Does Perception Have Content? 9780199756018

Table of contents :
Cover......Page 1
Half Title......Page 2
Series......Page 3
Does Perception Have Content?......Page 4
Copyright......Page 5
CONTENTS......Page 6
Contributors......Page 8
1 Introduction: Does Perception Have Content?......Page 12
Part One Content Views......Page 48
2 Empirical Problems with Anti-Representationalism......Page 50
3 Affordances and the Contents of Perception......Page 62
4 Looks, Reasons, and Experiences......Page 87
Part Two Against Strong Content......Page 114
5 The Problem with the Content View......Page 116
6 The Preserve of Thinkers......Page 149
7 Disjunctivism, Discrimination, and Categorization......Page 190
Part Three Reconciliatory Views......Page 208
8 The Relational and Representational Character of Perceptual Experience......Page 210
9 Experiential Content and Naïve Realism: A Reconciliation......Page 231
10 Love in the Time of Cholera......Page 253
Part Four Imagistic and Possible-Word Content......Page 274
11 Image Content......Page 276
12 What is the Content of a Hallucinatory Experience?......Page 302
Part Five The Constituents of Perceptual Content and the
Role of Perception......Page 320
13 What Does Vision Represent?......Page 322
14 Phenomenal Intentionality and Secondary Qualities: The Quixotic
Case of Color......Page 340
15 Which Causes of an Experience Are Also Objects of the Experience?......Page 362
INDEX......Page 382

Citation preview

Does Perception Have Content?

PHILOSOPHY OF MIND

Series Editor David J. Chalmers, Australian National University and New York University Self Expression Owen Flanagan

What Are We? Eric T. Olson

Deconstructing the Mind Stephen Stich

Supersizing the Mind Andy Clark

The Conscious Mind David J. Chalmers

Perception, Hallucination, and Illusion William Fish

Minds and Bodies Colin McGinn

Cognitive Systems and the Extended Mind Robert D. Rupert

What’s Within? Fiona Cowie

The Character of Consciousness David J. Chalmers

The Human Animal Eric T. Olson

Perceiving the World Bence Nanay (editor)

Dreaming Souls Owen Flanagan

The Contents of Visual Experience Susanna Siegel

Consciousness and Cognition Michael Thau

The Senses Fiona Macpherson (editor)

Thinking Without Words José Luis Bermúdez

Attention is Cognitive Unison Christopher Mole

Identifying the Mind U.T. Place (author), George Graham, Elizabeth R. Valentine (editors)

Consciousness and the Prospects of Physicalism Derk Pereboom

Purple Haze Joseph Levine Three Faces of Desire Timothy Schroeder

Introspection and Consciousness Declan Smithies and Daniel Stoljar (editors) The Conscious Brain Jesse J. Prinz

A Place for Consciousness Gregg Rosenberg

Decomposing the Will Andy Clark, Julian Kiverstein, and Tillmann Vierkant (editors)

Ignorance and Imagination Daniel Stoljar

Phenomenal Intentionality Uriah Kriegel (editor)

Simulating Minds Alvin I. Goldman

The Peripheral Mind István Aranyosi

Gut Reactions Jesse J. Prinz

The Innocent Eye Nico Orlandi

Phenomenal Concepts and Phenomenal Knowledge Torin Alter, Sven Walter (editors)

Does Perception Have Content? Edited by Berit Brogaard

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

9 8 7 6 5 4 3 2  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

vi    Contents 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 author (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 philosophy—including philosophy of mind, metaphysics, philosophy of language, epistemology, and metaethics.

vii

viii    Contributors

Mark Johnston is the Walter Cerf Professor of Philosophy at Princeton University and the author of many widely reprinted articles in philosophy of mind, metaphysics 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 concerning 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 philosophy 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 perceptual 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 philosophy 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 professor 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 considerable 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 intelligible. 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 factors 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, traditional debates about perception were typically concerned with a different question, 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 external 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 following 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 content (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 perceptual 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 perceptual 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 experience 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 minimally 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 experience 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 describe 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 available 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 believe that being contentful is essential to perceptual experience. On this view, it’s a fundamental feature of perceptual experience that it has representational content, 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, veridical 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 feature of perceptual experiences that they have representational contents. Though naïve realists with traditionalist leanings adamantly deny that perceptual 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 adjacent 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 example, 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 perceiving. 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 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). 2

Introduction    5

represent objects and features in the external world, sense-data are not themselves 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 representationalism 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 instantiated 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 relationalism. 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 incompatible 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

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. FIGURE 1.1 

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 explanation. 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 information. In a three-dimensional model of the Ebbinghaus illusion, the perceptual 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 perception 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 considers turns on the multimodality of perception. Information in one sensory modality can influence information processing in another, even at very early stages of information 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 perception as an attribution of properties by different sense modalities to the same perceived scene. Different sensory modalities represent the same scene as having different properties by matching two modality-specific representations. Anti-representationalists appear to have greater difficulties explaining multisensory perception. The enactivists’ talk about perception as an active exploration of the environment presupposes multimodal perceptual abilities. Likewise, the relationalists’ 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 perception, contrary to what anti-representationalists claim. Susanna Siegel: In her contribution to this volume Susanna Siegel brings the content debate to bear on Gibson’s notion of an affordance. On Gibson’s view, an affordance 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 ongoing 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 situation. 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 situation 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 articulated 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 perceptually 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 attention 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 content, 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, perceptual experiences have a phenomenology that is notably different from the phenomenology 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 experience, 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 experience 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.

10    Introduction

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 operator. So, perceptual contents can be first-order, even if they contain a ‘look’ operator. The reason, she argues, is simple: Propositional attitude operators create hyperintensional 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 justification 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 contexts. 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’ generates 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 hyperintensional 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 between 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 system 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

12    Introduction

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 distinctive 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 perception 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 distinction. 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 objectual attitude. Johnston’s argument is not easily refuted, as it seems right that content by itself cannot play an epistemically significant role and cannot account for illusions and hallucinations. To refute the argument, one might argue that perceptual experience is also fundamentally relational, as Schellenberg and Logue do (see below). One might also attempt to show that experience must be treated as representational 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 experience 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 historical 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 reserved 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 representational 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 covered 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 question 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.

14    Introduction

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 notion 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 particular 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 perceptual 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 colored. 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 indiscriminability 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 settings 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 relation 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 continua. 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 surrounding 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 attempting 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 example, 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 interpretation of the empirical study. The study showed that while subjects report that the

16    Introduction

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 controversial 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 interpretation 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 fundamentally a matter of representing the environment in a certain way and being perceptually 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 environment 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 description sentences. After presenting an argument for the association theory, Schellenberg goes on to defend the representational view. In broad outline, the argument can be summarized 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, perceptual 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 experience. 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 experience 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

18    Introduction

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 assumption (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 content 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 example, 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 accepting 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 sufficient 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 associated 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 perceptual 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 perceptual 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 defeaters 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 experience 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 perceptual 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 content 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 question of why perceptual experiences give rise to particular beliefs and not others.

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. FIGURE 1.2 

20    Introduction

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 associated 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 experiences 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 representationalist, 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 content 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 subject a certain way, we can say that the subject perceptually represents her environment 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 constrain the content of belief originating in the experience. So, this aspect of the epistemological role of experience is grounded in the way the subject perceptually 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 experience 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 representing 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 sometimes 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 seemings 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 distinguished 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, represents the number 64, whereas the inexperienced subject does not have a perceptual experience that represents the number 64. If one simultaneously holds that perceptual seemings are aspects of perceptual experiences, then one can straightforwardly explain the difference between the perceptual seemings of the subjects. Even if perceptual experiences and perceptual seemings come apart, however, 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 content 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

22    Introduction

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, unlike 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 different 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 provides 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 syntactically on clauses but on predicates, which means that they do not operate semantically 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 behavior 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 challenges as the strong versions of naïve realism. By insisting that perceptual experience 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 environment. Does this mean that there are two ways of being related to the environment: 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 unproblematic to grant that this is so.

24    Introduction

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 content. His earlier view was inspired by the problem posed by hallucinations. If the content of veridical perception is object-dependent, it contains concrete particulars 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 , the sentence ‘Sherlock Holmes is a fictional detective’ expresses the proposition . 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 hallucinations 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 . 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 hallucinations 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, specifically 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 composed 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 referent 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 phenomenal character and a function from the perceptual situation to the property that corresponds to the phenomenal character. The content of a veridical or hallucinatory 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

26    Introduction

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 hallucination 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 perceived 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 hallucination 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 representationalism that he favors, viz., the view that phenomenal character supervenes 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 perceptual situations to entities corresponding to the phenomenal character, hallucinations 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 . 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 spatially 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 features 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 experience 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 relations 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.

28    Introduction

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 perception can be meaningfully separated from cognition. According to the modularity of mind hypothesis, systems involved in producing 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 illusion 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 features perceptual states represent. If he is right about this, this might have consequences 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 include high-level natural kind properties can also be used to argue that the constituents 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 recognitional 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 (perceptual 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 perceive 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 cognitive states. But if aesthetic properties and other extremely high-level properties are

30    Introduction

not among the constituents of perception despite contributing to the differences in phenomenal character between our overall experiences, then Siegel’s argumentative strategy cannot be used to settle the debate about whether perception represents 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 representational, 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 experience, is a narrow and intrinsic feature of mentality. Phenomenal consciousness 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 phenomenology. Because phenomenal intentionality is narrow, there is an exact match of phenomenal intentionality between you and your envatted brain physical 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 constitutively depends both on phenomenal intentionality and the external environment. 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 constituted grounding presuppositions that are satisfied by the external environment. These include the presupposition that there is an object at a certain location, that 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. 6

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 project 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 properties 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 central 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 representing 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 consider 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.

32    Introduction

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 perceptual 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 chemicals 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 experience. A more plausible assumption is that the visual experiences of these blind individuals 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 dissociation 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.

34    Introduction 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 phenomenology 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 transcendental 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 representations. 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 perceptual 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 perceptual 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 supposed 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 construct representations at all and, more specifically, we do not need perceptual 1

Enactivism and relationalism often combine, see Noë (2004) and Hellie (forthcoming).

39

40    Content Views

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 distributed 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 objection 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 perceived objects. Perception is a genuine relation between the perceiver and the perceived object—and not between the agent and some abstract entity called ‘perceptual 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

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. 2

Empirical Problems with Anti-Representationalism    41

have object-involving, or maybe gappy, content), it does pose an important question. 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 something 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 representational 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 fundamental 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, especially 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 differentiating 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 perceptual 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.

42    Content Views

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 empirical 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 perceptual system seems to attribute two incompatible property-instances to the same object. If we accept that there are perceptual representations, this is easy to accommodate: 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 different regions of our central nervous system, the ventral and dorsal streams. To put it very simply, the ventral stream is responsible for identification and recognition, 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 malfunctioning, 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 illusion 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, sometimes 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 memory,” 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.’

44    Content Views

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 appealing to this property playing a role in the perceptual processing. One way of resisting this argument would be to argue that we should only consider 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 difference 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 response 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 unconscious 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 philosophical 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 experience it as two flashes (Shams et al. 2000). What is the most important for us from this literature is that the multimodality of perception presupposes that information from two different sense modalities

46    Content Views

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 system 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 insufficient: they can appeal to the active exploration of the multimodal environment, 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 information:  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 different 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 enactivists, who are often explicit about not limiting their attention to personal level phenomenon (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 perceptual 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 exploration 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, understood 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 perception. 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.

48    Content Views 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 illusion:  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 outside 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.

50    Content Views Vroomen, J., Bertelson, P., & de Gelder, B. (2001). Auditory-visual spatial interactions:  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 representations. 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 representation 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

52    Content Views

represented—or whether instead we simply experience affordances without representing 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 affordances. 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 experienced 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 activity, 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 perceptual experiences may have. Regarding Q3, if we could identify a proposition that would characterize the mandate that the subject experiences, then representational 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 section 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 representational 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 mandates 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 affordances 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 subject’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 colliding. If a subject perceptually experienced them, these proto-affordances would characterize how the ball, the rock, the path, and the pedestrians look to that subject. One might associate various actions with the proto-affordances, such as freeing 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.

54    Content Views

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 experienced 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 protection, 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

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. 6

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 motivation or urge to phi? Conceptually, solicitation in the sense used here and motivation 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 exhortation 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 sensitivity 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.

56    Content Views

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 solicitation 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 delusions 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 unavoidable 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 suddenly 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 environment soliciting you to do it. Experienced mandates involve affordances by the perceived environment that you’re currently in. In experienced mandates, solicitation 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 sidewalk 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 traversed 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 sidewalk. 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 mandatory 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.

58    Content Views

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 representation, 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 subsequent 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 motivation built in to experienced mandates, it might seem natural to assimilate these experiences to action that is fueled by dynamic interaction with the environment.

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. 12

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 intelligent 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 solicitations 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 appeals 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 representations 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 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 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.

60    Content Views

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 represented 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 representation 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 experienced mandates and representation?

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). 14

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 mandates are experiences with accuracy conditions, where the accuracy conditions characterize 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 experience 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 preliminary 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 experiences, 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 experienced 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 experiences, 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 certain 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

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

62    Content Views

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, positive 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 mandates involve perceptual experiences, different examples might call this assumption into question. And if there are experienced mandates that extinguish all perceptual 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 special 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 slipping 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 experienced mandates. But there is little reason to think that experienced mandates extinguish perceptual 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 stepping 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 experiments are correct. The usual moral is that our capacity for experience, or our capacity 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 familiar 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

64    Content Views

contents. Against the assumption that perceptual experiences have contents, do any of these contents pertain especially to the felt mandate? This question is sharpened 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 experiences as mandated? Let’s say that a content of an experienced mandate is explanatorily relevant, if it is relevant 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 perceptual experiences might represent, and then consider whether contents involving 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 entering 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 protection 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 mandated 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 standards to decide on or to justify her actions, the expert immediately responds to the current concrete situation. . . . When the Grandmaster is playing lightning 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 dispensable 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 nonrationalizing contents. Turning to contents that involve rationalizing properties, it might seem obvious that these properties play an explanatory role in making the action seem appropriate 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 rationalizing property. Here the forest mandates entering it, at least partly because it looks canopied. Alternatively, in the action-first direction, the component perceptual experience represents a rationalizing property, because you experience the mandate. In this case, the forest would look canopied, at least partly because the experience

18 19

Dreyfus 2005, p. 8 of web version. On attentional suppression, see van Rullen and Koch 2003.

66    Content Views

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 perceptual experience could clearly help explain why the subject performs the mandated 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 experienced 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 contributes to the perceiver’s intellectual coherence and integrity. Compare a case of psychological (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 balcony 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 balcony 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 psychological states into a type of cognitive harmony.21 To the extent that contentful states figure in these processes, they are not explanatorily idle. 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. 20

Affordances and the Contents of Perception    67

Let us turn to the third question about the role of representation in experienced 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 rationalize 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 motivating aspects of experienced mandates. Can we identify a proposition that could reflect both the soliciting and motivating 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 mandate 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 solicitations 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 mandates in accuracy conditions adds the issuing of a mandate by X (cake, forest, hair,

68    Content Views

etc.) to the ‘to-be-phi’d property. The result would be that experience has conjunction 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 motivate 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 reflected 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 streetcarhaving-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 disappeared; 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 directionality and distance have an implicit first-person component, so contents involving 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 mandate 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 soliciting 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-bedone,” 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 psychological validity, it seems reasonable to suppose that it figures in feelings of answerability 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 indifferently. 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 accuracy 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-bephi’d’ contents, it too can be entertained indifferently. The strongest answer to these substantial challenges combines structural considerations with closer attention to what is experienced in the feelings of answerability 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 structure seems to respect the fact that there are both motivated and indifferent ways to 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. 23

70    Content Views

experience soliciting affordances. Leaving a crucial element partly blank for now, the structure could be this: Experience: [It is 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 listening, 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 framework 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 experiences 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 differences 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 representation 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 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. 26

72    Content Views

mechanisms of social interactions including power relationships exerted through discourse, 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 mandates 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 solicited 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 solicitation content, but differ in whether they feel motivated to fulfill the solicitation 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 hypotheses that posit different accuracy conditions, or something other than accuracy 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 proposals. 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 soliciting one to perform an action that will relieve the tension one feels in that situation. 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 understanding, 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 relief 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 attention to an important fact about perception: sometimes our perceptual experiences are pervasively structured by our role as agents responding to social situations.

74    Content Views

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 phenomenologically, perception feels quite different depending on whether it is dominantly 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 performance, 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 perceptual 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 propositional 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 argument 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 perceptual 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

76

Asking whether experience has content immediately raises to two further questions: What are contents? And: What is it for a psychological state to have a content? 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 theoretical 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 perceptual 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 following: 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 commanded 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 expressions (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 abstract 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 properties are what uniquely qualifies the propositions for their role as the objects of

Cf. Davidson 1973, 1975, 1974, 2005; Lewis 1972, 1979. 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. 1

2

78    Content Views

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 psychological states to model a specific kind of structure among them, to locate individual 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 folkpsychological 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 explain why Joe ran across the room in terms of his believing that his stuffed dinosaur 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 relations. 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 question 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 logical 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 psychological 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 psychological 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 content: Do the states in question show the kind of inferential integration into their subject’s system of propositional attitudes that would come with assigning contents 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 attitudinal ‘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?

80    Content Views

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 integral part of our pre-theoretic conception of perceptual experience that experience 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 experience 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 nonstandard 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 consequences can be avoided, or so I have argued, if we construe the contents of experience 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’.

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. 6

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 chapter, 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 propositional 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 experience 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.

82    Content Views

One immediate question for (PB) thus is whether (i)  and (ii) are necessary 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 phenomenal 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 phenomenology, 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 advantages 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 For more on that, see below, section 4.5. 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. Lookscontents 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 12 13

Looks, Reasons, and Experiences    83

relation of reason providing we are interested in is a relation between two propositional 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 proposition: 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, citing 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 perspective 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 belief. 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 providing 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 analogously 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.

84    Content Views

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 reasons 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, however, 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 providing are the following: First, the evidential relation between the contents of experiences 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 experience 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 phenomenal 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 overridden 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

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. 19

86    Content Views

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 experiences and basic perceptual beliefs have identical contents, the relevant inferences ‘stutter’. Defeasibility and modularity go by the board. We avoided this by assigning 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

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, 20

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 implausible 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 concerned 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 inferences 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 exceptions, 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 follows, 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

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. 21

88    Content Views

originally suggested by Pollock (1974). To secure the justificatory power of experiential 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 justification are to be construed as involving prima facie reasons in this sense: justification provided by inductive reasoning, and justification provided by perception. Both suggestions seem plausible to me. Applying the idea to perceptual experience—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 perceptual 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 perceptual 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 sufficient 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 correlation 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 conditions 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 perfectly 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.

90    Content Views

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 perceptual 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, perceptual 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), however, 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 defeasible 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 integration without either jeopardizing modularity or falling prey to the dilemma of stuttering and silence. Silence is avoided by understanding experiential reasons as 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.): 26

(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).

92    Content Views

prima facie reasons in the Pollockian sense. A phenomenal semantics for experience 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 phenomenal 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 operator 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 attitude 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 contents 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 necessary 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 experience 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 available. 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 illustrated 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 subjects). 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

94    Content Views

of what these ways of looking are, an explanation that does not itself make (comparative) 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 comparative 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 natural language. There are some difficult questions regarding our use of ‘looks’, questions 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 phenomenal ‘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 plausible 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.

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. 29

96    Content Views

(7) is clearly well-formed, and that means that there is at least one uniform interpretation of (7). To me, it seems more difficult to read (7) as uniformly comparative 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 experiential 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’ phenomenally thus spells trouble for the idea that ‘looksp’ “indexes” the content of visual experience, he probably does not think that these observations ultimately prevent 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.

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. 32

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 phenomenally 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 equivalence. 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 everything 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 instantiated 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’.

98    Content Views

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) because 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. 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. 37

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 hyperintensional. 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 propositional attitude operator. Nothing we have encountered in the course of these considerations therefore prevents it from ‘going into’ the content of visual experience.

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 38

100    Content Views

In this chapter, I have further developed and defended the phenomenal belief 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 experience into our systems of beliefs and other propositional attitudes. This inferential integration in turn provides us with one of the best motivations for construing experience as having propositional contents in the first place. By offering 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.

102    Content Views 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 perceived 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 representational states postulated by perceptual psychology are not the states or conditions 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 representations 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 terrain, 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

106    Against Strong Content

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 psychology of perception. Only then will it be worth pursuing questions of emergence, 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 reflection 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 waving 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 genuine 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 sensory 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 epistemology, 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 perception 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 deliverances 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 significance 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

108    Against Strong Content

epistemic significance of perceptual experience, the newly “discovered” propositional 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 experiences—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 experience. 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 sensory 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 presented 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 constitutively involves a non-factive propositional attitude rather like the non-factive attitude 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 content 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 contents are “non-conceptual”, that there’s a different attitude for each of the different 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 characterization that is often found in the literature. For instance, in On Clear and Confused Ideas (p. 111), Millikan [2000] introduces “visaging”, “a general 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

110    Against Strong Content

postscript to his paper ‘How to Speak of the Colors’ (pp. 172–173), discusses the view that visual experience involves “a sui generis propositional attitude—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 perceptual 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 scenarios 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 perceptually 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 perceptual 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, structured 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 theoretically fruitful description of the phenomenon of perception, and not as a piece of unarticulated folk psychology, is helpful. For outside of the contemporary philosophical 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

112    Against Strong Content

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 plausibility 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 success 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 conditions 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 determinate conditions. As is now widely recognized, this argument from the merely determinable character of experience is at most an argument for the conclusion that experience is intentional 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 success. 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 different 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 bipolarity 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.

114    Against Strong Content

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 characteristic 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 perceptual 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 misleading 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

116    Against Strong Content

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 perception, 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 produced 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.

118    Against Strong Content

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 illusions 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 already 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 distinctive features, one which potentially bears on the distinctive epistemological 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 perceptual 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 beginning 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 content 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 experience 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 perceptual 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 experiential contact with the relevant items in the scene before him. Fortunately the content view is not, as such, committed to this kind of deflation 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 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. 1

120    Against Strong Content

The resultant state of EX-ing may thus deserve the name of a distinctive form of contact with the worldly items in question; for they enter into the individuation conditions 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 experienced 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 experience 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 distinctively 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 justification. (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 arrangement 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 truthmaker. The predicative structure of your judgment matches the complex exemplification 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 follows. 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 perceptual features, and not simply of predicative relations among items and features. The exemplifications can play an epistemicially significant role that no mere predication 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 theoretically 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 truthmakers, 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.)

122    Against Strong Content

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 hallucinating 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 exploitable 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 perceptual 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 propositional 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 predication. 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 proposition 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 between 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 plausibly 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 perceptual 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 redness” 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, contrary to the suggestion above, the content that the redness of the leather is causing

124    Against Strong Content

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 propositional 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 translation, 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 epistemic significance of perceptual experience; that is, its capacity to disclose truthmakers 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 perceptual 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 correct 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 counterpart 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 psychologists refer to by the determinable “Brain Grey”. This shade is the default presentation 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 exemplifying 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 content 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

126    Against Strong Content

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 truthmakers 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 memory 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 perceptual 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 perceptual 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 mischaracterize 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 perceptual 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 content 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 perceptually 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 luding 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 presentation of the fact that p. Now we can explain the distinctive epistemic significance of “fully successful” 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 reliability 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

128    Against Strong Content

and thereby secures the distinctive epistemic significance of perceptual experience. 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 linguistic 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 understood 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 exemplification 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 psychological 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 objectual perceptual attitudes directed at the exemplifications that are the truthmakers. 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 immediately 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 construed 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 understand 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 presentation mischaracterize their objects; they are veridical if and only if their manners 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

130    Against Strong Content

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 underlying 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 paraphrases 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 remarkable 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 knowledge.) 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 maintaining its utterly determinate character. Yet, the fact that that the calvados is astringent 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 peatiness of the single malt. The friend of facts may see a way forward here, by building in a specific determinateness 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-todegree 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 comparison along which Mary might find one of the respective tastes more remarkable 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,

132    Against Strong Content

without thereby experiencing just what their degrees along that dimension are. But the factive paraphrase, employed to solve the problem of Mary’s comparison, appears to place the degrees themselves in the facts experientially taken in. Thus, reverting to facts taken in rather than individual characteristics presented 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 propositional 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 reference 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 experience 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 environment. 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 subject 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 previous 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 object 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 subject, particular constitution and time of occurrence of token courses of experiences. We can say: A type of course of experience E = E* if and only if (subject, token constitution 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

134    Against Strong Content

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 processing 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 unfamiliar 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 committed 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, experience’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 involved “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 characteristic 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 experience. Moreover, they are theoretical posits, rather than episodes we actually report. Finally and most importantly, unlike the ASEs, the EX-ings do not model experience in such a way as allow for the distinctive epistemic virtue that attends our immediate 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 actually 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

136    Against Strong Content

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 believe 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 illusion 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 epistemically 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 distinctive 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 philosophy, 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 representing autorepresenting. Such will be a side issue here. By contrast, the authoring of a

138

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 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 suggesting 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 parsings). 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, something which is recognizable as what it is—so as occurring, present, or not—independent 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 (producer) 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 ‘allorepresent’ (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 autorepresenting 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.)

140    Against Strong Content

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 relation between one historical circumstance and another. Here one bit of history is what is represented. Another does the representing. That teetering rock represents æons of wind erosion. Pia’s haggard mien represents years of Sid’s grunting. 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 allorepresenting 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 experience 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 inevitably, in doing so thereby autorepresent.

6.1. Thinkers The notion of a thinker at work here is Descartes’. Aiming to distinguish res cogitans 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 infallibly 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 recognize 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 sensitivity 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 recognizable 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 inconceivable. 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 connection 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 continuing 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, allorepresenting is always a project, guided by sensitivity to the world’s bearing on the

142    Against Strong Content

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 perceptions 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 importantly, 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 represented 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 collective, 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 circumstances 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 wallpaper, 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 anything 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 illustration. 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

144    Against Strong Content

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, something, 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 author 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 something 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 anything). 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 different 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 sensory 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.

146    Against Strong Content

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 conceptual 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, correspondingly, 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 yellow. 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 differs 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 success 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 creations 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.

148    Against Strong Content

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 represented 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 particular 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 collectively 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 allorepresent, and how. We represent with an agenda. Such agenda may include contributing 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 (recognizably) 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

150    Against Strong Content

rises from the grill. There is something Tara understood Sid to say. That is a psychological 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 misunderstood. 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) recognizable. 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 appreciative 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 thinkers—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 competent, 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 intrinsic 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 relevant 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 generalizations 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 psychological 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 surroundings, 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

152    Against Strong Content

cloud in which that class of paths, cases of something being yellow are but an element. 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 perhaps 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 unthinkers. 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 incapacity 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 represent 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 content 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 flyers. 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, unless 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 projects 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,

154    Against Strong Content

as Frege puts it, the necessary seriousness were missing, e.g., if he could not properly 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 represented things as being given ways, for example, it would need to be able to represent 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 contribute 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 feature 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 structuring. 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 presence 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, because this presupposes an autonomous existence of concepts, but I think concepts 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 understood 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 isolation 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 represented 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 allorepresenting 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

156    Against Strong Content

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 oursins 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 pathological 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 responsibilities 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 section 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 responsibility 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 certain 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 possible 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 fillingin. 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 representing 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, necessarily, 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

158    Against Strong Content

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 representing 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 represents 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 (section 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 understood 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 contributed 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’ assertive (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 understandings of things being it.

160    Against Strong Content

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 section 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 version 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 seriousness 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 coincidence! 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 representing 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 allorepresenting; 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 allorepresenting 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 universal. 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 modifier) do so. So far, it is the presence of those empty seats which does the representing. 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

162    Against Strong Content

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 allorepresenting 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 allorepresenting 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 ourselves 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 occasion, 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 realizing 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 compositionality. If, say, it were detecting where animate things were, it would need indefinitely many dedicated outcomes for each of indefinitely many arrays of locations 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

164    Against Strong Content

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 ecumenical 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 intuition 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 recognition 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 outcomes 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, yellow 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 environment 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 recognize 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

166    Against Strong Content

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 detection 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 environment 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 application, 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 allorepresenting 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 represent 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 decision 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 yellow 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 represent 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 incorporation 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 detectors (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,

168    Against Strong Content

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 temperature 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 respective 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 creations, 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 allorepresenting 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 pattern, 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 recognizes 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 produced 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 maintaining, 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,

170    Against Strong Content

without supposing it to have occurred, or which might recognizably be present without recognizing it to have occurred. When it comes to reason-giving, allorepresenting 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) allorepresenting 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 allorepresenting 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 representing, 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; according 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 project 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

172    Against Strong Content

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 relevant 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 allorepresenting 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 allorepresenting 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), intrinsically 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 meaningless, 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 thinkers. . . . (cf. 1918, p. 77) Instantiating, thinkers’ doings are what it takes to place thoughts in allorepresenting. Instantiating again, perceptual experience is nothing like being represented to.

6.10. Afterword The question this volume poses is whether perception has content. I have understood content as the content of representing-as (representing something as something), 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 grammatical 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 perception (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

174    Against Strong Content

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 represented). 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 certain 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 yellow). 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 registers, 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 implausible 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 question: 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 doctrine—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 (representational) 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.

176    Against Strong Content

Such different forms of awareness serve different purposes. Perceptual awareness 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 something 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 awareness 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 contents, 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 doctrine 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 rationally 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 representing 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 perception’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 perceptual awareness—other than the presence, or occurrence, of that representing—of what such other awareness positions us to judge knowledgeably. (If in that representing 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 philosophische 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.

178    Against Strong Content 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 hallucinations. 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 qualitative 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 hallucinations are all experiences, but he claims that what unites them as such is just their indiscriminability from veridical perceptions—not their qualitative or intentional 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

180    Against Strong Content

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 epistemological condition. Rather than appealing to a substantive condition which an event must meet to be an experience, and in addition ascribing to us cognitive powers to recognise the presence of this substantive condition, it instead 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 negative 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 indiscriminable 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 conceptual truth that sensory experiences are indiscriminable from veridical perceptions. 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 experiences that they instantiate mental properties that realize, or underlie, indiscriminability 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 combinations 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 disjunctivist’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 occasions 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 experiences 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 nontransitivity 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 perceptions, hallucinations, and illusions—a conception that may satisfy the disjunctivist’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.

182    Against Strong Content

is understandable if one supposes that indiscriminable stimuli invariably appear the same; but in fact there are no such stimuli: not even physically identical stimuli viewed under identical conditions by the same observer always appear the same. (Psychophysicists use the term “false alarm” for instances in which physically 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/different 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 indiscriminability 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 recognitional 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 identify 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 relation. 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 nontransitive 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 indiscriminability series, in which neighboring members do, but the endpoints do not, appear the same in a certain percentage of comparisons. A phenomenal continuum 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 different, 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 phenomenal continuum is a continuous progression in appearance that is instantiated, for an observer at a time, by such a series. The phenomenology of phenomenal continua 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 appear the same (e.g., in 25 percent of same/different trials), there is nothing to prevent 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

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). 7

184    Against Strong Content

instantiating a phenomenal continuum. Any indiscriminability series can instantiate 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 perceptual experience as forming kinds. One might follow Graff ’s suggestion that there is simply no good reason to believe in the existence of phenomenal continua 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 indiscriminability, 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 nontransitive 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 phenomenal 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 locations 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 neighboring 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 subject made a judgment of “same,” a disk of colored light appeared in the center of 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. 9

186    Against Strong Content

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 systematically 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 physically 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 Disk setting in arbitrary units

950

Physically same Physically different

900 850 800 750 700 650 600 550 500 450 400

0

5

10

15

20

25

30

35

40

Stimulus pair (b)

1000 Disk setting in arbitrary units

950

Same Different

900 850 800 750 700 650 600 550 500 450 400 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 comparison 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 subjects 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.

188    Against Strong Content 1000 900 800 700 600 500

0 FIGURE 7.4 

5

10

15 20

25

30 35 40

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 neighboring 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 indiscriminability series can instantiate a phenomenal continuum, such a series is only physically, not phenomenally, stable. A claim of nontransitivity is true of the statistical 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 experimental 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 phenomenal 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 “impersonally” discriminable. On the contrary, all that is required for the existence of a phenomenal 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.

190    Against Strong Content

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 perception of the street as the street.” What exactly is this supposed to mean? The locution “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 constitutive 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 immediate succession. Martin’s and Siegel’s talk of the (in)discriminability of experiences 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 typeidentical 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 experience. He needs a standing, non-occurrent relation that can hold between experiences that don’t actually occur simultaneously and aren’t actually compared. Siegel writes: For all my complaints about the link Martin sees between indiscriminability 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 general 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, hallucinations, 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

192    Against Strong Content

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 veridical 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 subject 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 recognize (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 species 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 categorization, viz., the judgment that it is a hallucination, could be confused with a discrimination or same/different judgment between an occurrent veridical perception and a hallucination. This much cannot be the whole story, of course, for Martin’s thesis is specific:  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 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.) 18

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 introspectively it seems category-identical to a veridical perception of red28? Consider that we already have in hand non-discriminatory identity conditions 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 component 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 identify 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

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. 19

194    Against Strong Content

into the definition of an experience.21 From the disjunctivist’s point of view, replacing 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 estimation: 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.

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. 21

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 fundamentally representational but not fundamentally relational. According to austere relationalists, perception is fundamentally relational but not fundamentally representational.1 Against both, I argue that perceptual relations to the environment and the content of experience should be recognized to be mutually dependent in any explanation of what brings about perceptual awareness of the environment. Another way of expressing the idea is that being acquainted with particulars 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

200    Reconciliatory Views

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 particulars neither ground nor explain perceptual representations. In contrast, to austere representationalists I argue that perceptual representations neither ground not explain 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 representing 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 particular 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 perceptual experience. One choice point is how to understand the relationship between the experiencing subject and the content of her experience. In order to 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. 2

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 account 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 experience 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, consider 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 experience proper and not merely associated with the experience. So the Content Thesis differs from the Association Thesis in kind. After all, if experience is fundamentally 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 content 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 relationship 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 posits 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 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. 3

202    Reconciliatory Views

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 relation the ex-ing relation; Pautz (2010, p. 54) calls it the sensorily entertaining relation; 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 characterizes 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 understood 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 environment without being committed to these more contentious theses. The Content Thesis is neither committed to the thesis that the content of experience is a proposition, 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 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). 6

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 either that content is grounded in sensory character or that sensory character is grounded in content. Or, one might treat content and sensory character as independent 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 nonconceptually structured.8 A third is whether or not the content of experience is propositionally 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 independent 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 between subjectively indistinguishable experiences in distinct environments is a difference in the causal relation between the experiencing subject and her environment. According to austere representationalists, this difference in causal relation 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 representationalism 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.

204    Reconciliatory Views

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 representationalist 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 relational 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 representational 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 representational content (e.g., Travis 2004). The Accuracy Condition Objection: Perception is a relation between a perceiving 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 accurate 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 qualitative 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 perceptual 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 content 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 perceptually 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 understood 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 essentially 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 perceptually related to a mind-independent object that instantiates a perceivable property.

206    Reconciliatory Views

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, austere 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 aspect of perception proper and whether the thesis that experience is representational is fundamental in an account of perceptual experience. Second, austere relationalists 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 personal level. Finally, austere relationalists need not deny that we can articulate propositions to express what we experience. Acknowledging that a subject can articulate such propositions entails no commitment to positing that her experience itself has the content articulated. It might just be that the propositions articulated are merely associated 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 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). 11

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 forward 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 inherently 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 perceives 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 suffering 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 environment 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 perception), 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 different 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.

208    Reconciliatory Views

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 connection 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 commitments 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 perceptually 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 representing. 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 environment 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.

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. 14

210    Reconciliatory Views

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 perceptual capacities, then S has a perceptual experience that is fundamentally 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 sensory 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 understood as conceptual capacities or as nonconceptual capacities. There are powerful 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 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. 15

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 successfully 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 employing 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 perceptual 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 established by the Master Argument. Since S is arbitrarily chosen, Conclusion 3 holds for

212    Reconciliatory Views

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 employing 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 particular 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 perceptual 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, perceptual 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 perceptual capacities employed are the very same in subjectively indistinguishable experiences, such experiences have the same content type. Individuating experiences 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 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). 17

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 attitude 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 experiences 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 perceived object. Olfactory, gustatory, and tactile experiences may not have an attributional 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, gustatory, and tactile experiences among other kinds of experiences. Austere relationalism has it that for perceptual experience to ground perceptual 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 perceptual 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.

214    Reconciliatory Views

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 premises, 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 compatible with just the Association Thesis. Accordingly, Brewer considers a strengthening 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 interpretation. 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 representational 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 fundamentally 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 representational 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 truthmakers. 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 argument that strengthens his case against representationalism. This argument has the same form as Brewer’s argument and preserves the intuitions guiding his argument, but is about mental content rather than sentences: If S sees a mind-independent physical object o, then there are certainly (perhaps 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).

216    Reconciliatory Views

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 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 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 several 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 considers 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 represent when we perceive a cup at an angle is that we represent the shape of the cup in two ways: . 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 (nonconstant) 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 features 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 regardless 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 For a development of the notion of situation-dependent properties, see my (2008). For a critical discussion, see Jagnow (2012) and Madary (2012). 23

The Relational and Representational Character of Perceptual Experience    217

While the single and double primed properties are different kinds of properties, a perceiving subject need not represent these metaphysical facts. The perceiving 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 metaphysical 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 . 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 situational 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 perceptual 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.

218    Reconciliatory Views 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 experience. 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 banana 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 affirmative 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 experiences (namely, those in which one perceives one’s environment as it is) fundamentally 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 issues. 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

220

Thanks to Adam Pautz, as well as audiences at York, Edinburgh, and Nottingham for helpful 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 yellow′). 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 proposition 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 (whatever 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

222    Reconciliatory Views

(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 structure 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 proposition 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 experiential 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 crescentshaped 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 propositional attitude—one could hold that there is a proposition associated with an experience, but deny that the experience consists in the subject perceptually representing it. On this sort of position, although the proposition figures in the characterization 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 experience fundamentally consists in the subject perceptually representing her environment as being a certain way—one could hold that experience is a propositional 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.)

224    Reconciliatory Views

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 experience (e.g., its phenomenal character, or some aspect of its epistemological role) is ultimately grounded in the subject perceptually representing her environment 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 supervene on its content? And does the content of an experience supervene on its phenomenology?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 fundamentally consists in my perceiving the banana. On this view, the ultimate psychological 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 structurally 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 perceptual relation to something. They just give a different account of what that something 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, material 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

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 6

226    Reconciliatory Views

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 conditions. 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 properties, 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 content? 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 experience. 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 experience and a hallucination will render Naïve Realism largely explanatorily redundant (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 content, 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?

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. 10

228    Reconciliatory Views

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 phenomenal 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 experience purple things. Putting the two together, two subjects could have experiences with the content that there’s something yellow before them (because they’re having 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 character 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 phenomenology (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 [environment] 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 epistemological, 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 perceptual 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 experience 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 experience 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 Thanks to Dave Ward for raising this important issue. 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. 12 13

230    Reconciliatory Views

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 argument 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 dialectically 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 argument 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 incompatible 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 sympathetic 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 excellent 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 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. 15

232    Reconciliatory Views

Mild sense). Prima facie, this argument is much more promising than the argument 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 situation from the perspective of a Naïve Realist who takes her view to be incompatible 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 experience 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 environment, 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 propositional 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).

234    Reconciliatory Views

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 yellowness 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 vindicate 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 assessable 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 doxastic 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 appear 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 accuracy 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 yellow crescent-shaped banana naturally gives rise to the belief that there’s a yellow 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 appearance 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 associated 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 perceptually 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 surroundings—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 explanation. 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 crescentshaped 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 having 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 explanation seems to be the least we can get away with. The more promising option is to challenge whether we even need a psychological 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

236    Reconciliatory Views

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 disposition 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 features—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 structure, 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 perceptual 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 experience 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 credibility 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 antecedent 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 proposition 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

238    Reconciliatory Views

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 perception 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 example, 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 experience, 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 veridical 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 epistemological, behavioral, and phenomenological features of an experience. We’re left without an account of what experience fundamentally consists in, if not perceptual 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 environment 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 environment 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 yellow. 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 veridical 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 philosophical 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 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. 22

240    Reconciliatory Views

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 propositional 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, despite compelling arguments in its favor (the arguments from accuracy and appearing). 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 incompatibility 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. 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. 23

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 informational 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 experiencing. 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.

242

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 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 perceptual 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 experiencing, 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 surround 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 experience” 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 organismal 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 consciousness is something more like a window past which aspects of the objective 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 explanation applying to the stream of consciousness is not efficient-causal. Aspects of the stream of consciousness are explicable, if at all, rationalized by other aspects of the stream of consciousness. 2

For more on this, see Hellie 2013.

244    Reconciliatory Views

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 nonpsychological, 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 experience” 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 phenomenological veracity. Namely, attention. Attention is, at the very least, the targeting of aspects of the perceptual surround for inclusion within the stream of consciousness. Recast in this idiom, the nature of the relation between perception 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 surround—cannot fail to be “correct.” And moreover, the content is not the “narrow” sort embraced by Lewis (1994). When an aspect of the perceptual surround is targeted by attention, it is present within the stream of consciousness: the stream of consciousness is characterized by a distinctively 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.

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. 3

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 “frictionless 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 “sensorimotor 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 distinctive: 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, epistemological, 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 threatens 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 following. At each moment of a creature’s life, the creature and its environment

246    Reconciliatory Views

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 knowledge 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 dissociated 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 perception 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 sensorimotor processes should reflect and explain that. Resolving the vexation stemming 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 phenomenological 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.

248    Reconciliatory Views

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 candidate target of Sam’s attention is exhausted by what is a constituent of her sensorimotor 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, assembling 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 phenomenologically. 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 sensorimotor 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 ingredient of the objective world, so that its constituency is an objective matter. Distinguishing the objective phenomenon of the sensorimotor from the nonobjective 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 sensorimotor 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 constituents 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

250    Reconciliatory Views

“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 propositional 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 perception 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 presentation is worth the bother. In a nutshell: without presentation, rational psychology—and probably conscious life—in an objective world would be impossible. In outline, a sort of “transcendental 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 accumulates evidence by gradually making more precise one’s certainties about the objective world as regards the evolving sensorimotor processes of a certain creature and discharges one’s plans by contouring those same sensorimotor process. So unless 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 semantically 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 doctrines 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 rationalizing 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 political 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 sympathize 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 “representationalism” (Harman 1990b; Chalmers 2004). Separated from its focus on exclusively

252    Reconciliatory Views

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 sensorimotor 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 singling 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? creature 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 presentations 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 presentation. 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 consciousness, 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 evidence 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 subject 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

254    Reconciliatory Views

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 dialectical 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 situation 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 alienated 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, presupposition: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.

256    Reconciliatory Views

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 correct view on which color this patch of grass is: the cosmopolitan recognizes 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 falsehoods:  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 semantic 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 undermine 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.

258    Reconciliatory Views

The second leap toward cosmopolitanism may be one we only perform sometimes, 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 parochial biases; in the capital, as easily abandoning them in a cosmopolitan spirit of national compromise. Diachronically inconsistent; less than wholeheartedly sincere 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 idealism 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 circumstance, 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 between 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, representing 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 reasonably asked and universally answered; alternatively, it is the totality of correct propositions. A question that cannot be universally answered is in one respect nonobjective: 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 superficial, 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 peculiarities 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 affirmative). 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 possible worlds. In a context-sensitive propositional semantics, a declarative sentence c ϕ receives a proposition ϕ  as its semantic value against a context c. In a truth-conditional semantics, affirmation is verification:  c verifies ϕ (c |= ϕ) just if c wc ∈ϕ  , for wc ∈W the “world of the context.” In a mindset semantics, affirmac tion is support: c supports ϕ (c  ϕ) just if ic ⊆ ϕ  , for ic ⊆ W the “information 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 charc acteristic rigidifier is A, the familiar “actuality” operator, with  Aϕ  = {w : c ⊧ ϕ}; in mindset semantics, the characteristic rigidifier is ∇, the “Veltman rigidifier” (Veltman c 1996; Yalcin 2007), with ∇ϕ  = {w : c  ϕ}—with an intuitive reading along the 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 interacts with the classical connectives only classically, ∇ exhibits surprising interactions 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 evaluated 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.

260    Reconciliatory Views

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 entertainment 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 conn sciousness, if there are propositions about consciousness, there are 22 proposi2n 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

266    Imagistic and Possible-Word Content

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 imagistic, 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 content 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. Letter to Herodotus, § 51 (= Long and Sedley 15A11). Note: ‘portrait’ = eikon. 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. 1

2

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 characteristics 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 particular 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 specified situation—it has to be applied to, or asserted of, a particular situation or existentially quantified over a range of situations. Perceptual experience applies its

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. 3

268    Imagistic and Possible-Word Content

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 perceptual 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 situation, 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 situations. 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 mistake, 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 clarifying 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 consequently 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”) unproblematically support certain simple sentences, which they called “protocol sentences.” 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 disagreement, let us just stipulate that there is a class of qualities directly given by sense perception. 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 perception—how? According to the logical empiricists, such beliefs are derived from perception 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.

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. 6

270    Imagistic and Possible-Word Content

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 directly from perceptual images to “rich” perceptual beliefs by-passing the intermediary 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 empiricists 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 perceptual 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 individuals. 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. 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. 8

Image Content    271

Finally, I’ll argue that on natural assumptions about sensory images, deflationary 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 episodic memory. In this section and the next, I will lay out some general characteristics 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 sensory “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 perception 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 perceived as properties of objects, and objects must therefore be included in perceptual 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.

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. 10

272    Imagistic and Possible-Word Content

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 . 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 perspective. 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 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 , 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, properly speaking, even though they have some kind of spatiality. Spatial connectedness plays itself out somewhat differently in different modalities.

274    Imagistic and Possible-Word Content

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] correspond 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 represent 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.

276    Imagistic and Possible-Word Content 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 presentation 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 principle. 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 property 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 similarly 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 presenting 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).

278    Imagistic and Possible-Word Content

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 phenomenological 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 movement 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 phenomenology 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 phenomenology 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 experience 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 represented? 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”.

280    Imagistic and Possible-Word Content

in the typical experiences of object-seeing, objects are presented as being denizens 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, moreover, 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, content 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 conclusion 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 represented 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 interesting historical counterparts. Consider Hume’s argument that existence is not imagistically represented: I do not think there are any two distinct impressions, which are inseparably 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.

282    Imagistic and Possible-Word Content

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 impression. (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 represent 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 feature. “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 presupposed 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 perceiver 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 attention 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 disadvantage 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 features 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

284    Imagistic and Possible-Word Content

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 egocentrically 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 orientations 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 counterparts (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 information 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 specification 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 experience 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 stimulation 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 perceptual 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 illusions 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 contemporary 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

286    Imagistic and Possible-Word Content

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 philosophical 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 judgment. 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 judgment, 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 “comparative,”—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 relevant 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 experiences. The reports capture what the subject wishes to convey about the imagistic 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 apparent 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 position 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 content. 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 argument from hallucination, which goes like this: A perceptual experience of (a mind-independent object) O is phenomenally identical to some hallucinatory experience of O.

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. 26

288    Imagistic and Possible-Word Content

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 experience 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.

290    Imagistic and Possible-Word Content Uebel, Thomas E. (1992). Rational reconstruction as elucidation? Carnap in the early protocol 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 experiences, then, have accuracy conditions. Where there are accuracy conditions, there is representational content. Historically, the idea that perceptual experiences have representational content was not the usual one. The most popular view was that content resides further downstream in the beliefs that are formed on the basis of perceptual experience. Perceptual experiences themselves 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 experience 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 perceptual 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

292    Imagistic and Possible-Word Content

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 hallucinogenic drug (see http://stylefrizz.com/200803/keith-richards-for-louis-vuittonad-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 experience 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 supported 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

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. 1

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 experience 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 combination 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 adverbial. 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—content 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 consequences of these reflections for the thesis of representationalism about phenomenal character.

Not everyone accepts this claim. One notable exception is Travis 2004. See also Brewer 2008. 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). 3

4

294    Imagistic and Possible-Word Content

12.1.  The Trouble with Gappy Content Consider the singular content that object, O, is red. On the Russellian view, this content 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

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. 8

What is the Content of a Hallucinatory Experience?    295

surrounded by the relevant piece of cheese. What Casati and Varzi show (at a minimum) 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 hallucinatory 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 spatial 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.

296    Imagistic and Possible-Word Content

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 illumination 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 complicating 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 delivers 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 oversimplified, 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 component 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 singular 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 ordered pair view and all of the obscurity that goes with it.11 Furthermore, if we do make this proposal about gappy content then, to preserve 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 property P, a >>> Modifying the account of gappy content correspondingly, we have: < AND >, < 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

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. 11

298    Imagistic and Possible-Word Content

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 understood 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 straightforward. 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), specifically, 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 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. 12

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 representation 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 correspondingly 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.

300    Imagistic and Possible-Word Content

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 developed 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 appropriate set? Answer: the set of worlds at which the objects picked out by representational 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 hallucination, there are no possible worlds at which the objects picked out by the representational 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:

So,

(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). (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 component 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 experience 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), 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.

302    Imagistic and Possible-Word Content

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 indexicals (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 something 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 sentence 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 entities: 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. 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. 17

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 particular. 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 phenomenally) 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 experience. It is to an examination of this issue that I turn next.

18 19

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

304    Imagistic and Possible-Word Content

12.4. Representationalism Strong representationalism (or intentionalism) is the view that phenomenal character is one and the same as a certain sort of representational content. Weak representationalism, 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 phenomenally 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 representationalism 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 content*. 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 hallucination 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 function 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 experiences, 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

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 20

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 representationalists should endorse: (R) Necessarily, visual experiences that predicatively represent the same property 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 different properties.23 To summarize: on the alternatives presented in this chapter, we have the following packages of views: Package 1:  Each visual experience has a representational content. Such content is unstructured:  a set of possible worlds. Each visual experience also has a phenomenal character. Such character is (or supervenes on) an external property complex. The complex is predicatively represented 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).

306    Imagistic and Possible-Word Content

Package 2: Each visual experience (individuated phenomenally) has a representational content*. Such content* is a function from contexts in which an object is seen to singular contents made up of the seen object and the property complex predicatively represented by the visual experience in that context.25 Within this package of views, as on classical 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 content*. 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 phenomenal 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 property 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 . 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.

308    Imagistic and Possible-Word Content 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 represents 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

312    The Constituents of Perceptual Content and the Role of Perception

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 further 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 properties 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 simply 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 evidence 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

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. 2

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 nonconsciously 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 specifically 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

314    The Constituents of Perceptual Content and the Role of Perception

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 between 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 processes 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 philosophical opponents in the tradition of Kuhn and Hanson (and Ernst Gombrich), most notably Churchland (1988). On the latter view, vision is penetrated by background 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 supposes, 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 penetration 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

316    The Constituents of Perceptual Content and the Role of Perception

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 involving 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 knowledge, 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 particular 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 invalidated 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 earlier 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 borderline 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 distinction 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 chunked, 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 contrasting 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.

318    The Constituents of Perceptual Content and the Role of Perception

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 experience (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 recognitional 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 anything. 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 contrast 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 meaning 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 measurement 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 expressing 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 representation 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).

320    The Constituents of Perceptual Content and the Role of Perception

environment, in order for the representation to have propositional content. 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.

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”: 11

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 external 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 version. 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.

322    The Constituents of Perceptual Content and the Role of Perception

(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 question:  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 qualitative properties given in sensory experience—colors, smells, tastes, textures. A paradigm 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 contents 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 counterexamples to the Representational theory. (Notice I just assumed that vision represents some high-level properties.) I extend the layering view to accommodate color constancy 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 difference 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.”

324    The Constituents of Perceptual Content and the Role of Perception

A Duck/Rabbit Picture

B 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, vision 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

Right

Left

Down

Left

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 unambiguous 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 suggestion 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

326    The Constituents of Perceptual Content and the Role of Perception

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 thoroughly 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 pertinent 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 interpretation. 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.

328    The Constituents of Perceptual Content and the Role of Perception 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 distinguish 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 articulated, 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 content 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 generalizable to all kinds of secondary-quality intentional content. I will also propose a way to draw a general primary/secondary distinction that is plausible, theoretically 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 perceptual color-content within the framework of the GTH account. This approach has

329

330    The Constituents of Perceptual Content and the Role of Perception

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 position, 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 intentionality 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 scenarios. 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 plausible 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 phenomenal 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 contextual 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 phenomenal aspects of conscious mentality are non-intentional, whereas the intentional 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 phenomenally conscious and being “access conscious.”) On an orthodox version of separatism, sensory-perceptual experiences were regarded as lacking representational content (despite having so-called “phenomenal character”), whereas the perceptual 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 phenomenal 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 experience 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 phenomenal 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.

332    The Constituents of Perceptual Content and the Role of Perception

14.1.  The GTH Framework: Phenomenal Intentionality and Externalistic Intentionality Central to the GTH approach is the role of phenomenology or phenomenal consciousness, 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 consciousness 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, apparently 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 propositional attitudes, including conative attitudes like occurrent wishes and cognitive 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 phenomenological 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, intentional 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 particular 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 specific 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., TwinMichele) 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 purport 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 constituted 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 duplicate, 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

334    The Constituents of Perceptual Content and the Role of Perception

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 various 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 acquaintance with such properties—and hence it makes no difference to mental reference to such properties—whether or not the properties with which one becomes experientially 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 referencefixation crucially involve phenomenally constituted grounding presuppositions (as we call them). Thus, phenomenal intentionality is more basic than externalistic 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 constituted 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 particular 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 externalistic: 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 mentally 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 colorproperties 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 presentational 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 phenomenology alone) and externalistic presentational content (which is constituted in part by perceptual-experiential phenomenology and in part by suitable externalistic 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 television, 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 following 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).

336    The Constituents of Perceptual Content and the Role of Perception

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 systematically 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 presumably 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

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. 6

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 theoretical 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 powerfullooking 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 commonsensical beliefs that are widely held and are widely regarded (including by epistemologists) 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-experiences. But this approach is not credible, because claims (1) and (2) express facts about the phenomenology of visual perception that are introspectively self-evident. (Or so I maintain, and will henceforth assume. I realize that many philosophers 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 scientifically 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.

338    The Constituents of Perceptual Content and the Role of Perception

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 everyday color-judgments and color-ascriptions are very often both true and epistemically 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 content 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 colorcontent 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 colorproperties (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 presuppositions 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: 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. 8

Phenomenal Intentionality and Secondary Qualities    339

(i) if presentational color-properties are actually instantiated in one’s ambient 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 factors to play, in the fixation of the reference of the judgment-constituents that purport 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 presentational 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 disposition: a disposition of an object, when perceived under suitably favorable viewing circumstances, to cause a visual-perceptual presentation of itself as-of instantiating 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 conditions, 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 combinations of light-reflectances, or certain physical properties that are the categorical 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 judgments to be epistemically well warranted:  since judgmental color-properties are response-dependent dispositional properties, color-attributing judgments

340    The Constituents of Perceptual Content and the Role of Perception

normally are very well warranted by the visual-perceptual experiences that generate them, viz., visual-perceptual experiences as-of the instantiations of the corresponding, presentational, color-properties. And yet this account also honors the introspectively evident phenomenology of visual-perceptual experience by recognizing 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 presentational color-properties are indeed instantiated on the surfaces of external objects—and there is excellent scientific reason to believe that our world is not actually 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 distinguishing 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 overwhelmingly 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 presentational 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 covariational, 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 presentational 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 framework appeals to the inherent, intrinsic, intentionality of phenomenology:  the phenomenological character of visual experience inherently presents the experiencing 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 reference 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 alternative versions, it might be difficult to make sense of the notion of a presentational color-property. For instance, the version embraced by Uriah Kriegel (2011) treats phenomenal intentionality as (what I  would call) hyper-intrinsic:  this approach repudiates the idea that there can be phenomenologically constituted acquaintance 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 comparison 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

342    The Constituents of Perceptual Content and the Role of Perception

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 monitoring 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 normal 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 constituted grounding presuppositions governing judgmental color content interact constitutively 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 judgment attributes to the object the property being disposed to reliably cause in oneself, in suitably favorable viewing circumstances, experiential visual experiences with presentational color-content as-of presentational red. The invert’s judgment is true. Concerning the meanings of public-language color-words:  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 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 competent 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 neither 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 directional 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-theright (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 relations—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.

344    The Constituents of Perceptual Content and the Role of Perception

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 perceptual/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 instantiated 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 linguistic 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’ directionascribing sentences are true, because these sentences conform with competent practice 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 framework 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 processes occur throughout its life that exactly match the neural processes in the ordinary 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 judgmental 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 redness. 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 disjunctive 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 perceive genuine objects at all in its ambient environment—even though it has systematically 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 environment, 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 visually 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 spectralinversion 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 sensory-perceptual experience, via apparent instantiations of them by apparent objects in one’s apparent ambient environment; (b) this experiential acquaintance constitutes presentational mental reference to them; (c) presentational primary qualities are actually instantiated in the world; and (d) presentational primary qualities are identical to the corresponding properties 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;

346    The Constituents of Perceptual Content and the Role of Perception

(b) this experiential acquaintance constitutes presentational mental reference 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 properties); 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 corresponding 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 collaborative, I have advocated a general conception of truth that I call contextual semantics (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 framework of contextual semantics in the following way. That discussion pertains primarily to default settings of semantic parameters, for both presentational content and judgmental content. Those settings may well work in such a way that the default 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 philosophical 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 sentences 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 common 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 argument that ontological vagueness is impossible. This excludes from the right ontology most of the posited objects and posited properties of common sense

348    The Constituents of Perceptual Content and the Role of Perception

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 concerned, 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 presentational primary qualities, as characterized in section 14.7, includes not just presentational properties like shape, size, relative position, and the like, but numerous 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 standards. When one’s visual experience presents a cup on a table, the presentational content of this experience can indeed be veridical—even though putative properties 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 nonvague 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 cuphood and cathood are real and also are instantiated in the world, and that secondary 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 language 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 correspondence. 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 experience: 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 phenomenology 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.

350    The Constituents of Perceptual Content and the Role of Perception 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, experiences are not produced by external senses. Bodily sensations, possibly some types of emotional experiences, or experiences of the passage of time may involve sensory 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 observation 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

352    The Constituents of Perceptual Content and the Role of Perception

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 experience: 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 extensive 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 investigation 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 different view in this paper.

15.2.  Causes and Objects of Experiences We include under ‘perceptual experiences’ both veridical and non-veridical (illusory or hallucinatory) experiences. All successful perceptual experiences have an object. This is simply another way of saying that in a perceptual experience, something seems to be perceived—what seems to be perceived is the object. Among successful 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 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. 1

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 successful 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 causation are events, then we say that when someone sees a glass, the glass is a constituent 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 experience, 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 seeing 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

354    The Constituents of Perceptual Content and the Role of Perception

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 perceptual 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 investigate 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 disagree 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 question, 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 experiences, 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 experiences. 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.

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). 2

356    The Constituents of Perceptual Content and the Role of Perception

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 question 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 specific to a sense yet not involved in every experience through that sense, and something 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 perceived 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 understood 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 constituent 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 similarly to those of roses. The point is not that the nose can be fooled. Au contraire; 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 olfactory experiences are only odors. Lycan is persuaded here by Ruth Millikan’s teleosemantics-based arguments. On the teleosemantics story, perception indicates for the organism the presence of those things which are useful for its survival, 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.

358    The Constituents of Perceptual Content and the Role of Perception

However, Lycan is reluctant to give up the view that olfactory experiences represent 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 different 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 perception are: those whose presence and properties are responsible specifically for perceptual, 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 sensory 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 appropriate 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 survival than the presence of the sound; this consideration then would rule in favor

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. 7

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 different: 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 discussion in Psychosemantics (Fodor 1987, ch. 4). Fodor’s question is somewhat different 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-inthe-dark. This is clearly unacceptable, says Fodor, and we all agree. Another, similar 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 experience 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 probably 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 phenomenal 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,

360    The Constituents of Perceptual Content and the Role of Perception

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 information about states of the retina, states of the optic nerve; perhaps a visual experience 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 experience is meant to be compatible with pretty much any theory of perceptual experience. 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 uncertainty 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 perceptual 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 phenomenological 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 therefore 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

362    The Constituents of Perceptual Content and the Role of Perception

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 associated 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 significance: 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 experience is an odor, on other occasions, it is (also) the source of the odor. This supports 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 phenomenally 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 association 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 condition 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 properties 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 recognitional 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 illumination 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

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. 9

364    The Constituents of Perceptual Content and the Role of Perception

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 perceptual 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 hollow face illusion; until the difference is demonstrated and shown to have explanatory 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 experience, we need further, and decisive reasons to dismiss the testimony of phenomenal 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 sensations, 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 identified 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 experience, 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 exemplifying 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 phenomenal 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 recognitional 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’.

366    The Constituents of Perceptual Content and the Role of Perception

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 phenomenally 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 experiences. 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 experiences: bodily parts, and objects immediately touching the body. In addition, further 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 membrane. 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 manipulating. 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).

368    The Constituents of Perceptual Content and the Role of Perception

Both Merleau-Ponty’s description and the reports about using sensory substitution 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 exploratory 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 reflection 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

372    Index 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

Index    373 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

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

374    Index 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

Index    375 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

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

376    Index Robust intentional properties, 180 Rock, I., 324 Russell, B., 4, 202–203, 202n6, 253–254, 294 Salva veritate substitutability, 99n38 Sanford, David, 356 Sartre, J. P., 68 Sauret, Wesley, 319n9 Scenario content, 269n6 Schellenberg, Susanna, 12, 16–18, 199, 224n3, 321 Searle, John, 12, 92, 111, 112, 122–124, 280 Second order states, 83–84 Seems-content link, 208, 209n14 Selection process, 148–149 Self-referential causal content, 124 Semantic value, 24–25 Sense-datum view, 4–5, 16, 221, 225, 253, 291–293, 311, 317 Sensible profiles, 121n2 Sensorily entertaining relation, 202 Sensory consciousness, 245–250, 252 Sensory illusion, 265. See also Illusions Sensory imagination, 229 Sensory modalities, 31–32, 45–47, 81, 356–366 Sensory-motor habit, 62 Sensory qualities, 322–323 Sentence-sentence inference, 270 Separatism, 331, 331n4 Seuss, T. G., 57n11 Shared content principle, 280, 283, 288 Shared image content, 280–281, 283–284 Siegel, Susanna, 2, 7–9, 13, 14, 18, 19, 28–29, 51, 110, 180, 181n4, 183, 190, 193, 201n5, 202, 224n3, 230, 233, 233n18, 279–280, 313, 314n6, 315, 316–319, 355 Siewert, Charles, 268n4 Signaling, 360 Similarity coding, 276 Sine wave illusion, 116 Skill-formation, 62 Smaller Center Illusion, 117 Smell.  See Olfactory experiences Soliciting affordances, 54, 54n7, 70, 72 Soliciting experiences, 55 Sosa, Ernie, 21 Spatial connectedness, 273–274 Spatial matrices, 274n15, 282 Spatiotemporal connectedness, 273–275, 295, 334, 356 Specialized skill actions, 60 Spectator-first approach, 51, 51n2 Spectrum inversion, 95n29, 228, 256–257, 341–345 Stefanucci, J. K., 66n20 Stoicism, 265 Stream of consciousness, 243–244

Strong content views, 1–5 Strong externalism, 340–341 Strong modularity thesis, 315 Sturgeon, S., 190n16 Stuttering and silence dilemma, 86 Stuttering inference argument, 80n8 Subjective rationality, 84, 256–257 Subject-raising verbs, 22n5, 23 Sub-personal/personal distinction, 47 Substantive properties, 180 Substitutability, 99n38 Substitution principle, 10 Success conditions, 112 Syntax, 154 Tactile experiences, 274–275, 278, 364–368 Tactile visual substitution system (TVSS), 367, 367n11 Teleofunctionalism, 320n11, 357 Tension, 59, 73 That-clause constructions, 353 Thought (Harman), 111 Tienson, John, 329, 331. See also Graham-Horgan-Tienson (GTH) account Timmons, Mark, 330 Token ASEs, 129 Token content, 212, 212n18 Token experiences, 190–191, 303 Touch sensory experiences, 274–275, 278. See also Tactile experiences Transcendental aesthetic, 273 Transcendental idealism, 258 Transparency, 82n14, 259, 276n21 Travis, Charles, 12–14, 31, 96n32, 132–133, 138, 287n26 Trope nominalism, 121n4 Truth, 77, 346 Truthmakers, 121–122, 238 TVSS (tactile visual substitution system), 367, 367n11 Two-state view, 3 Tye, Michael, 24–26, 291, 323n13 Unconscious inference, 362–364, 363n9 Undercutting defeaters, 88 Unified views, 16–23, 199–240, 250–255 Uniformity, 300 Varzi, Achille, 294–295 Vehicle, 139 Velleman, David, 276 Ventral stream, 5, 42 Ventriloquism, 45 Veridical hallucinations, 118, 123 Veridical illusions, 123 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

Waterfall illusion, 305n23

The Zax (Seuss), 57n11

Xu, Fei, 271n11