppositions; these
thought constituents thereby purport to refer to color-properties in a way that
constitutively incorporates certain externalistic factors. The grounding presup-
positions combine with facts about the agent’s external environment to jointly
fix the referents (if any) of the judgment-constituents that purport to refer to
color-properties. (I will call those referents judgmental color-properties.) The
grounding presuppositions work disjunctively, more or less as follows:

8
Since the proposal concerns phenomenal intentionality, it applies to a brain-in-vat duplicate
of an ordinary human experiencer in the same way it applies to the human experiencer. More on this
below, in section 14.6.
 Phenomenal Intentionality and Secondary Qualities   339

(i) if presentational color-properties are actually instantiated in one’s am-

bient environment, then judgmental color-properties are identical to the
corresponding presentational color-properties; however,
(ii) if presentational color-properties are not actually instantiated in one’s
ambient environment, then each judgmental color-property is identical
with a disposition to reliably cause in oneself, when instantiated by an
object one perceives in suitably favorable viewing circumstances, a visual
experience as-of that object’s instantiating a specific presentational color
property.

These grounding presuppositions provide a constitutive role for externalistic fac-

tors to play, in the fixation of the reference of the judgment-constituents that pur-
port to refer to color-properties. In an Edenic world, the referents of color-referring
judgment-constituents turn out to be just identical with the corresponding presen-
tational color-properties. But in a non-Edenic world like ours—a world in which
presentational color-properties are never instantiated—the referent of a given
color-referring judgment-constituent instead turns out to be a Lockean disposi-
tion: a disposition of an object, when perceived under suitably favorable viewing
circumstances, to cause a visual-perceptual presentation of itself as-of instantiat-
ing a certain specific presentational color-property. (A perceived object possesses
the property judgmental redness, for instance, just in case the object is disposed to
cause visual-perceptual presentations of itself, under suitably good viewing condi-
tions, as-of instantiating the property presentational redness.)
Why go Lockean in the formulation of clause (ii), rather than identifying
judgmental color-properties with (say) certain light-reflectances, or certain com-
binations of light-reflectances, or certain physical properties that are the catego-
rical bases of the Lockean dispositions? My principal reasons are two. First, this
accommodates the multiple physical realizability of judgmental colors. (There is
considerable empirical evidence for such multiple realizability.) Second, Lockean
dispositions are reasonably uniform, reasonably natural, properties—rather
than being un-lovely, un-natural, disjunctive properties. However, the generic
treatment of color-content that I am recommending—viz., allowing judgmental
color-properties to differ from presentational color-properties—could, of course,
be implemented in some non-Lockean way. Moreover, even if one goes for a
Lockean implementation, there might be reasons to refine conditions (i) and (ii)
in one way or another; I am not firmly wedded to this specific characterization of
the grounding presuppositions governing color-referring judgment-constituents.
This proposed account of color-content meets the desiderata I set out at the
end of section 14.3. It allows everyday color-attributions to be true: objects really
do instantiate the color-properties attributed to them in people’s color-judgments
and in the statements that express those color-judgments. It allows these judg-
ments to be epistemically well warranted: since judgmental color-properties
are response-dependent dispositional properties, color-attributing judgments
 The Constituents of Perceptual Content and the Role of Perception
340   

normally are very well warranted by the visual-perceptual experiences that gen-
erate them, viz., visual-perceptual experiences as-of the instantiations of the cor-
responding, presentational, color-properties. And yet this account also honors the
introspectively evident phenomenology of visual-perceptual experience by recog-
nizing that such experience presents color-properties not as dispositional at all,
but rather as intrinsic, objective, features of the surfaces of external objects.
Admittedly, this account of color-content is an error theory of sorts: it treats
presentational color-content as systematically non-veridical. Visual experience
presents us with an apparently Edenic world, an apparent world in which pres-
entational color-properties are indeed instantiated on the surfaces of external
objects—and there is excellent scientific reason to believe that our world is not ac-
tually Edenic at all. But there is no serious theoretical cost in embracing this kind
of error theory, because doing so just amounts to embracing that scientific moral
while still acknowledging the actual, introspectively obvious, phenomenological
character of visual color-experience. It would be much more theoretically costly
to embrace an error theory asserting that everyday color-attributing judgments,
and the statements that express such judgments, are systematically and radically
false. But that high cost is precisely what my proposal avoids via the device of dis-
tinguishing between presentational color-content and judgmental color-content.
Admittedly too, this account goes contrary to the idea, which initially seems
very plausible, that the content inherited by color-judgments from the corresponding
color-experiences is literally identical to the presentational color-content of the
experiences themselves. But that idea, although quite natural from the perspective
of naïve common sense, has an epistemic status that is deeply hostage to how things
go in empirical science. Naïve common sense also cleaves very strongly—more
strongly, really—to the idea that visual color experience normally provides over-
whelmingly strong epistemic justification for everyday color-attributing judgments.
Given the pertinent science, something has to give, as far as naïve common sense
is concerned. I submit that on balance, in terms of comparative theoretical costs
and benefits, what should give is the naïve common-sensical belief that presenta-
tional color-content and judgmental color-content are identical. Once that belief is
jettisoned, everyday color-judgments turn out to be frequently true after all: objects
really do possess judgmental color-properties, and these are the properties that are
attributed to them by color-judgments.

14.5. Underscoring the Role of the GTH Framework

It bears emphasis why, and how, the availability of the account of colors and
color-content here proposed depends upon the wider GTH framework. For some
time now in philosophy of mind, the dominant approaches to mental intentionality
have been one or another version of a view which Graham and Tienson and I call
strong externalism: roughly and generically, the view that all mental intentionality
 Phenomenal Intentionality and Secondary Qualities   341

depends constitutively on certain external connections (e.g., causal, and/or covari-

ational, and/or historical, and/or evolutionary) between a mental subject and that
subject’s wider environment. Most any version of strong externalism is apt to have
trouble making sense of what I call presentational color-content, because presen-
tational colors are properties that are never instantiated here in the actual world.
Presentational colors are not out there in the actual external environment—which
they would have to be in order to be eligible, according to strong externalism, to
become part of the content of visual experience. By contrast, the GTH frame-
work appeals to the inherent, intrinsic, intentionality of phenomenology: the
phenomenological character of visual experience inherently presents the experi-
encing subject with apparent instantiations of presentational color-properties by
external objects (or by apparent external objects, if the experiencing subject is
an envatted brain), and that presentational acquaintance constitutes mental ref-
erence to presentational color-properties. Under the GTH framework, it doesn’t
matter that those properties are never actually instantiated in the subject’s external
environment.
A growing minority of contemporary philosophers repudiates strong
externalism and instead embraces the thesis that the fundamental kind of mental
intentionality is phenomenally constituted and narrow. Some members of this
minority group construe phenomenal intentionality somewhat differently than it
is construed in the GTH framework. It remains to be seen whether—and if so,
how—various alternative approaches to phenomenal intentionality can embrace
my recommended account of color-content, or something like it. On some alter-
native versions, it might be difficult to make sense of the notion of a presenta-
tional color-property. For instance, the version embraced by Uriah Kriegel (2011)
treats phenomenal intentionality as (what I would call) hyper-intrinsic: this ap-
proach repudiates the idea that there can be phenomenologically constituted ac-
quaintance with properties that are instantiated, or are apparently instantiated, by
external objects (or are apparently instantiated by apparent external objects, in the
case of an envatted brain). So much the worse, I say, for such alternative construals
of phenomenal intentionality.

14.6. Color-Spectrum Inversion, Left-Right Inversion, and Envatment

It will be useful, by way of further elaboration of my proposal, to discuss it in

relation to several thought-experimental scenarios. First, suppose that within
the human population there is a small subpopulation of lifelong color-spectrum
inverts—people whose visual experiences are always spectrally inverted in com-
parison to those of normal humans. The inverts, since they learn color-words as
part of public language, apply these words to external objects in the same way
that visually normal humans do. Neither the inverts nor anybody else is aware
of the differences in color experiences between the inverts and the normals, and
 The Constituents of Perceptual Content and the Role of Perception
342   

neither the inverts nor anybody else are even aware that there are inverts among
the human population. (The differences are detectable in principle, via the right
kinds of third-person monitoring of neural activity in the visual cortex of an invert
while presenting the invert with objects of various colors; but so far no such moni-
toring has been carried out.) Given my account of presentational color-content
and judgmental color-content, what should one say about the inverts, concerning
the content and the truth values of their color judgments? And what should one
say concerning the meaning that public-language color-words have for them?
Concerning presentational contents: The inverts’ presentational contents
are different from, and are the spectrally inverted counterparts of, those of nor-
mal humans. When an invert sees a green object under suitable illumination, the
object is visually presented as instantiating the property presentational red.9 This
presentational content is nonveridical, because presentational color-properties are
not instantiated in the world. But of course the same is true for the presentational
color-content of the visual experiences of normal humans.
Concerning judgmental contents: For an invert, the phenomenologically consti-
tuted grounding presuppositions governing judgmental color content interact con-
stitutively with externalistic factors in such a way that the invert’s color-judgments
attribute Lockean dispositions to external objects that are inverted relative to the
Lockean dispositions that are attributed by the color-judgments of normal humans.
When an invert sees a green object under suitable illumination, and the invert forms
a judgment that he/she expresses by applying the word ‘green’ to the object, the judg-
ment attributes to the object the property being disposed to reliably cause in oneself, in
suitably favorable viewing circumstances, experiential visual experiences with presen-
tational color-content as-of presentational red. The invert’s judgment is true.
Concerning the meanings of public-language color-words: One can cor-
rectly say either of two things about this, but not in the same breath. On one
hand, there is an important sense in which the words have different meanings
as employed by the inverts from the words’ meanings as employed by normal
humans: viz., the words as used by the inverts express judgments with different
judgmental color-content than judgments that normal humans express with
those words. (The color-ascribing sentences are true given what they mean by the
color words they are employing, but are false given what normal humans mean
by those words.) On the other hand, there is also an important sense in which
the words have the same meanings as employed by the inverts: viz., the publicly
observable linguistic behavior of the inverts fully conforms to that of the wider
human population, insofar as the use of color-words is concerned. (The inverts’
color-ascribing sentences are true, because these sentences conform with compe-
tent practice in deploying color-words; in that respect, the sentences are correctly

9
For expository convenience I am here pretending, as is common in philosophical discussion of
spectral-inversion scenarios, that red and green are spectral inverses of one another.
 Phenomenal Intentionality and Secondary Qualities   343

affirmable.10) Which way of using ‘meaning’-talk is appropriate, vis-à-vis color

words as employed by inverts, will depend on context.
It is instructive to compare the color-inversion scenario to a different
thought-experimental scenario, this time involving primary qualities with which
one is directly acquainted with in perceptual experience. Suppose that within the
human population there is a subpopulation of lifelong left-right inverts—people
whose experiential presentations as-of left and right are systematically inverted (in
all pertinent sensory modalities) relative to those of normal humans. When a dog,
say, is positioned to the front-left of such an invert, the dog is visually presented
as being to the front-right. When the dog barks, the sound is auditorily presented
as emanating from the front-right. When the invert reaches out with the left
hand (in the front-left direction) to pat the dog, the invert’s visual and kinesthetic
experience is as-of reaching out with the right hand (in the front-right direction)
to pat the dog. And so forth. The inverts, since they learn direction-words as part
of public language, apply these words to external objects in the same way that
visually normal humans do. Neither the inverts nor anybody else is aware of the
differences in left-right experiences between the inverts and the normals, and nei-
ther the inverts nor anybody else is even aware that there are inverts among the
human population. (The differences are detectable in principle, via the right kinds
of third-person monitoring of an invert’s neural activity while the invert moves
about in the world and deploys direction-talk; but so far no such monitoring has
been carried out.) Given the GTH framework concerning the nature of mental
intentionality, what should one say about the inverts, concerning the content and
the truth values of their judgments about left and right? And what should one say
concerning the meaning that public-language direction-words have for them?
Concerning presentational contents: Are the inverts’ presentational contents
are different from, and are the left-right inverted counterparts of, those of normal
humans? When an invert experientially interacts with an object as-of to the front
left, the interaction is experienced presentationally as being to the front right. This
presentational content is nonveridical, because although (a) self-oriented direc-
tional relations like left-front and right-front are indeed instantiated in the world,
nonetheless (b) the invert’s experiences systematically misrepresent these relations
by systematically presenting as to-the-left (or to-the-right) what is actually to-the-
right (or to-the-left). In normal humans, by contrast, the presentational-direction
content of perceptual/kinesthetic experience is veridical.
Concerning judgmental contents: For both normal humans and left-right
inverts, the presentational content of perceptual/kinesthetic experience provides
the experiencer with direct acquaintance with self-oriented directional rela-
tions—acquaintance that constitutes mental reference to these relations. Mental

10
More below, in sections 14.8 and 14.9, on truth as correct affirmability under contextually
operative semantic standards.
 The Constituents of Perceptual Content and the Role of Perception
344   

reference to such relations is thus a matter of purely phenomenal intentionality;

no externalistic factors figure constitutively in securing mental reference to these
primary qualities. Thus, when a directional judgment arises directly from per-
ceptual/kinesthetic experience, the content of that judgment (its judgmental
content) is identical to the presentational directional content of the experience
itself. And of course, such presentational directional relations really are instanti-
ated in the world—unlike presentational color-properties. The upshot is that the
left-judgments and right-judgments of normal humans are true, whereas those of
left-right inverts are systematically false. The inverts systematically misjudge the
world, as far as left and right are concerned.
Concerning the meanings of public-language words for right and left: One can
correctly say either of two things about this, but not in the same breath. On one hand,
there is an important sense in which the words have different meanings as employed
by the left-right inverts from the words’ meanings as employed by normal humans:
viz., the words as used by the inverts to express judgments with different judgmental
direction-content than judgments that normal humans express with those words.
(The direction-ascribing sentences are false given what they mean by the direction
words they are employing, but are true given what normal humans mean by those
words.) On the other hand, there is also an important sense in which the words
have the same meanings as employed by the inverts: viz., the publicly observable lin-
guistic behavior of the inverts fully conforms to that of the wider human population,
insofar as the use of words like ‘left’ and ‘right’ is concerned. (The inverts’ direction-
ascribing sentences are true, because these sentences conform with competent prac-
tice among the normals in deploying direction-words; in that respect, the sentences
are correctly affirmable.) Which way of using ‘meaning’-talk is appropriate, vis-à-vis
direction words as employed by left-right inverts, will depend on context.
Turn now to envatted-brain scenarios. A major advantage of the GTH frame-
work is it vindicates the strong pre-theoretic intuition that an envatted brain that
is a duplicate of an ordinary human brain (say, your brain), in which neural pro-
cesses occur throughout its life that exactly match the neural processes in the or-
dinary brain, will have a mental life that exactly matches the mental life of the
ordinary human. Its phenomenal intentionality is exactly the same.
Given my proposed treatment of presentational color-content and judg-
mental color-content, however, a certain worry now arises. The envatted brain
is hooked to a computer that continuously monitors the physical activity of the
brain’s motor neurons and continuously stimulates the brain’s sensory neurons, in
such a way as gives the brain ongoing experience as of being an embodied human
being that actively interacts with the ambient environment. Thus, certain states of
the computer systematically dispose the computer to cause the brain to undergo
experiences as-of the instantiation, for instance, of the property presentational red-
ness. Often when the brain undergoes such an experience, it forms on that basis
a judgment that there is something red in front of it. The worry is this: Doesn’t
my account of judgmental color-content require me to say that such a belief is
 Phenomenal Intentionality and Secondary Qualities   345

true, even though neither the computer nor any of its components are red? After
all, doesn’t the computer, when it is in its current monitoring-state, have the very
Lockean-dispositional property that constitutes judgmental redness?
Well, no. Look again at the formulation, in section 14.4, of the proposed dis-
junctive grounding presuppositions that I said govern judgmental color-content.
Clause (ii) says that the pertinent Lockean dispositions are properties possessed by
objects one perceives in one’s ambient environment. The envatted brain doesn’t per-
ceive genuine objects at all in its ambient environment—even though it has system-
atically nonveridical experiences as-of perceiving external objects. In particular, it
does not perceive the computer that is generating its inputs, and it does not perceive
any states of that computer either. Thus, even though the computer, when in certain
states, is indeed disposed to generate in the brain experiences as-of the instantiation
of presentational redness by apparent objects in the envatted brain’s apparent envi-
ronment, this does not suffice to make it the case that the computer, when in that
state, instantiates the property of judgmental redness. In order for the computer to
instantiate that property, it would need to be disposed to reliably cause experiences
as-of presentational redness in an experiencer who visually perceives it (and does so
under suitably favorable viewing circumstances). The envatted brain doesn’t visu-
ally perceive anything—even though it seems to itself to see an external world.

14.7. The Primary/Secondary Distinction

The remarks in the preceding section about the differences between spectral-
inversion scenarios and left-right inversion scenarios suggest the following way of
drawing a general distinction between primary qualities and secondary qualities
within the GTH framework. Primary qualities have these features:
(a) one is directly acquainted with presentational primary qualities in sen-
sory-perceptual experience, via apparent instantiations of them by ap-
parent objects in one’s apparent ambient environment;
(b) this experiential acquaintance constitutes presentational mental refer-
ence to them;
(c) presentational primary qualities are actually instantiated in the world;
and
(d) presentational primary qualities are identical to the corresponding prop-
erties one refers to in judgment (i.e., judgmental primary qualities).
Secondary qualities, on the other hand, have these features:
(a) one is directly acquainted with presentational secondary qualities in
sensory-perceptual experience, via apparent instantiations of them by
apparent objects in one’s apparent ambient environment;
 The Constituents of Perceptual Content and the Role of Perception
346   

(b) this experiential acquaintance constitutes presentational mental refer-

ence to them;
(c*) presentational secondary qualities are not actually instantiated in the
world;
(d*) presentational secondary qualities are not identical to the corresponding
properties one refers to in judgment (i.e., judgmental secondary proper-
ties); rather, judgmental secondary properties are Lockean dispositions
of perceived objects to reliably cause, in suitably favorable perceptual
circumstances, experiences as-of those objects instantiating the corre-
sponding presentational secondary qualities.
The distinction between primary and secondary qualities has been with us since
the dawn of modern science, and has been perennially plausible. It is a theoretical
advantage of the account I have proposed of the contents of color-experiences and
color-judgments that this account suggests a natural and attractive way of drawing
the primary/secondary distinction.

14.8. Contextual Semantics and the Judgmental Content of

Perceptual Belief

In this section and the next, I will situate the preceding discussion within a wider
philosophical perspective that incorporates certain of my views about matters of
truth and ontology. In a number of writings over the years, sometimes collabora-
tive, I have advocated a general conception of truth that I call contextual seman-
tics (Horgan 2001; Horgan and Timmons 2002; Horgan and Potrč 2006, 2008;
Barnard and Horgan 2006). Some leading ideas are these:

(1) Truth is semantically correct affirmability, under contextually operative

semantic-correctness standards.
(2) Standards for semantic correctness are contextually variable; they
involve contextually appropriate settings of implicit parameters.
(3) Often (indeed, very often) the contextually operative
semantic-correctness standards involve implicit parameter-settings
under which the “referential apparatus” of one’s thought and discourse
does not carry ontological commitment to the posited items. In these
contexts, truth (i.e., semantically correct affirmability) is an indirect
form of correspondence between thought/language and the world.
(4) Sometimes (e.g., in contexts of serious ontological inquiry) the
contextually operative semantic-correctness standards involve
implicit parameter-settings under which the referential apparatus
of one’s thought and discourse does carry ontological commitment
to the posited items. In these contexts, truth is a direct form of
correspondence between thought/language and the world.
 Phenomenal Intentionality and Secondary Qualities   347

To offer one illustration of truth as indirect correspondence, the statement ‘There

are exactly three public universities in Arizona’ would normally be affirmed under
contextually operative parameters that do not require, in order for the statement
to be true, that the right ontology contain items like universities, nations, or states.
I would situate my discussion in sections 14.1–14.7 within the wider frame-
work of contextual semantics in the following way. That discussion pertains pri-
marily to default settings of semantic parameters, for both presentational content
and judgmental content. Those settings may well work in such a way that the de-
fault satisfaction conditions, for any or all of the kinds of content I have discussed
(viz., presentational secondary-quality content, judgmental secondary-quality
content, presentational primary-quality content, and judgmental primary-quality
content), involve only indirect correspondence rather than direct correspondence.
(More on this theme in section 14.9.)
Contextual semantics also can accommodate certain contextually appropriate
non-standard settings, for judgmental content in particular. In the context of phil-
osophical discussion about the metaphysics of color, for instance, it can become
contextually appropriate to deploy the judgmental notion of color in a non-default
way that coincides with the presentational notion. In such a context, one can truly
say this: Nothing in the world has color. Some things philosophers say—like that
one—are strange but true: strange because they contravene deeply held beliefs that
are normally regarded as overwhelmingly well warranted, but true because they
are being affirmed under unusual, but contextually appropriate, settings of implicit
semantic parameters.
Contextual semantics also fits well with my remarks, in section 14.6, about
being able to go different ways as to the truth values of the color-attributing sen-
tences of color-inverts and the direction-attributing sentences of left-right inverts,
and being able to go different ways as to sameness or difference in meaning. The
different ways of going involve different settings of implicit contextual parameters
on ‘truth’-talk and on ‘meaning’-talk.

14.9. Austere Realism and the Contents of Perceptual Experience

Matjaz Potrč and I advocate a metaphysical position we call austere realism

(Horgan and Potrč 2006, 2008). Metaphysically, austere realism asserts that the
right ontology, whatever exactly it is, excludes numerous posits of everyday com-
mon sense and even science. Semantically, austere realism asserts that almost
all human thought and discourse, including much serious scientific thought and
discourse, is governed by indirect-correspondence default semantic standards
(and is often true, under those default standards). We mount various arguments
in support of austere realism, but perhaps the most fundamental one is an ar-
gument that ontological vagueness is impossible. This excludes from the right
ontology most of the posited objects and posited properties of common sense
 The Constituents of Perceptual Content and the Role of Perception
348   

and science. For, such putative objects and properties, if they were included in
the right ontology, would be ontologically vague; and ontological vagueness, we
argue, is impossible.
In sections 14.6 and 14.7, I maintained that although perceptual experience
is nonveridical insofar as its secondary-quality presentational content is con-
cerned, it is typically veridical in its primary-quality presentational content. But
I also would claim that much of the presentational primary-quality content of
perceptual experience is vague. (Within the GTH framework, the rubric of pres-
entational primary qualities, as characterized in section 14.7, includes not just
presentational properties like shape, size, relative position, and the like, but nu-
merous others too, some of which are surely vague—e.g., properties like being
a cup and being a table). Can the veridicality of primary-quality presentational
content be reconciled with the vagueness of presentational primary qualities? My
answer is yes. The key to reconciliation is to embrace the claim that the default
primary-quality presentational content of experience is not only often vague in
various respects, but is also governed by indirect-correspondence semantic stan-
dards. When one’s visual experience presents a cup on a table, the presentational
content of this experience can indeed be veridical—even though putative prop-
erties like cuphood and tablehood are vague and therefore cannot belong to the
right ontology. This vague presentational content bears the contextually pertinent
relation of truth—in this case, a relation of indirect correspondence—to the non-
vague world.11
One final point deserves mention. Thought and discourse often is governed
by indirect-correspondence semantic standards even in metaphysics-oriented
philosophical inquiry. For instance, in urging the difference between
primary-quality presentational content and secondary-quality presentational
content, I myself am happy to say that presentational primary qualities like cup-
hood and cathood are real and also are instantiated in the world, and that sec-
ondary presentational qualities like redness are real too (albeit not instantiated
in the world). In positing such properties, and in calling them real despite their
vagueness, I am deploying property-talk in a manner that itself is governed by
indirect-correspondence standards. By my lights this is a theoretically legitimate
way to talk, in context, even though I also hold that the right ontology cannot
include vague properties. Given austere realism, this kind of double-talk is to
be expected, even in philosophical inquiry about the metaphysics of color. Such

11
One might think that under contextual semantics, the presentational color-content of
visual-perceptual experience turns out to be veridical itself—albeit with its veridicality being a form
of indirect correspondence that does not require presentational color-properties to be instantiated
in the world. But I would maintain that ordinary visual experience does present certain determinate
color-properties as being instantiated in spatiotemporally determinate ways, and thereby has
direct-correspondence truth-conditions insofar as determinate presentational color-properties are
concerned. (Although determinable presentational colors are vague, vagueness does not intrude as
regards the determinate color-content of visual experience.)
 Phenomenal Intentionality and Secondary Qualities   349

double talk is useful and theoretically illuminating in philosophy, as long as one

keeps track of the contextually operative, dynamically shifting, score in the lan-
guage game.12

References

Barnard, R., & Horgan, T. (2006). Truth as mediated correspondence. The Monist, 89, 28–49.
Block, N. (1995). On a confusion about a function of consciousness. Behavioral and Brain
Sciences, 18, 227–247.
Chalmers, D. (2006). Perception and the fall from Eden. In T. Gendler & J. Hawthorne
(Eds.), Perceptual Experience (pp. 49–125). Oxford: Oxford University Press.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Farkas, K. (2008). Phenomenal intentionality without compromise. The Monist, 91, 273–293.
Georgalis, N. (2006). The Primacy of the Subjective. Cambridge, MA: MIT Press.
Graham, G., Horgan, T., & Tienson, J. (2007). Consciousness and intentionality. In M.
Velmans & S. Schneider (Eds.), Blackwell Companion to Consciousness (pp. 468–484).
Oxford: Basil Blackwell.
Graham, G., Horgan, T., & Tienson, J. (2009). Phenomenology, intentionality, and the unity
of mind. In B. McLaughlin, A. Beckermann, & S. Walter (Eds.), The Oxford Handbook
of Philosophy of Mind (pp. 512–537). Oxford: Oxford University Press.
Horgan, T. (2001). Contextual semantics and metaphysical realism: Truth as indirect cor-
respondence. In M. Lynch (Ed.), The Nature of Truth: Classic and Contemporary
Perspectives (pp. 67–95). Cambridge, MA: MIT Press.
Horgan, T. (2007). Mental causation and the agent-exclusion problem. Erkenntnis, 67,
183–200.
Horgan, T. (2011). Phenomenal intentionality and the evidential role of perceptual experi-
ence: Comments on Jack Lyons, Perception and Basic Beliefs. Philosophical Studies, 153,
447–455.
Horgan, T., & Graham, G. (2009). Phenomenal intentionality and content determinacy. In
R. Schantz (Ed.), Prospects for Meaning (pp. 321–344). Amsterdam: de Gruyter.
Horgan, T., & Potrč, M. (2006). Abundant truth in an austere world. In M. Lynch & P.
Greenough (Eds.), Truth and Realism: New Essays (pp. 137–167). Oxford: Oxford
University Press.
Horgan, T., & Potrč, M. (2008). Austere Realism: Contextual Semantics Meets Minimal
Ontology. Cambridge, MA: MIT Press.
Horgan, T., & Tienson, J. (2002). The intentionality of phenomenology and the phenom-
enology of intentionality. In D. J. Chalmers (Ed.), Philosophy of Mind: Classical and
Contemporary Readings (pp. 520–533). Oxford and New York: Oxford University Press.
Horgan, T., & Timmons, M. (2002). Conceptual relativity and metaphysical realism.
Philosophical Issues, 12, 74–96.

12
Thanks to Mark Timmons for helpful comments and discussion.
 The Constituents of Perceptual Content and the Role of Perception
350   

Horgan, T., Tienson, J., & Graham, G. (2004). Phenomenal intentionality and the brain
in a vat. In R. Schantz (ed.), The Externalist Challenge: New Studies on Cognition and
Intentionality (pp. 297–317). Amsterdam: Walter de Gruyter.
Kriegel, U. (2011). The Sources of Intentionality. Oxford: Oxford University Press.
Kriegel, U. (2013). The phenomenal intentionality research program. In U. Kriegel (Ed.),
Phenomenal Intentionality (pp. 1–26). Oxford and New York: Oxford University Press.
Loar, B. (2002). Phenomenal intentionality as the basis of mental content. In D. J. Chalmers
(Ed.), Philosophy of Mind: Classical and Contemporary Readings (pp. 000–000).
Oxford and New York: Oxford University Press.
Lycan, W. (1995). Consciousness and Experience. Cambridge, MA: MIT Press.
Lyons, J. (2009). Perception and Basic Beliefs: Zombies, Modules, and the Problem of the
External World. Oxford and New York: Oxford University Press.
Maund, B. [1997] (2006). Color. Stanford Encyclopedia of Philosophy, http://plato.stanford.
edu/entries/color/.
McGinn, C. (1988). Consciousness and content. Proceedings of the British Academy, 76,
219–239. Reprinted in N. J. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of
Consciousness: Philosophical Debates. Cambridge, MA: MIT Press, 1997.
Pitt, D. (2004). The phenomenology of cognition; or what is it like to think that P? Philosophy
and Phenomenological Research, 69, 1–36.
Siewert, C. P. (1998). The Significance of Consciousness. Princeton: Princeton University
Press.
Strawson, G. (1994). Mental Reality. Cambridge, MA: MIT Press.
Strawson, G. (2008). Real intentionality 3: Why intentionality entails consciousness. In id.,
Real Materialism and Other Essays (pp. 281–305). Oxford: Oxford University Press.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.
15

Which Causes of an Experience Are Also Objects

of the Experience?
Tomasz Budek and Katalin Farkas

15.1. The Content of Perception

By ‘perceptual experiences’ we mean here experiences associated with the process

of apparently gaining information through the five external senses: vision, hearing,
touch, smell, taste. Perceptual experiences form a variety of sensory experiences
in general; other varieties may include experiences that are not clearly associated
with the process of external perception. In visual or auditory imagining, experi-
ences are not produced by external senses. Bodily sensations, possibly some types
of emotional experiences, or experiences of the passage of time may involve sen-
sory modalities other than those associated with the five external senses.
Perception provides our primary access to the contingent features of the world
around us, and at the first sight, this is what distinguishes perceptual experience
from other sensory experiences. Through perception, we gain a conception of the
world, and we acquire knowledge of its nature. The term ‘content of experience’
may be used in different senses (see Siegel 2013), but here we mean by the ‘content’
of an experience simply the way the experience presents the world. Philosophical
interest in this feature is motivated by a number of considerations.
According to a strong current in the empiricist tradition, perceptual obser-
vation is the neutral arbiter among different theories of the world. To use Quine’s
metaphor, theories face “the tribunal of experience”. The metaphor suggests that
we expect the tribunal to be impartial among the different theories. But is this
expectation well founded? It has been suggested that the sharp distinction between
raw and uninterpreted perceptual observation on the one hand, and subsequent

Earlier versions of this paper were presented at a workshop at Central European University, and at
talks at the University of Bristol and the University of Cardiff. The authors are very grateful for audiences
for their comments. Research for this paper has received funding from the European Commission’s
Seventh Framework Programme FP7/2007-2013 under grant agreement no. FP7-238128, and from the
project BETEGH09 supported by MAG Zrt. 351
 The Constituents of Perceptual Content and the Role of Perception
352   

theorizing on the other, is untenable. According to this suggestion, the way we see
the world, the way it appears to us, is already a result of joint work between theory
and observation. This is one of the important questions about the content of expe-
rience: just what is it exactly that perceptual experience presents us?
This question has always interested philosophers: below we shall quote for
example Aristotle’s view on the objects of perceptual experience. However, it is
worth mentioning some factors in and outside philosophy that have shaped the
recent development of the issue. Starting already in the nineteenth century, the
empirical study of perception in cognitive psychology and later, in the twentieth
century, in neuroscience, has become a huge enterprise, and we have now exten-
sive experimental data and a whole array of theories about the psychological and
physiological aspects of perception. We now know that the seemingly simple act of
seeing an object is made possible by a complex operation of the perceptual system,
using both hardwired and learned mechanisms. This gave a new impetus to asking
the question of whether perception can be separated from the rest of our cognitive
operations. This question, in turn, may have far-reaching consequences for our
epistemological theories.
Among the senses, vision has always been in the primary focus of investiga-
tion both in philosophy and psychology. More recently, however, there has been a
growing interest in the other senses, and it seems that focusing on vision may even
have distorted our view of perception. As we mention below, some philosophers
have even considered the view that for example the olfactory sense is not in the
business of presenting the world at all: olfactory experiences have no content, they
are mere sensations, mere modifications of one’s consciousness. We defend a dif-
ferent view in this paper.

15.2. Causes and Objects of Experiences

We include under ‘perceptual experiences’ both veridical and non-veridical (illu-

sory or hallucinatory) experiences. All successful perceptual experiences have an
object. This is simply another way of saying that in a perceptual experience, some-
thing seems to be perceived—what seems to be perceived is the object. Among suc-
cessful perceptual experiences we include illusions or misperceptions too, that is,
perceptual experiences where some object is actually perceived, but it’s perceived
to be different from the way it is (so not all successful experiences are entirely
veridical). But we don’t include hallucinations: what distinguishes hallucinations
is that the object one takes oneself to perceive is not actually perceived; it may not
even exist.1

1
The boundary between successful misperceptions and hallucinations is probably blurred; it’s not
clear how serious a misperception can be so that we can still say that some object is actually perceived.
But even if the distinction is vague, it is still a distinction, and an important one.
 Which Causes of an Experience Are Also Objects of the Experience?   353

Our ordinary ways of speaking allow a wide variety of categories to be objects

of perception: a property, a thing, an event, or in that-clause constructions, a fact.
We see the color of a dress, we taste a strawberry, we hear the train approaching,
we feel with our fingers that the surface of the table is smooth. In case of a suc-
cessful perceptual experience (when someone actually sees, hears, smells, tastes
etc. something), the object of perception contributes to the cause of the perceptual
experience. There are various views about the relata of the causal relation: they
may be substances, events, facts, perhaps properties. We want to remain neutral
on this issue, and would like to formulate our view so that it is compatible with
either of these theories. For example, if someone believes that the relata of causa-
tion are events, then we say that when someone sees a glass, the glass is a constit-
uent of some event that is the cause of the perceptual experience. From now on, we
will say that the object is a cause of the experience without adding the qualification
‘or constituent of an event or the fact which is a cause’, but this will be understood
implicitly.
An object of a perception is only one element in the causal chain leading to the
experience. Someone wanted a drink, went to the kitchen to fetch a glass of water,
the glass is now in front of her, light is reflected from its surface, some stimulus
reaches the eye, changes are brought about in the retina, impulses are forwarded
in the optic nerve, and so on. On various views of causation, all elements in this
causal chain can be called the causes or ‘some’ causes of the perceptual experience.
That is why it is better to say that the object of perception is ‘a’ cause, rather than
‘the’ cause of the experience.
It may seem that certain naïve realist theories of perception deny that the
object of a perceptual experience is its cause, because they would say that the
object is a constituent of the experience, and we don’t normally regard constituents
as causes. This is correct, but there is a way of modifying the claim which gives
us everything we want, while taking naïve realist views on board. Naïve realists
could agree that the object of perception is part of a causal chain that stands in a
close relation to perceptual experience; however, on their theory, the final link in
the causal chain would be not the perceptual experience itself, but rather a brain
state that partly constitutes the experience. Thus, naïve realists would agree that if
no element in the causal chain leading to the brain-state is the object of the expe-
rience, then the experience is a hallucination (this is a sufficient, but perhaps not
a necessary condition for hallucinating). The object of perception is an element in
the causal chain leading to the experience, or to a brain-state that is a constituent
of the experience. In what follows, we will mean this disjunct when we refer to the
causal role of the object.
It is necessary for a successful perceptual experience that its object is found
among the causes of the experience. However, it should be obvious that not all
causes of the experience are also its objects; returning to the above example of see-
ing a glass, what one certainly sees in the situation is the glass, but this cannot be
said of the links in the causal chain before and after: say, the hand placing the glass
 The Constituents of Perceptual Content and the Role of Perception
354   

on the table, or the surface of one’s retina. Even if the subject of experience knows
that these things participate in the causal chain, and hence she can infer to their
presence or existence from the fact that she has the perceptual experience, it would
still not sound right to say that these are the objects of her experience.

15.3. A Question

In the previous section, we established that an object of an experience must be a

cause of the experience, but not all causes are objects. Can we say something in
general about which elements in the causal chain—leading to a successful percep-
tual experience—are the objects of the experience? What is it about the glass, and
about the surface of the retina, that makes it so compelling that the first is the
object of perception, and the second isn’t? We do not assume that there must be a
general answer to this question—maybe there are different cases, or maybe this is
just a primitive fact; but the question is worth asking.
This issue hasn’t received much attention in this general form (and won’t be
resolved entirely in this essay). Some particular cases and some issues in the close
vicinity have been discussed in some detail, for example in the debate about the
immediate objects of perception; on this, we will say more in the next section.
There is a debate for example on the objects of olfaction: do we perceive odors,
or sources of odors? In the tactile modality, do we perceive the skin or the object
touching the skin?
In this paper, we would like to argue that the question concerning the objects
of a given perceptual modality cannot always be given a uniform answer. For
example, in the olfactory modality, one can’t say that we always smell an odor, or
that we always smell the source of odor. We claim that on some occasions we smell
one, on some occasions the other or both. Perception can take different kinds of
things as objects, in a sense we shall explain below: different elements in the causal
chain leading to the experience can become the objects of perception. We shall in-
vestigate this claim in more detail in the case of olfactory and tactile experiences,
but we believe that similar considerations may apply to all sensory modalities.
If we can answer the question of what the object of experience is, we gain
knowledge also of the content of the experience. The content is plausibly regarded
as representing the objects as being in a certain way. If the objects of perception
vary, then contents also vary.

15.4. Related Questions

Since the possible objects of perceptual verbs show a great variety in a number of
dimensions, philosophers sometimes raise the question of whether some objects
are special in a certain sense. First, the investigation has to be narrowed—if
 Which Causes of an Experience Are Also Objects of the Experience?   355

possible—to what we literally regard as cases of perceiving something. We are not

interested in all cases when it is simply natural to say that we perceive something.
For example, it is natural to say that we can see that it’s cold outside when we see
people hurrying on the street huddled in their coats. However, many would disa-
gree with the claim that we can literally see temperatures, and would say instead
that we learn about the temperature in this case by seeing the shapes and colors of
people moving. One could ask for a general theory of all cases where perceptual
verbs take objects in natural everyday constructions. This is an interesting ques-
tion, but not the one in the focus of this paper. We are looking for those objects
which are literally perceived (assuming, of course, that this distinction can be
made.)
An issue related to our discussion is the question of which properties can be
represented2 in experiences in a sensory modality. This issue can be regarded as
the complement of our question concerning which things are the objects of an
experience. Assuming that the content is the representation of things as having
certain properties, one issue is what kind of properties, another issue is what kind
of things can enter the content of a perceptual experience.
Recently there has been a lively debate of the question of which properties
are represented in experience. The debate is largely concerned with visual expe-
riences, where on one side of the debate we find claims that visual experiences
represent only a limited range of properties: colors, shapes, motion, perhaps a few
more, while on the other there are views according to which visual experiences
can represent properties like that of being a tree or of being a house.3 Susanna
Siegel, who defends the second, more liberal view, discusses for instance the case
of someone who gradually develops a recognitional capacity for pine trees (Siegel
2006, e.g., pp. 491–500). Siegel thinks that while previous to this development,
the subject may have represented only the shape and color of pine trees in her
experiences, once something’s being a pine tree becomes visually salient to her,
this feature becomes part of the content of her visual experiences. There are some
notable differences between the case of representing properties and representing
things, but these differences aside, we believe that in some circumstances, a similar
phenomenon occurs in connection to the things that are the objects of our experi-
ences. And as we shall see below, there are important parallels between the issues
of which properties and which things enter the content of a perceptual experience.

2
Which properties can be represented—or presented, if someone doesn’t want to commit herself
to the idea that perceptual experiences are representational. Hopefully, taking sides on this issue is
orthogonal to the question we are asking. Hence we shall often talk of representation and content, but
what we say should be relevant for those who believe that perceptual experience is not representational
(e.g., Travis 2004).
3
An example of the former, called sometimes ‘sparse’ or ‘conservative’ view, is Tye (1995, pp. 140–
141). Some of the recent defenses of the ‘rich’ or ‘liberal’ view include Siegel (2006) and Bayne (2009).
 The Constituents of Perceptual Content and the Role of Perception
356   

A further question about the objects of perception concerns the proper objects
of a sense, that is, the objects that can be perceived by that sense only4; for example,
smells can arguably be perceived only by the olfactory sense. Yet another ques-
tion is about the primary objects of perception: these are objects that are always
perceived whenever something is perceived in a sensory modality. For example,
people have argued that whenever we hear something, we always hear a sound—
indeed, if we hear anything, we hear it by or in virtue of hearing a sound. As David
Sanford (1976) makes it clear, though sounds are also proper objects of hearing,
being proper and being primary are different conditions; something could be spe-
cific to a sense yet not involved in every experience through that sense, and some-
thing could be involved in every such experience without being specific to that
sense.
Proper and primary ‘objects’ are sometimes conceived as qualities: for
example, when it is claimed that colors are proper objects of vision. In this case,
‘object’ is not an ontological category, but simply signals that something is per-
ceived in a certain modality. But proper and primary objects need not always be
qualities. A ‘sound’ can be understood as a spatio-temporally located physical
existent.5 Similarly, smells or tastes (or perhaps ‘odors’ and ‘flavors’) can be un-
derstood not as qualities, but as quantities of certain chemicals. If we wanted to
find an equivalent of sounds, odors and flavors in the visual modality, it would be
something like a colored expanse, rather than simply a color.
Developing a theory of the proper and primary objects of perception—if it
can be done—goes some way towards answering our question, but doesn’t go the
whole way. For one thing, saying that a colored expanse is the proper and primary
object of sight doesn’t answer the question of why it is the glass we see, but not the
retina or the glassmaker, given that all three have colored surfaces. Further, we are
interested in all objects of perception, including improper and secondary ones,
and anyone who allows for such objects will still have to single out some elements
in the causal chain. Finally, our question stands even if there are no proper and
primary objects of perception. So the debate about the proper, primary or direct
objects of perception, though relevant to our purposes, is not exactly what we are
interested in.

15.5. Objects of Olfactory Experiences

Our question was: which elements in the causal chain leading to (the final constit-
uent of) perceptual experiences are the objects of experience? That is, where, along

4
See Sanford 1976, p. 190. The question goes back to Aristotle’s theory of ‘proper sensibles’. It
is sometimes thought to have significance for the issue of distinguishing different sensory modalities
from each other; see Nudds (2004).
5
This question is discussed for example in O’Callaghan (2007).
 Which Causes of an Experience Are Also Objects of the Experience?   357

the causal chain, do we locate the experience-independent thing that we perceive?

We do not have a general explanatory answer to this question, but we have a claim
that puts a constraint on possible explanation: that within at least some sensory
modalities, different kinds of things can be the objects of a perceptual experience.6
Let us illustrate this first on the example of olfactory experiences.
On some views, olfactory experiences have no object. Christopher Peacocke,
for instance, allows for the possibility that “a sensation of smell ( . . . ) may have no
representational content of any sort, though of course the sensation will be of a
distinctive kind” (1983, p. 5). As ‘content’ is used in this paper, if the sensation has
no content, it has a fortiori no object either.
William Lycan in “Layered Perceptual Representation” considers a
different view:
( . . . ) there are objects other than roses that set off the rose smell—artificial
rose smells can be made of any substance whose molecules are shaped sim-
ilarly to those of roses. The point is not that the nose can be fooled. Au con-
traire; it is that in the artificial case, the nose is not fooled, and the rose smell
is not incorrectly tokened. An artificial rose that produces the rose smell
is smelled correctly, for it does have that smell even though it is not a rose.
(Lycan 1996, p. 90)
Someone who accepted this line would agree that we often classify particular
smells with respect to the objects likely to give out those smells—in this case with
reference to roses. However, she would think that we correctly say that we smell
‘rose smell’ whenever rose odor is present; hence our experiences are correct or
incorrect not with respect to roses, but rather with respect to rose odor. Therefore,
odors are better candidates for being the objects of olfactory experiences.
At the end, however, Lycan turns against the view that the objects of ol-
factory experiences are only odors. Lycan is persuaded here by Ruth Millikan’s
teleosemantics-based arguments. On the teleosemantics story, perception indi-
cates for the organism the presence of those things which are useful for its sur-
vival, and in the case of olfaction, these things are presumably not odors, but the
source of odors. Hence:
( . . . ) if smells do represent anything, they do after all represent environmental
objects of potential adaptive significance. Surely that is what olfaction is for, to
signal food, predators, shelter, mates, and other objects of interest ultimately
derivative from those . . . (Lycan 1996, p. 92)

6
The phenomenon is general: given that some bodily sensations arguably have an object, the same
observation applies to bodily sensations too. In this paper, we focus mainly on perceptual experiences,
but we discuss for example tactile sensations (i.e., awareness of bodily parts in the tactile mode), since
they have a very intimate connection with external tactile perception.
 The Constituents of Perceptual Content and the Role of Perception
358   

However, Lycan is reluctant to give up the view that olfactory experiences repre-
sent odors, so he settles for a ‘multi-object’ view: olfactory experiences represent
both roses and odors, indeed they represent the former ‘by’ representing the latter.
The difference between smelling the odor of coffee and smelling a thing like
coffee in the cup illustrates the sense in which olfactory experiences can have dif-
ferent type of objects: different elements are singled out in the causal chain leading
to the experience. This means that there are different conditions for the experience
being successful. If the object of smell is coffee, and there is no coffee in the room,
then any apparent experience of coffee is hallucinatory. But if the object is an odor
which happens to be emitted by coffee, then an experience can be veridical even if
there is no coffee around, but only the smell lingers.7
At first sight, this relation between perceptual error and the content of
experience may suggest a way of finding out what the proper objects of percep-
tion are: those whose presence and properties are responsible specifically for per-
ceptual, rather than other kinds of error. Unfortunately, this observation, while
valid, doesn’t advance the debate at this stage, because we don’t really have direct
intuitions about perceptual error independently of what we take to be the objects
of perception. All parties agree that in the case of mistakenly thinking that one is
smelling a real rose rather than an artificial one, there is an error. But the question
of whether the mistake is perceptual or cognitive, is the same as the question of
what the object of perception is.

15.6. The Phenomenal Presence of Objects

We are interested in the question of which causes of a perceptual experience are

also objects of the experience. In this paper, we argue that at least within some sen-
sory modalities, there is no uniform answer to this question. We have illustrated
this claim so far on olfactory experiences: the object can be an odor, or the source
of an odor. Or in any case, these are candidates for being the object of the appro-
priate experiences. But perhaps there is a way of deciding the issue?
There is a suggestion we may want to pick up from the previous section.
Perhaps teleosemantic theories can give a principled answer to the question of
where to locate the objects of perception: namely, where objects of ‘potential
adaptive significance’ are found. For example, in the case of auditory perceptions,
the presence of the thing emitting a sound is presumably more significant for sur-
vival than the presence of the sound; this consideration then would rule in favor

7
In fact, the view that smells have no objects at all can be regarded as a ‘limiting case’ of moving
up in the causal chain, closer and closer to the experience (or relevant brain state). Perhaps one
could extend the argument of this paper to show that moving from a pure qualitative condition to an
experience with some content is a process in some ways similar to the process of the object moving
along the causal chain.
 Which Causes of an Experience Are Also Objects of the Experience?   359

of regarding things, rather than sounds, as the object of auditory experiences. If

other reductive theories of content—theories that attempt to reduce intentional
content to causal-nomological connections—were successful, they would also be
contenders for locating the objects of perception.
Reductive accounts of content face many objections, and we think there are
good reasons to think that none of the theories we are familiar with is completely
satisfactory. Be that as it may, our approach to this question is completely dif-
ferent: we propose to regard the phenomenal presence of the object as our guide in
establishing what is perceived in an experience. This approach is compatible with
the eventual success of reductive theories.
To introduce the idea, let us reflect for a moment on how reductive accounts
of intentional content are developed and evaluated. Take for example Fodor’s dis-
cussion in Psychosemantics (Fodor 1987, ch. 4). Fodor’s question is somewhat dif-
ferent from ours: he wants to know how mental symbols represent properties, and
he is interested not only in perceptual experiences, but in mental states in general.
Nonetheless, as we shall see, the issue is straightforwardly relevant to our problem
as well. Fodor first considers the suggestion that whatever causes the tokening
of a mental symbol is represented by that symbol. This would have the result
that tokenings of a certain symbol, say HORSE, represents horses and cows-in-
the-dark. This is clearly unacceptable, says Fodor, and we all agree. Another, sim-
ilar objection may be that if experiences (conceived either as tokening of mental
symbols or in another way) represented their causes, then visual experiences could
represent states of the retina. This is clearly unacceptable too.
The question is: why are these consequences unacceptable? It seems that even
before we start to design a systematic theory, we already have some idea of what is,
and what can possibly be—or cannot possibly be—the object of a perceptual expe-
rience or the content of a representation. We know that the object of an olfactory
experience of a rose smell is not the surface of the inside of our nose, nor it is the
gardener who planted the rose. If a theory had that consequence, it would prob-
ably count as a reductio against that theory.
Where do these initial ideas about the objects of experiences come from? We
propose to further elaborate the matter by an appeal to the phenomenal character
of experiences. That is, we assume, along with many other philosophers, that the
nature of a perceptual—and in general, sensory—experience is given by its phe-
nomenal character. The question about the objects of perception arises precisely
because it is part of the phenomenology of perceptual experiences that they seem
to present to us objects as being in a certain way. All further investigations are
shaped by this initial observation.
The object of a perceptual experience is what seems to be present when having
that experience. In other words, it is the object that is phenomenally present. This
is clearly not an attempt at a reduction: indeed, the claim is almost tautological.
However, it is still important, because we implicitly rely on it when evaluating
theories of perception or intentionality. Whatever account of perception we give,
 The Constituents of Perceptual Content and the Role of Perception
360   

one of the data we have to account for is that we see the glass in front of us, but we
don’t see the optic nerve, nor the hand that placed the glass here.
There are various notions, like ‘signaling’ or ‘carrying information’ that may
adequately cover the relation between a perceptual experience and any member of
the causal chain that leads to the experience. Perhaps experiences carry informa-
tion about states of the retina, states of the optic nerve; perhaps a visual experi-
ence of the lights being on next door signals that the neighbor is at home. But
the reason why we say that neither the retina, nor the neighbor is the object of
the perception—that we don’t see the retina or the neighbor—is that they are not
phenomenally present in the experience. We don’t seem to see these things when
having the experience—we see (or at least seem to see) simply the lights.
The idea that the objects of perception are phenomenally present in an expe-
rience is meant to be compatible with pretty much any theory of perceptual expe-
rience. Suppose we accept a representationalist theory of perception, which holds
that the nature of a perceptual experience is determined by its representational
properties, which, in turn, determine the accuracy conditions for the experience.
A particular visual experience is of a glass, because the experience is veridical if
the glass—rather than the retina, or the glass-maker—has certain properties. Of
course, we might agree with that. But how did the representationalist know where
to look for the accuracy conditions? Why didn’t she suggest that the experience is
veridical if the retina is in a certain state? The simplest answer, we suggest, is that
the retina doesn’t seem to be the object of the experience—in other words, it’s not
phenomenally present in the experience.8
Appealing to the idea of phenomenal presence is not going to settle all issues
about the objects of perception. First, there may not always be a consensus on what
is phenomenally present in someone’s experience. Each of us has to rely on their
first-person reports, and some people may be better than others in attending to the
phenomenal features of their experiences. In some cases, there may be some un-
certainty at first whether something is phenomenally presented in an experience
or its presence is evoked as a result of some association, and then different kinds
of considerations might be used to decide the question. Besides, as we noted, the
guidance of phenomenal presence is compatible with many, otherwise different
theories, so clearly, not all questions will be solved this way.
But even if there will be unclear cases, it seems obvious that there are clear cases
as well. As we said before, most philosophers will agree that one’s own olfactory bulb
is normally not the object of an olfactory experience. To argue that the olfactory bulb
is the object of an olfactory experience requires significant further work. So it seems
that we do use the condition of phenomenal presence as the default position to
exclude certain objects as candidates for being the object of a perceptual experience.

8
A similar notion of phenomenal presence is used for a similar purpose—i.e., to locate objects
of representation among the many causes of representation—by Galen Strawson in Strawson (2008).
 Which Causes of an Experience Are Also Objects of the Experience?   361

15.7. Phenomenal Presence and Different Kinds of Objects

It’s part of the phenomenology of perceptual experiences that objects seem to be

presented to us; so the first and foremost clue when locating the objects of percep-
tual experiences is to see what is phenomenally present in our experiences. And
our reason for insisting that perceptions can take different kinds of objects in a
given modality is that we think that these different kinds of things can all have the
appropriate phenomenal presence in our perceptual experiences. Without a further
argument against these things being the object of an experience, we suggest we
should take phenomenology at face value.
To clarify: we do think that for any particular episode, one can single out the
perceived object(s) among the causes. We don’t think that the very same token
olfactory experience can take an odor or a flower as its object, depending on some
further factors. When we argue for the possibility of different kinds of objects, we
merely say that we cannot in general establish that all olfactory experiences take
only one or other kind of thing as their object.
Compare two situations: you enter a room and an unexpected smell hits
you—or you enter a room and you smell coffee; we say that from a phenomeno-
logical point of view, the odor and the coffee are present in your experience in the
same apparently unmediated way. So in the first case, the object is only the odor,
in the second case, the objects include the source of odor (and possibly the odor
as well). Another case: you feel a soft caress your arm—or your are searching in
your bag for your key and suddenly your fingers touch it; in the first case, you are
aware only of some sensation on your skin, in the second case, (also) of the object
that touches the skin. Here again, we say that your skin (as you feel the sensation
in your arm) and the key have the same sort of phenomenal presence, and there-
fore both count as objects of perception. (More on the objects of bodily and tactile
experiences in sections 15.9 and 15.10; the foregoing was just one example of how
something may be added to an experience as an object.)
As these examples show, one factor that often plays a role in determining the
object of a particular perceptual episode is the totality of our previous experiences.
Coffee becomes phenomenally salient in our experience after repeated encounters
with coffee; familiarity with the shape of the key and our purpose in the search
makes the key phenomenally present. In the second case, the focus of attention
is an additional factor in determining the phenomenal presence: if we are in
an exploratory mode, shapes might strike you differently than in the case of an
unexpected stimulus.
There are two parts to the defense of the claim that the object of, say, an
olfactory experience can be located at different points in the causal chain: first, we
have to show that in some cases, the only object is the odor; second, we have to
show that in some other cases, the source of the odor is really an object. We need
both parts, because otherwise someone could hold a view, similar to Lycan’s view,
that olfactory experiences always have multiple objects: we perceive the source by
 The Constituents of Perceptual Content and the Role of Perception
362   

perceiving an odor. If that was the case, then we could establish uniformly, for all
cases of the olfactory modality, which elements in the causal chain are objects of
the experience. But our contention is that we cannot establish this.
The most plausible cases for smelling only odors are those of dispersed and
unfamiliar smells. You are walking on a deserted country road and suddenly a
peculiar, faint and slightly stale smell hits your nose. You have no conception of
the origin of the smell; the faintness and staleness is characteristic of a lingering
smell rather than a freshly produced smell. We think that it would be implausible
to say that in this case, any object beyond the odor itself is phenomenally present
in your experience.
A different kind of case is when you feel a feel a familiar smell, clearly asso-
ciated with a source, perhaps even answering to some need or expectation. You
arrive at a friend’s house for brunch after a bad night sleep; you smell freshly
brewed coffee, and you immediately look for the coffee pot. Maybe this is one of
the cases that fits Lycan’s idea of perceiving objects with potential adaptive signif-
icance: what matters for your survival that morning is the coffee, not its odor, and
this may explain its phenomenal salience in your experience. Since the coffee is
presented in your experience, it is an object of the experience.

15.8. Unconscious Inference or Association?

We have been arguing that on some occasions, the object of an olfactory experi-
ence is an odor, on other occasions, it is (also) the source of the odor. This sup-
ports our claim that within a sensory modality, one cannot fix the object of the
experience for all cases. Our main argument for claiming that the object may vary
from case to case is based on an appeal to the phenomenal presence of the object.
Sometimes it’s the odor, sometimes it’s (also) the source of the odor that is phe-
nomenally present in an experience.
A possible objection is that we don’t in fact smell the coffee in the second case,
but we merely make a quick unconscious inference to the presence of coffee, or that
an immediate association of the smell and the coffee brings the idea of coffee to
mind (in the latter case, we need not say that any inference took place). Coffee is
not the object of the perceptual experience, according to this objection, but of a
separate mental state, probably a judgment. The coffee case then would be similar
to the case of ‘seeing’ that it’s cold outside: we don’t literally see the cold, we only
learn that it is cold on the basis of a visual experience.
According to the view which is behind this objection, the inference or asso-
ciation is unconscious. This is important, because it doesn’t seem that in the
specific cases we have in mind, there is an intermediate step. The idea of coffee
comes to you directly upon entering the house, and immediately prompts a
reaction. (Similarly, to anticipate the discussion of tactile experiences, when your
fingers close around the key in your bag, the process doesn’t seem to consist of two
 Which Causes of an Experience Are Also Objects of the Experience?   363

steps: first, the identification of the sensation on your skin, and then matching it
with an awareness of the external object. Instead, you just feel the key.)
We do not want to say, at this point, that when you smell the coffee, you cease
to be aware of the odor, and when you feel the key, you cease to be aware of a con-
dition of your skin. In fact, we’d like to leave this question open. What we want to
say is merely that it doesn’t seem that awareness of the coffee or the key is reached
through an intermediate step. It seems possible to be aware of two different objects
without the phenomenology of indirectness. An analogous case, involving proper-
ties rather than objects, might be the example we mentioned above about seeing
pine trees. Susanna Siegel’s claim that experiences (can come to) represent natural
kind properties does not entail that when they do, they cease to represent simpler
properties like colors or shapes. If Siegel is right, and when developing a recogni-
tional capacity for pine trees, you start to be perceptually aware of something being
a pine tree, this doesn’t mean that you cease to be perceptually aware of the shapes
and colors that determine that something looks like a pine tree. Furthermore, and
that’s the crucial point, the property of being a pine tree that according to Siegel
enters the content of the perceptual experience, does not seem to be represented
in the content of a mental episode as somehow being inferred from, or following
upon, the original perceptual experience of shapes and colors.
Now back to the suggestion that we don’t perceive, but merely unconsciously
infer, or make an association with, the presence of coffee. Our reply is that this
objection is based on a false contrast between perception and the involvement of
unconscious inference or association. If an unconscious inference or association is
involved in a perceptual process, this doesn’t mean that the inference starts from a
perceptual experience and leads to a separate mental state. It’s very much possible
that the unconscious process precedes the formation of the perceptual experience
itself.
There is overwhelming evidence that our perceptual experiences are results of a
substantial amount of unconscious work performed by our perceptual system. This
is especially well-documented in the case of visual perception.9 Take for example the
hollow-face illusion (Gregory 1973, pp. 49–96). The stimuli that reach the sensory
organ are compatible with both a concave and a convex surface under different illu-
mination conditions, and yet the resulting experience is always of a convex surface.
It seems that the best way to explain this fact is to assume that our perceptual system
plays an active part in producing a perceptual experience. One idea is that because
of previous experiences of faces which were all convex, the visual system ‘judges’ it
to be much more likely that the present experience is of a convex surface, and hence

9
There is a debate about how this ‘work’ is best characterized: for example, as an unconscious
inference following Bayesian principles, or merely as a process conforming to various laws. This debate
is not our concern here: we simply want to rely on the fact that the content of perceptual experiences
is not entirely determined by the stimuli on our perceptual organs, but requires the active contribution
of the perceptual system as well.
 The Constituents of Perceptual Content and the Role of Perception
364   

infers from the ambiguous stimuli to the presence of a convex surface. Of course,
if there is inference here, it must be unconscious, and some people even object to
saying that the visual system can do anything like judging or inferring, and some
other expression should be used. Be that as it may, the unconscious inference cannot
be understood as starting from a perceptual experience and resulting in a separate
mental state. Instead, the inference precedes the experience itself: there is no doubt
that we have a genuine visual perception of the face as convex.
The relevant examples vary a great deal, but the upshot is the following. Just
because unconscious association or interpretation, based on previous experiences,
is involved in a perceptual process, this doesn’t mean that there is first a percep-
tual experience, and then the interpretation produces a separate judgment. The
unconscious interpretation may well be part of the formation of the perceptual
experience itself. We need an explanation of why unconscious interpretation in
the coffee case is different from unconscious interpretation in the case of the hol-
low face illusion; until the difference is demonstrated and shown to have explan-
atory advantages over our proposal, phenomenal presence remains the best guide
to what the object of perception is.
We did say above that phenomenal presence may not settle all issues about
the objects of perception. We do insist, however, that it is our primary guide, and
if phenomenal presence singles out something as an object of a perceptual experi-
ence, we need further, and decisive reasons to dismiss the testimony of phenom-
enal presence. On this basis, coffee qualifies as an object, and the fact that this is
the result of an unconscious interpretation does not disqualify its status as object
of perceptual experience.
One question may remain, however: what is then the difference between the
coffee case and the case of ‘seeing’ that it’s cold? Why doesn’t the second qualify
as a genuine case of perceiving temperature? We shall answer this question in the
next section.

15.9. Objects of Tactile Experiences

In this section and the next we shall illustrate how the object of perception can
move along the causal chain in the case of experiences in the tactile modality.
Philosophers have expressed different views about what should be regarded as the
object of touch. On the most ‘minimal’ version, the object is some part of the
skin: in touching something, we feel a pressure on our skin. A possible variation
on this view would also include the muscles communicating the magnitude of
pressure.
Perhaps some people don’t regard the skin or the muscles as proper ‘objects’
of experience; they may argue that these cases are better described as mere sensa-
tions, that is, sensory experiences without a perceptual object. We do not need
to take issue with this, because this way of looking at things still fits our general
 Which Causes of an Experience Are Also Objects of the Experience?   365

thesis: the object of experience is not fixed for a certain sensory modality. However,
what matters here is that tactile objects and bodily parts are different links identi-
fied along the same causal chains that result in perceptual experiences. For the sake
of simplicity, we will continue talking of ‘tactile experiences’ as covering both cases
of bodily and tactile perception.
If the skin and what’s within the skin are the only objects of a tactile experi-
ence, anything that extends further than the skin must be the object of a separate
mental state. This is the view held by Fred Dretske, for example:
. . . even if we can tell the shape of things by touch (i.e., feel that it is square),
I do not think this shows we feel the object’s shape. What we feel when we
tell the shape of objects in this way is pressure and (if this is really different)
texture. It is differences in pressure we feel as we move our hand around the
object that (together with what we know about how we are moving our hand)
tell us what shape an object is. [ . . . ] If this is to count as feeling shape, then
wine connoisseurs must be tasting colors. (Dretske 2000, pp. 458–459)10
To be precise, Dretske is talking here about the question of which properties, rather
than which things are perceived. But there is an obvious connection between these
questions: perceiving a thing is perceiving it as being in some way, that is, as exem-
plifying some properties. Presumably, the properties that enter the content of an
experience are the ones exemplified by the things we perceive. Hence if shape and
texture are not something we literally perceive when having a tactile experience,
then the object that has the shape and texture is not the object of the experience.
Dretske’s claim about tactile experiences falls within our discussion from the
previous section and so all the considerations and conclusions there apply to the
present case. That is, since objects outside the skin often have an immediate phe-
nomenal presence in tactile experiences, on our account, they qualify as objects
of perception.
We agree that it would be odd to grant that wine connoisseurs taste colors,
but we deny that tactile experience of extrabodily objects is analogous to ‘tasting
colors’. First, even for expert wine tasters, it is questionable how much the color of
the wine is phenomenally present in a gustatory experience. Take one of the more
easily identifiable grape varieties, for example pinot noir. It seems conceivable that
similarly to the example of the pine tree, someone can develop a reliable recog-
nitional capacity for pinot noir through tasting, and being pinot noir becomes
part of the content of their gustatory experience. Pinot noir is red, so one can
readily identify the color of the wine on the basis of the flavor. But this isn’t the
same as having redness phenomenally present in one’s gustatory experience: the

10
That is, the claim that extrabodily objects are not the objects of tactile experiences is Dretske’s
view only in case he regards feeling texture as “really different” from feeling pressure. We’ll assume here
that this is his view while the reader should keep in mind the qualification ‘unless texture is different
from pressure’.
 The Constituents of Perceptual Content and the Role of Perception
366   

identification of the pinot noir flavor does seem to be an intermediary step on the
way to the identification of color.
Why is there an intermediary step? Because it seems that there is no phenom-
enally unified category of the flavor of red (or the flavor of white, for that matter).
Red wines of different grape varieties or by different makers taste very different,
and then we haven’t even considered other red beverages, like raspberry juice or
rooibos tea. This is why the identification of color on the basis of flavor seems to us
indirect: the flavor that indicates red is too diverse to make it possible for red to be
presented phenomenally in a gustatory experience.
To come back to the question we left unanswered at the end of the previous
section: the same holds for our earlier example of ‘seeing’ that it’s cold. The visual
appearance of it’s being cold is just too diverse to form a category that could be
directly phenomenally present in experience. We learn that it’s cold when we see
snow, ice, people wearing thick coats, their having red noses and blue lips, their
breath steaming in the cold air, and so on. There is no such thing as phenomenal
category of ‘the look of cold’.
But the case of felt shapes, textures and temperatures is different. In these
cases, there is a phenomenally robust quality that is present in tactile experi-
ences. That is why we find it overwhelmingly plausible that in these cases, the
thing which has the shape or texture can be a genuine object of the perceptual
experience.
As before, to show that the object of experience moves along the causal chain,
we need to present instances both of feeling only the condition of the skin or bodily
part, and other instances when we feel (also) the object that touches the skin. We
have already mentioned a couple of relevant examples. The most plausible cases of
feeling only a bodily part are cases of unexpected, fleeting and unfamiliar tactile
sensations. If the sensation is unpleasant—for example a sort of sting verging on
the slightly painful—then it’s all the more plausible that the center stage of one’s
phenomenal awareness in the experience is taken up by the skin, with no room left
for the offending object that caused the sensation.
How different this is from a purposeful search among familiar objects. The
earlier example was searching for a key; another one may be finding the utterly
familiar location of the switch of your bedside lamp in the dark. It seems to us that
as soon as one’s finger hits on the switch, we do genuinely feel its presence.

15.10. Further Explored Objects

So far we mentioned two categories of possible objects for tactile experi-

ences: bodily parts, and objects immediately touching the body. In addition, fur-
ther explored objects may become the objects of touch. Aristotle remarks that an
object that comes in contact with the skin is at once perceived. Furthermore,
 Which Causes of an Experience Are Also Objects of the Experience?   367

if the experiment is made of making a sort of membrane and stretching it

tight over the flesh, as soon as this web is touched the sensation is reported
in the same manner as before, yet it is clear that the organ is not in this mem-
brane. If the membrane could be grown on to the flesh, the report would
travel still quicker (Aristotle, Book II, §11 422b33‒423a12).
A contemporary example could be a surgeon wearing a surgical glove. Although
the glove is the only thing that is strictly speaking in contact with the skin, the
surgeon can plausibly be said to feel also the shape of the scalpel she is manipu-
lating. More striking examples involve putting more distance between the skin
and the perceived object, for example in the case of exploring the shape and
surface of objects with a stick—that is, if one finds the following description by
Merleau-Ponty plausible:
Once the stick has become a familiar instrument, the world of feelable things
recedes and now begins, not at the outer skin of the hand, but at the end of the
stick. . . . the habit does not consist in interpreting the pressures of the stick on
the hand as indications of certain positions of the stick, and these as signs of
an external object, since it relieves us of the necessity of doing so. The pressures
on the hand are no longer given; the stick is no longer an object perceived by
the blind man, but an instrument with which he perceives. (Merleau-Ponty
1945, pp. 175–176)
An interesting example of tactile experience acquiring new objects can be the case
of the users of tactile visual substitution system (TVSS, see Bach-Y-Rita1996).
A camera records an image which is simplified into a black-and-white pixellated
image, which in turn is converted into vibratory or electric stimuli produced by a
plate against the skin of the subject. The subject feels the converted shapes through
tactile experience. After sufficient training, subjects learn to locate and manipulate
objects in space. Interestingly
when the man-machine interface, the electro—or vibrotactile array, is moved
from one area of skin to another (e.g., from the back to the abdomen or to the
forehead), there is no loss of correct spatial localization, even when the array is
switched from back to front, since the trained blind subject is not perceiving the
image on the skin, but is locating it correctly in space. (Bach-Y-Rita 1996, p. 500)
On the basis of considerations put forward above, we regard it as plausible that the
object whose image is conveyed through the array becomes here the object of the
experience.11

11
Tactile visual substitution system (TVSS) is but one of the many sensory substitution systems
that have been developed and tested, and other cases seem to confirm that the subjects using the
systems come to perceive new objects previously not experienced in the modality (see Bach-y-Rita
1996).
 The Constituents of Perceptual Content and the Role of Perception
368   

Both Merleau-Ponty’s description and the reports about using sensory substi-
tution systems raise a question that we already touched upon in passing: when a
more remote link in the causal chain is added as an object, do we cease to perceive
the more proximate links? Going back to the case of olfaction, the question would
be whether we cease to perceive an odor, once its source becomes the object of the
olfactory experience. Or similarly for tactile modality—do we cease to perceive
what happens to the palms of our hands once we engage in an active explora-
tory touch of an extrabodily object? We cannot settle the question here, but some
possible answers were already mentioned. Lycan’s layered-representation view of
olfaction holds that we always perceive the odor as well as the source of the odor.
Merleau-Ponty has a different view in the case of touch: once the stick becomes a
familiar tool, it ceases to be an object of perception, and the only things felt are the
explored surfaces.

15.11. Conclusion

It’s part of the phenomenology of perceptual experiences that objects seem to be

presented to us. The first guide to objects is their perceptual presence. Further re-
flection shows that we take the objects of our experiences to be among the causes
of our experiences. However, we do not think that all causes of the experience are
also objects of the experience. This raises the question indicated in the title of this
paper. We argued that taking phenomenal presence as the guide to the objects of
perception, we can see that at least in two sensory modalities, smell and touch,
there is no uniform answer to this question. The objects of olfactory and tactile
experiences can move along the causal chain.

References

Aristotle (1995). On the soul. Trans. by J. A. Smith. In Jonathan Barnes (Ed.), The Complete
Works of Aristotle (pp. 641–692). Princeton: Princeton University Press.
Bach-Y-Rita, Paul (1996). Sensory substitution and qualia. In Alva Noë & Evan Thompson
(Eds.), Vision and Mind (pp. 497–514). Cambridge, MA: MIT Press (reprinted in 2002).
Bayne, Tim (2009). Perceptual experience and the reach of phenomenal content. The
Philosophical Quarterly, 59, 385–404.
Dretske, Fred (2000). Reply to Lopes. Philosophy and Phenomenological Research, 60,
455–459.
Fodor, Jerry A. (1987). Psychosemantics. Cambridge, MA: MIT Press.
Gregory, Richard (1973). Illusion in Nature and Art. London: Duckworth.
Lycan, William G. (1996). Layered perceptual representation. Philosophical Issues, 7, 81–100.
Merleau-Ponty, Maurice (1945). Phenomenology of Perception. Trans. by Colin Smith.
London: Routledge & Kegan Paul, 1962; translation revised by Forrest Williams, 1981;
reprinted, 2002.
 Which Causes of an Experience Are Also Objects of the Experience?   369

Nudds, Matthew (2004). The significance of the senses. Proceedings of Aristotelian Society,
New Series, 104, 31–51.
O’Callaghan, Casey (2007). Sounds: A Philosophical Theory. Oxford: Oxford University
Press.
Peacocke, Christopher (1983). Sense and Content: Experience, Thought and their Relations.
Oxford: Oxford University Press.
Sanford, David H. (1976). The primary objects of perception. Mind, 85, 189–208.
Siegel, Susanna (2006). Which properties are represented in perception? In Tamar Gendler
Szabo & John Hawthorne (Eds.), Perceptual Experience (pp. 481–503). Oxford: Oxford
University Press.
Siegel, Susanna (2013). The contents of perception. In Edward N. Zalta (Ed.), The Stanford
Encyclopedia of Philosophy. \ http://plato.stanford.edu/archives/spr2013/entries/
perception-contents/.
Strawson, Galen (2008). Real intentionality 3. In id., Real Materialism and Other Essays (pp.
281–305). Oxdford: Oxford University Press.
Travis, Charles (2004). The silence of the senses. Mind, 113(449), 57–94.
Tye, Michael (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.
INDEX
Access conscious, 331
Accuracy, argument from, 230
Accuracy conditions, 61, 70, 71, 204
Action-first approach, 51, 51n2, 52, 65, 66
Adverbialism, 4, 16, 33, 179–180
Aesthetic rightness, 71
Affordances, 7–951–75; and experienced
mandates, 53–73; and representation, 61–73
Allorepresentation, 12–14, 138–178; agreement,
149–152; audience, 151; and Cartesian
thinkers, 140–144; collapse, 166–173;
deference, 152–155; finding and presenting,
161–166; force, 159–161, 170; generality,
144–147; recognition, 155–158; selection,
148–149
Analogue representation, 81n10
Analytic inferences, 90
Answerability contents, 55–56, 70, 71, 72–73
Anti-representationalism, 5–7, 39–50; and dorsal
perception, 42–45; and enactivism, 39–40;
and multimodality of perception, 45–47; and
relationalism, 40–42
Appearances, argument from, 8, 14, 231–232
Appears-looks conception of experiential
content, 223n2, 224n3
A-relation, 202
Argument from accuracy, 230
Argument from appearances, 8, 14, 231–232
Argument from belief generation, 234
Argument from illusion, 313n5
Argument from phenomenal contrast, 28–29, 72,
315, 316–318, 355–356
Aristotle, 352, 356n4, 366
Armstrong, David, 111, 136
Aspect-perception, 321–322, 323–326
Association, 17, 201, 214, 362–364
Attention, 65, 244–245, 244n4, 248, 319, 319n9
Attentive sensory episodes (ASEs), 129–132, 134
Attitude operator argument, 9–10, 92
Auditory experiences, 274, 278, 356
Austere realism, 330, 347–349
Austere relationalism, 199–219; accuracy
condition objection, 204; argument for
relational content, 209–213, 214; association
thesis, 201, 214; content thesis, 200–201,
204–213; epistemological objection, 205;
grounding objection, 205; indeterminacy
objection, 204; master argument, 207–209,
214–217; objections, 204–206; particularity
objection, 204; phenomenological objection,
205; propositional attitude thesis, 201–202
Autorepresenting, 12–14, 138–139, 169
Availability coding, 276
Ayer, A. J., 4
Ballard, Dana, 40
Barnard, Robert, 330
Barwise, J., 297n11
Batty, Clare, 275n19, 279n24
Bayesian principles, 363n9
Behavior-based AI, 40
Belief generation: argument from, 234; and
image content, 266, 268–271; and naïve
realism, 20, 221, 234; and reasons, 80, 81–85
Belief-independence of experience, 82n14, 85
Bipolarity of experience, 112–113, 126
Block, Ned, 325, 331
Boghossian, Paul, 276, 276n21
Brain Grey, 125–126
Breckenridge, Wylie, 4
Brewer, Bill, 80, 199n1, 208, 214–216, 214n20,
215n21
Bridge principles, 87, 87n21
Broad, C. D., 4
Brogaard, Berit, 1, 99n38
Brown, Angela, 185
Budek, Tomasz, 31–32, 351
Byrne, Alex, 9, 80n7, 95n30, 96, 100n38, 109, 111,
132–133, 136, 201n5, 202
Camp, Elisabeth, 270n9, 276n22
Campbell, J., 199n1, 200, 206
Capacity: allorepresentation, 151, 158, 168;
discriminatory, 210n16; perceptual, 210, 212
Carey, Susan, 277
Carnap, Rudolf, 269n7
Carroll, Lewis, 87
Carrying information, 360
Cartesian thinkers, 140–144, 148–149
Casati, Roberto, 294–295
371
 Index
372   
Categorization, 182n6, 191–194
Causal relations, 31–32, 351–369; and association,
362–364; further explored objects, 366–368;
and objects of experience, 352–354; olfactory
experiences, 356–358; phenomenal presence
of objects, 358–362; tactile experiences,
364–366; and unconscious inference,
362–364; and visual representation, 312
Chalmers, David, 31, 337, 337n7
Change blindness, 188n14
Chisholm, Roderick, 4, 14, 93, 106, 110
Chomsky, Noam, 143
Churchland, P., 315, 316
Clark, Austen, 269n6, 271n10, 312n2
Clarke, Thompson, 160
Co-extensionality, 97, 99, 99n38
Cognitive phenomenology, 81n11, 332
Co-intensionality, 98, 99n38
Color: and phenomenal intentionality, 31,
334–340; subjectivity of, 323, 336
Common kind theory, 179
Communicative efficacy, 71
Compartmentalization, 80n7
Computational theory of vision, 51
Conative phenomenology, 332
Conduct, M. D., 190n16, 192
Conjunctivism, 127
Constitutive dependence, 30n6
Content, 61–74; answerability, 72–74; appearance,
76–100; and austere relationalism, 200–201,
204–213; causal, 122–124; content view, 3,
5–11, 105–108; Fregean, 124–126, 154–155, 203;
gappy, 24, 294–299; imagistic, 26–27, 266–284;
layered, 319–321; modifying content view,
126–128; and naïve realism, 221–230; neutrality
of, 27; and objects of experience, 351–352;
propositional, 76–100, 108, 259–260, 267;
representational, 61–72, 299–300; Russellian,
24–25, 119–121, 203; Searlean, 122–124; source
of content view, 111–118; strong, 3, 207–209;
and visual representation, 175–177, 242–245;
weak, 2, 18–22, 200–202, 230–237
Content-first approach, 65
The Contents of Visual Experience (Siegel), 2
Contextual semantics, 330, 346–347, 348n11
Co-phenomenality, 10–11, 97, 98–99, 100n38
Copular verbs, 22n5, 23
Correctness conditions, 77
Cosmopolitanism, 257–258
Crane, Tim, 236, 268n4
Cussins, Adrian, 7, 8, 52
Datable sensory episodes, 106
Davidson, Donald, 77, 77n3
Davies, M., 293n5
Deep nonobjectivity, 258–260
Defeasibility, 88
Deflationary views, 284–287
Delusions of reference, 56
Descartes, René, 140. See also Cartesian thinkers
Desire, 79
Diamond, Cora, 155
Dilemma of stuttering and silence, 86
Discrimination, 182n6, 191–194, 210n16. See also
Indiscriminability
Disengagement, 253–256
Disjunctivism, 14–16, 179–195; categorization,
191–194; and disengagement, 254–256; and
hallucinations, 292, 293n4; indiscriminability,
15, 181–184, 191–194; Martin,s argument
against, 229n13; and naïve realism, 225–230;
nontransitivity, 184–190; positive, 229–230;
and relationalism, 41; and strong content
view, 3–4, 126–128; token experiences,
190–191; and visual representation, 311
Distality, 320n11
Distributed cognition, 40
Dorsal stream, 5–6, 42–45, 62
Doxastic account, 79–80, 81, 85
Dretske, Fred, 188, 188n14, 314n8, 365, 365n10
Dreyfus, Hubert, 7, 8, 52, 59, 60, 62, 73
Dynamical systems theory, 40
Ebbinghaus illusion, 6, 43
Ecological psychology, 40
Effect-representing, 12–14, 144–147
Efficient-causal explanation, 251
Egocentric space, 42
Embodied cognition, 40
Empirical realism, 258
Enactivism: affordances, 51–74;
anti-representationalism, 3, 5–7, 39–40;
mandates, 53–74; multimodality of
perception, 46, 275
Envatment, 341–345
Epicurus, 265–266, 285
Epistemology of perception: and austere
relationalism, 204–206; and beliefs, 81–91;
and content view, 107, 110–111; and contextual
semantics, 346–349
Evaluation switcher, 100n38
Evidential support, 86
EX-ing, 109–111, 133–136, 202
Experienced mandates, 7–9, 53–60; answerability
contents, 72–73; as challenge to centrality
of representation, 58–60; defined, 52;
motivation of, 67–69; and representation,
61–73

Experiences: auditory, 274, 278, 356;
belief-independence of, 82n14, 85; bipolarity
of, 112–113, 126; and justification, 85–91;
modularity of perceptual experience, 28–30,
80n7, 82n14, 315–316; objects of, 352–354;
olfactory, 31–32, 275, 275n19, 278–279, 356–358;
presentational sensory experience, 107–108;
and reasons, 81–85; soliciting, 55; tactile,
274–275, 278, 364–366; token, 190–191, 303
Experiential representation, 81n10
Externalistic intentionality, 332–334, 340–341
Facts, perception of, 128–129
Farkas, Katalin, 31–32, 351
Farrenikova, A., 312n3
Feature-placing structures, 271–273, 272n13,
277–279
Felt solicitation, 55
Ferrante, D., 324
Fish, Bill, 229
Flavor, 275n17, 279
Fodor, Jerry, 312, 315, 359
Folk-psychology, 82
Force-selection, 160, 170
Frege, Gottlob, 145, 146, 154, 172–173, 175, 203
Fundamentality, 201
Further explored objects, 366–368
Gappy content, 24, 293, 294–299, 303
General compositionality, 100n38
Generalizations, 144–147, 161
Gerbino, I., 324
Gestalt psychology, 271
Gibson, J. J., 51, 275
Glüer, Kathrin, 9–11, 76
Gombrich, Ernst, 315
Graff, D., 184, 189
Graham, George, 331. See also
Graham-Horgan-Tienson (GTH) account
Graham-Horgan-Tienson (GTH) account,
329–350; and austere realism, 347–349; and
colors, 334–340; and contextual semantics,
346–347; and envatment, 341–345; and
left-right inversion, 341–345; phenomenal
intentionality vs. externalistic intentionality,
332–334; primary/secondary distinction,
345–346; role of, 340–341; and spectrum
inversion, 341–345
Grice, P., 314n8
Grounding presuppositions, 30, 205, 334
Gupta, Anil, 285, 286, 287n26
Habit-discontinuity (HD) thesis, 63
Habit formation, 62
Hallucinations, 24–25, 291–308; and austere
Index   373
relationalism, 211; and causal relations, 352;
and content view, 111–113; and disjunctivism,
179; gappy content view, 24, 294–299;
indexicals view, 302–303; misperception vs.,
352n1; and naïve realism, 226–227; negative
account of, 226–227; phenomenally different,
305, 305n24; positive account of, 226, 227n10;
possible worlds conception, 299–302; and
presentation, 253; and representationalism,
304–306; and strong content view, 3–4
Haptic experiences, 275
Harman, Gilbert, 111
HD (habit-discontinuity) thesis, 63
Heck, Richard, 78–79, 80
Heim, Irene, 25
Hellie, Benj, 22, 242
Hering Illusion, 115, 117
High-level properties, 18, 21, 28–30, 311–313
Hollow face illusion, 43, 363
Horgan, Terry, 30–31, 329, 331, 331n4, 335n5.
See also Graham-Horgan-Tienson (GTH)
account
Hume, David, 267n3, 269, 281–282
Hyperintensionality, 10–11, 92–93, 99n38
Hyper-intrinsic approach, 341
Iconic representation, 81n10
Illusions, 11–12, 112–136; argument from,
313n5; and austere relationalism, 216–217;
and beliefs, 80n7; and causal relations,
352; and disjunctivism, 179; and dorsal
perception, 43; and naïve realism, 225–226;
and presentation, 253; and redundancy of
EX-ings, 134–135; and shifting aspects, 153
Image content, 24–26, 265–290; and belief, 268–
271; deflationary views, 284–287; elements
of, 271–277; limitations, 279–280, 281–282;
and naïve realism, 287–288; part-whole
composition, 277; and perception, 266–268;
predicative feature-placing, 271–273, 277–
279; property coding, 276; and propositional
content, 267–268; shared, 280–281, 283–284;
spatiotemporal connectedness, 273–275
Immediacy, 82n14
Imperatival contents, 71n26
Inattentional blindness, 63
Indeterminacy, 204
Indexicals, 25, 302–303
Indiscriminability, 15, 180, 181–184, 191–194
Information-carrying, 44–45
Instancing, 145, 146
Instantiation, 183
Intelligibility, 250
Intentionality, 59, 73, 200n2, 224, 304. See also
Representation
 Index
374   
Intentionality (Searle), 111
Internal sense-data, 246
Introspection, 313
Inverted spectra, 95n29, 228, 256–257, 341–345
Jackson, Frank, 1, 2, 4, 93, 95
Johnston, Mark, 11–12, 105, 109–110, 120, 276,
305n22
Judgmental color-properties, 338–339
Judgmental content, 335, 335n5, 338, 343
Kanizsa compression illusion, 43
Kant, Immanuel, 257, 273
Kaplan, David, 25, 155, 302–303
Kaplan,s paradox, 260
Kelly, Sean, 52, 52n4, 58n12
King, Jeff, 298, 298n12
Klein, C., 71n26
Kriegel, Uriah, 331, 341
Larger Ball Illusion, 117
Layered representation, 357, 368
Left-right inversion, 341–345
Levels of distality, 320n11
Lewis, David, 77, 242, 242n1, 244, 255
“The Limits of Self-Awareness” (Martin), 226
Lindsey, Delwin, 185
Logue, Heather, 8, 12, 18–22, 220
Looks-contents, 9, 81, 87n22, 91–100, 313–314
Looks-indexed, 82n14
Lycan, William G., 28–30, 311, 320, 323, 357, 358, 368
Lyons, J., 312, 331n4
Mach square/diamond, 324
Macpherson, Fiona, 323, 323n13, 324–325
Magnitude, 131, 193
Mailbox proposal metaphor, 296
Marr, David, 51, 311
Martin, Michael G. F., 15, 94n28, 179, 180–181,
183, 183n7, 189, 190, 191–192, 199n1, 226, 227,
229, 229n13, 286
Materialist Theory of Mind (Armstrong), 111
Matthen, Mohan, 26–27, 265, 271n10
Maund, B., 336n6
McDowell, John, 46, 80, 127, 138, 250, 270
McGinn, Colin, 274n15, 293n5, 305n22
Meaning-constitutive inferences, 90, 312
Memory: and image content, 267; perception
filled in by, 121–122; perceptual, 193
Merleau-Ponty, M., 52, 59, 59n13, 367, 368
Metaphysical supervenience, 30n6
Millikan, Ruth G., 109, 312, 314n8, 319–320,
320nn11–12, 357
Misleadingness, 82n14
Misperception, 314, 352n1
Modes of presentation, 246
The Modularity of Mind (Fodor), 312
Modularity of perceptual experience, 28–30,
80n7, 82n14, 315–316
Monitoring, 319n9
Motion perception, 278
Motivation, 67–69, 79
Moving Object phenomenon, 277–278
Müller-Lyer illusion, 19, 28, 43, 85, 116–118,
216–217, 232, 335
Multimodality of perception, 7, 45–47
Music perception, 274n16
Naïve realism, 18–22; and content views,
225–230; and hallucinations, 4, 292; and
image content, 285, 287–288; Medium
Content View, 221, 223, 239; Mild Content
View, 221, 223, 230–237; non-propositional
experiential content, 221–225; and objects
of perceptual experience, 353; reconciling
content views, 237–240; and relationalism,
199n1; Spicy Content View, 221, 224, 239,
240n23; and strong content view, 3; and
visual representation, 311, 314n7; and weak
content view, 3
Nanay, Bence, 5–7, 23, 39, 203n9
Negative account of hallucination, 226–227
Neurath, Otto, 269n7
Neutrality of content, 27
Nickel, Bernhard, 325
No-Content View, 285–286
Noë, A., 320
Non-ambiguity coding, 276
Non-classical connectionism, 40
Non-conceptual content, 213, 292
Non-experiential belief, 80, 82
Non-illusory/illusory distinction, 108, 113–114,
118, 126
Nonobjectivity, 258–260
Non-propositional content, 213, 221–225
Non-psychological facts, 150
Non-soliciting affordances, 54, 54n7
Nontransitivity, 180, 181n4, 184–190
Nonveridical phenomenology, 334. See also
Hallucinations
Object-directed idioms, 106
Objectivity, 246–247, 250
Objects of experience, 352–354
Object view, 199n1
Odors. See Olfactory experiences
Olfactory experiences, 31–32, 275, 275n19,
278–279, 356–358
On Clear and Confused Ideas (Millikan), 109
Optic ataxia, 42
Orlandi, N., 321

Pace, Michael, 86n20
Pagin, Peter, 100n38
Parochialism, 258
Particularity, 82n14, 204–206, 209–214
Part-whole composition, 277
Pautz, Adam, 202, 223n2, 224n3
Peacocke, Christopher, 269n6, 272n13, 275n18,
320, 324, 325, 357
Perceived conditions, 62
Perception: A Representative Theory (Jackson), 1, 2
Perception and its Objects (Brewer), 214
Perception as belief, 76–100, 135–136
Perceptual verbs, 22, 22n5
Perry, J., 297n11
Personal/sub-personal distinction, 47
Phenomenal consciousness, 332
Phenomenal contents, 80
Phenomenal continuum, 183–184
Phenomenal contrast, 28–29, 72, 315, 316–318,
355–356
Phenomenal intentionality, 30–31, 329–350;
and austere realism, 347–349; and colors,
334–340; and contextual semantics, 346–347;
and envatment, 341–345; GTH framework,
332–334; and justification, 86n20; and
left-right inversion, 341–345; primary/
secondary distinction, 345–346; and
spectrum inversion, 341–345
Phenomenally different hallucinatory
experiences, 305, 305n24
Phenomenally equivalent predicates, 97–98
Phenomenal presence of objects, 358–362
Phenomenology: appearances, 14, 76–100,
329–332; and austere relationalism, 204;
looks, 9, 22–23, 87n22, 91–100, 313–314;
nonveridical, 334
Phenomenology of Perception (Merleau-Ponty),
59n13
Plato, 265
Pollock, John, 88, 90
Portrait-like sensory appearance, 266
Positive account of hallucination, 226, 227n10
Positive disjunctivism, 229–230
Possible worlds conception, 26–27, 299–302
Potrč, Matjaž, 330, 347
Pragmatic contradiction, 254
Predicative feature-placing, 271–273, 272n13,
277–279
Presentation, 242–261; challenges to, 253–258;
and disengagement, 253–256; importance
of, 250–253; modes of, 246; semantics for
deep nonobjectivity, 258–260; and sensory
consciousness, 245–250; and subject’s
contribution, 256–258
Presentational color-content, 340
Index   375
Presentational content, 335, 338, 343
Presentational sensory experience, 107–108
Price, H. H., 4
Primary/secondary distinction, 345–346
Pro-attitudes, 79
Probabilification, 86
Proffitt, D. R., 66n20
Proper sensibles concept, 356n4
Property coding, 276, 276n22, 279
Propositional attitudes, 10, 77n5, 84, 106–107,
108, 130, 201–202
Propositional content, 77, 77n3, 79, 267–268
Proto-affordances, 53–54
Protocol sentences, 269, 269n7
Proto-objects, 312n2
Proximal provocation, 161–162
Pryor, Jim, 90n24
Psychological facts, 150
Psychosemantics, 314, 323n13
Psychosemantics (Fodor), 359
Public space interpretation, 269n7
Putnam, Hilary, 140
Pylyshyn, Z., 312n2
Radical Naïve Realism, 233–234, 233n18
Raffman, Diana, 14–16, 179, 187n13
Rationalization: of mandated actions, 64; and
presentation, 243, 250; psychological, 66; and
reasons, 83; subjective, 84
Read, Thomas, 3
Reasons, 9–11, 76–102; and beliefs, 81–85; and
contents, 76–81; and experiences, 81–85; and
justification, 85–91; and “looks” conception,
91–100
Rebutting defeaters, 88
Reference-constituting experiential
acquaintance, 334
Reid, Thomas, 18
Relationalism, 5–7, 16, 40–47, 94, 199–219.
See also Austere relationalism
Reliably truth-preserving, 86
Relief, 73
Representation, 30–31, 138–148; affordances,
58–60; and aspect-perception, 323–326;
and content views, 12–13; experienced
mandates, 61–73; and hallucination,
304–306; instantiation, 165–173; and
phenomenal intentionality, 329–332,
331n2; and presentation, 251–252; and
recognition, 158–161; and relationalism,
206; representationalism, 2, 16, 39, 51–74,
199–205, 216, 304–306, 323–326; visual. See
Visual representation
Representing-as, 139
Rigidifying operator, 259
 Index
376   

Robust intentional properties, 180 Strong content views, 1–5

Rock, I., 324 Strong externalism, 340–341
Russell, B., 4, 202–203, 202n6, 253–254, 294 Strong modularity thesis, 315
Sturgeon, S., 190n16
Salva veritate substitutability, 99n38 Stuttering and silence dilemma, 86
Sanford, David, 356 Stuttering inference argument, 80n8
Sartre, J. P., 68 Subjective rationality, 84, 256–257
Sauret, Wesley, 319n9 Subject-raising verbs, 22n5, 23
Scenario content, 269n6 Sub-personal/personal distinction, 47
Schellenberg, Susanna, 12, 16–18, 199, 224n3, 321 Substantive properties, 180
Searle, John, 12, 92, 111, 112, 122–124, 280 Substitutability, 99n38
Second order states, 83–84 Substitution principle, 10
Seems-content link, 208, 209n14 Success conditions, 112
Selection process, 148–149 Syntax, 154
Self-referential causal content, 124
Semantic value, 24–25 Tactile experiences, 274–275, 278, 364–368
Sense-datum view, 4–5, 16, 221, 225, 253, 291–293, Tactile visual substitution system (TVSS), 367,
311, 317 367n11
Sensible profiles, 121n2 Teleofunctionalism, 320n11, 357
Sensorily entertaining relation, 202 Tension, 59, 73
Sensory consciousness, 245–250, 252 That-clause constructions, 353
Sensory illusion, 265. See also Illusions Thought (Harman), 111
Sensory imagination, 229 Tienson, John, 329, 331. See also
Sensory modalities, 31–32, 45–47, 81, 356–366 Graham-Horgan-Tienson (GTH) account
Sensory-motor habit, 62 Timmons, Mark, 330
Sensory qualities, 322–323 Token ASEs, 129
Sentence-sentence inference, 270 Token content, 212, 212n18
Separatism, 331, 331n4 Token experiences, 190–191, 303
Seuss, T. G., 57n11 Touch sensory experiences, 274–275, 278.
Shared content principle, 280, 283, 288 See also Tactile experiences
Shared image content, 280–281, 283–284 Transcendental aesthetic, 273
Siegel, Susanna, 2, 7–9, 13, 14, 18, 19, 28–29, 51, Transcendental idealism, 258
110, 180, 181n4, 183, 190, 193, 201n5, 202, Transparency, 82n14, 259, 276n21
224n3, 230, 233, 233n18, 279–280, 313, 314n6, Travis, Charles, 12–14, 31, 96n32, 132–133, 138,
315, 316–319, 355 287n26
Siewert, Charles, 268n4 Trope nominalism, 121n4
Signaling, 360 Truth, 77, 346
Similarity coding, 276 Truthmakers, 121–122, 238
Sine wave illusion, 116 TVSS (tactile visual substitution system), 367,
Skill-formation, 62 367n11
Smaller Center Illusion, 117 Two-state view, 3
Smell. See Olfactory experiences Tye, Michael, 24–26, 291, 323n13
Soliciting affordances, 54, 54n7, 70, 72
Soliciting experiences, 55 Unconscious inference, 362–364, 363n9
Sosa, Ernie, 21 Undercutting defeaters, 88
Spatial connectedness, 273–274 Unified views, 16–23, 199–240, 250–255
Spatial matrices, 274n15, 282 Uniformity, 300
Spatiotemporal connectedness, 273–275, 295,
334, 356 Varzi, Achille, 294–295
Specialized skill actions, 60 Vehicle, 139
Spectator-first approach, 51, 51n2 Velleman, David, 276
Spectrum inversion, 95n29, 228, 256–257, Ventral stream, 5, 42
341–345 Ventriloquism, 45
Stefanucci, J. K., 66n20 Veridical hallucinations, 118, 123
Stoicism, 265 Veridical illusions, 123
Stream of consciousness, 243–244 Visual agnosia, 3, 42
 Index   377

Visual perception, 116. See also Visual
representation
Visual phenomenology, 279
Visual representation, 311–328; and
aspect-perception, 321–322, 323–326;
conservative views on, 311–312; disputes over,
313–315; layering views, 319–321; modularity
thesis, 315–316; sensory qualities, 322–323;
Siegel’s view on, 316–319
Wayward causal chains, 112
Weak content views, 1–5
Weak modularity thesis, 315
Westerståhl, Dag, 100n38
Williamson, T., 179, 184, 189
Wittgenstein, Ludwig, 138, 172, 311, 321, 322
Wollheim, Richard, 322
Xu, Fei, 271n11

Waterfall illusion, 305n23 The Zax (Seuss), 57n11