Philosophy of Perception: A Contemporary Introduction [Second ed.] 9781138485426, 9781138485433, 9781351049504

Citation preview

Philosophy of Perception

The philosophy of perception investigates the nature of our sensory experiences and their relation to reality. In the second edition of this popular book, William Fish introduces the subject thematically, setting out the major theories of perception together with their motivations and attendant problems. While providing historical background to debates in the field, this comprehensive overview focuses on recent presentations and defenses of the different theories, and looks beyond visual perception to take into account the role of other senses. Key Features and Benefits • • •

•

The only single-authored textbook on philosophy of perception currently available Devoted to contemporary theories and topics, but with appropriate historical coverage for fuller understanding of contemporary work Includes coverage of topics such as: the phenomenal principle; perception and hallucination; perception and content; naïve realism and disjunctivism; intentionalism and representationalism; the nature of content; qualia theories and phenomenal intentionality; perception and empirical science; color and color science; theories of non-visual perception; Molyneux’s problem; cross-modal illusions; multimodality Each chapter includes a chapter overview, questions for further consideration, and an annotated list of Suggested Reading

Key Changes to the Second Edition • • • • • • • •

The division of the book into two major parts: Part I on philosophical theories of perception, Part II on key interdisciplinary topics in perception The addition of two new chapters on color and color vision, and interaction between different sense modalities More topics from the last 25 years of philosophy of perception Combined chapters on belief acquisition theories and intentional theories into one larger chapter More material on the growing intersection of the philosophy and psychology of perception Includes coverage of Molyneux’s problem and of cross-modal illusions Updated chapter summaries, references, and Suggested Reading lists at the end of each chapter A summary table and a more extensive index

William Fish is Professor of Philosophy at Massey University, New Zealand. He is the author of Perception, Hallucination, and Illusion (2009) and the editor of Perception: Critical Concepts in Philosophy (Routledge, 2016).

Routledge Contemporary Introductions to Philosophy Series editor: Paul K. Moser, Loyola University of Chicago

This innovative, well-structured series is for students who have already done an introductory course in philosophy. Each book introduces a core general subject in contemporary philosophy and offers students an accessible but substantial transition from introductory to higher-level college work in that subject. The series is accessible to nonspecialists and each book clearly motivates and expounds the problems and positions introduced. An orientating chapter briefly introduces its topic and reminds readers of any crucial material they need to have retained from a typical introductory course. Considerable attention is given to explaining the central philosophical problems of a subject and the main competing solutions and arguments for those solutions. The primary aim is to educate students in the main problems, positions and arguments of contemporary philosophy rather than to convince students of a single position. Recently Published Volumes: Philosophy of Western Music Andrew Kania Phenomenology Walter Hopp Philosophical Logic John MacFarlane Philosophy of Action Sarah K. Paul Animal Ethics Bob Fischer Philosophy of Time Sean Enda Power Philosophy of Perception Second Edition William Fish For a full list of published Routledge Contemporary Introductions to Philosophy, please visit www.routledge.com/Routledge-Contemporary-Introductions-to-Philosophy/ book-series/SE0111

Philosophy of Perception A Contemporary Introduction SECOND EDITION

William Fish

Second edition published 2021 by Routledge 605 Third Avenue, New York, NY 10158 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Taylor & Francis The right of William Fish to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. First edition published by Routledge 2010 Library of Congress Cataloging-in-Publication Data A catalog record for this title has been requested ISBN: 9781138485426 (hbk) ISBN: 9781138485433 (pbk) ISBN: 9781351049504 (ebk) Typeset in Garamond by codeMantra

For Freya, Anya and Finlay

Contents

Acknowledgments (First Edition) Acknowledgments (Second Edition) 1

xi xiii

Introduction 1 Overview 1 The Two Hats 1 The Interaction of Phenomenology, Epistemology, Metaphysics, and Science: Vision in Modern Philosophy 3 Conclusion 8 Questions 9 Further Reading 9

PART I

Philosophical Theories of Visual Perception 11 2

Sense Datum Theories 13 Overview 13 The Phenomenal Principle 13 Mental Objects and the Common Factor Principle 15 Sense Datum Theory 20 Sense Datum Theory and the Two Hats 23 Sense Datum Theory as a Theory of Seeing 29 Questions 34 Further Reading 34

3

The Representational Principle and Intentional Theories 35 Overview 35 The Representational Principle 36 Representational Variants of Sense Datum Theory 39 Belief Acquisition Theory 43 Intentional Theories 48 Higher-Order Theories and Representationalism 51 Theories of Perceptual Content 53

viii Contents

How Do Experiences Get Their Contents? 61 Representationalism and the Two Hats 64 Questions 68 Further Reading 69 4

Adverbialism and Qualia Theories 71 Overview 71 Adverbialism 72 The Many Property Problem 76 Adverbialism and the Two Hats 80 Color Adverbialism 83 Qualia Theory 84 Qualia Theory and the Two Hats 91 Phenomenal Intentionality 95 Questions 99 Further Reading 99

5

Naïve Realism 101 Overview 101 Naïve Realism 102 Naïve Realism and Disjunctivism 106 Naïve Realism and the Two Hats 111 Naïve Realism and Theories of Hallucination 115 Naïve Realism and Illusion 122 Naïve Realism and the Causal Theory of Perception 125 Is Naïve Realism Refuted by Empirical Science? 128 Questions 133 Further Reading 133 Philosophical Theories of Visual Perception: Summary Table

134

PART II

The Philosophy of Perception and the Sciences of the Mind 135 6

The Philosophy of Perception and Vision Science 137 Overview 137 Cognitive Science and the Information Processing Paradigm 137 J.J. Gibson and Ecological Psychology 142 Perception and Cognition 147 The Admissible Contents of Experience 155 Perceptual Consciousness and Reportability 158 Questions 162 Further Reading 162

Contents ix

7

Color, Color Vision, and Color Science Overview Color Science Philosophical Theories of Color Synesthesia and Color Synesthesia Questions Further Reading

165 165 166 173 184 191 192

8

Perception and the Nonvisual Sense Modalities Overview Individuating the Senses Hearing Touch The Chemical Senses Questions Further Reading

193 193 193 204 209 215 223 224

9

Multimodality Overview Minimal Multimodality and Phenomenal Unity Intermodal Feature Binding and Object Unity Molyneux’s Question Cross-Modal Illusions Multimodality Looking Forward Questions Further Reading

225 225 227 229 232 236 239 243 244 245

References Author Index Subject Index

247 265 269

Acknowledgments (First Edition)

Thanks to everybody I’ve talked philosophy of perception with over the years – there really are too many of you to mention individually, but I’m very grateful to you all. Particular thanks are due to my recent graduate philosophy of perception class, Rhys Burkitt, Malcolm Loudon, Justin Ngai, Louise Nicholls, Hayden Shearman, Jeremy Smith, and Marcel Zentveld-Wale, who, over the course of a semester, worked through a draft of this book with me. I am also very grateful to Ned Block, Alex Byrne, Tim Crane, Dave Chalmers, Kati Farkas, Heather Logue, Stephen Hill, Susanna Schellenberg, and two anonymous readers, who provided me with valuable advice and suggestions at various stages of the project. I am indebted to them all for their input. Thanks also to The Journal of Consciousness Studies for permission to reproduce the synesthesia pop-out figures; Random House, for permission to reproduce the picture of the vase-face illusion; Behavioral and Brain Sciences as well as Alex Byrne and Dave Hilbert for permission to reproduce the cone sensitivity and metamer graphs; and the Hackett Publishing Company for permission to reproduce the opponent processing schematic and graph. Finally, thanks to Beth for all of her support and to my children – Freya, Anya, and Finlay – who make it all worthwhile. This book is dedicated to them.

Acknowledgments (Second Edition)

There has been such a lot of interesting research done on the philosophy of perception in the decade since I wrote the first edition of this book – to everybody who has contributed to this fascinating field, thank you for your work. In addition to those already mentioned in the first edition’s acknowledgements, I’d like to particularly thank three anonymous readers for Routledge who made numerous suggestions of ways in which the first edition could be expanded and improved, as well as Keith Allen, Clare Batty, Robert Briscoe, Mazviita Chirimuuta, Jonathan Cohen, Max Cresswell, Ophelia Deroy, Jakob Hohwy, Uriah Kriegel, Mohan Matthen, Casey O’Callaghan, Adam Pautz, Louise Richardson and Adriane Rini, who were all kind enough to offer comments or suggestions on different aspects of the revisions for this edition. I’d also like to thank Beth Greener for reading a complete draft of the manuscript to try and make sure everything made sense, my daughters – Freya and Anya – for helping me get clear on the physiology, and Finlay, for his help compiling the index. As always, any errors that remain are mine and mine alone.

1

Introduction

Overview In this chapter we begin by introducing two hats that a philosophical theory of perception needs to wear. The first is the epistemological hat, which focuses on perception’s role of providing us with information about the external world; the second is the phenomenological hat, which focuses on the conscious aspects of perceptual experiences. By focusing on these two hats – these two different jobs that a theory of perception must play – we can identify “tests” for an adequate philosophical theory of perception. The remainder of the chapter then considers the way that philosophical thinking about perceptual experience developed from Aristotle through the Modern Philosophers, as the theories that were developed and the assumptions that shaped them also serve to set the scene for the development of the philosophy of perception as an independent sub-discipline in the twentieth century. Over the course of this book, we will be thinking philosophically about our capacity for sense perception – our capacity to perceive the world and our bodies by means of our sense organs.

The Two Hats If one is of a scientific bent, one might wonder just what the role of philosophical theorizing about perception is: Isn’t empirical science in the process of discovering what the nature of a perceptual experience is and what is going on when we perceive? The relationship between the philosophy of perception and the associated sciences of the mind will be discussed in more detail in Part II, but for now, let us simply note that philosophical thinking about perception has a remit that is broader than that of the sciences. In particular, philosophical theorizing about perception is informed in part by considerations from other branches of philosophy.

2 Introduction

So whilst philosophers are indeed concerned with many of the questions that concern empirical investigators – questions of how our capacity to perceive is related to our brains, bodies, and environment, for instance – philosophical theories of perception are also explicitly fashioned to take more philosophical considerations into account. Two considerations that are of particular importance to the philosophy of perception are the following: Epistemology: One of the main reasons for philosophical interest in perception is that it is the primary source of our knowledge of the world in which we live. So any philosophical theory of perception is going to both inform, and be informed by, epistemological considerations. One consideration for a theory of perception, then, will be how well it can make sense of perception’s role as a source of empirical knowledge. Phenomenology: Perceptual experiences are paradigmatically conscious experiences: they have a phenomenology or there is, in Nagel’s influential terminology, something it is like to perceive (1974/1979). And given that there is something it is like to perceive, we can ask what it is like to perceive: what, specifically, it is like to see a pink elephant, to smell coffee, or to touch sandpaper? A further test of a philosophical theory of perception will be how accurately it can capture what it is like to have perceptual experiences. To put it metaphorically, these considerations suggest that an adequate philosophical theory of perception has (at least) two different hats to wear – an epistemological hat and a phenomenological hat. How successfully a particular theory wears each of these hats will therefore be an important choice point when it comes to assessing different theories of perception. Of course, these are not the only important considerations to bear in mind when it comes to evaluating a theory of perception. In addition, there are metaphysical considerations to take into account, such as the fact that any theory of perception will claim that certain things exist, and can hence be assessed in part by querying whether or not these ontological commitments are metaphysically acceptable given the alternatives available. Furthermore, as we shall see, certain philosophical theories of perception incorporate metaphysical commitments about the world itself; if there are reasons to think that these commitments are mistaken, this will constitute a problem with that theory of perception. And finally, whilst it is true that philosophers of perception have a range of concerns that distinguish them from empirical scientists, the conceptual tools they use in philosophical thinking about perceptual experience have often been forged by empirical science. So despite the differences in emphasis, science does exert a significant influence on the philosophy of perception, in part by providing conceptual tools that are then utilized in developing philosophical theories, and in part by discovering various phenomena that philosophical theories of perception are then required to accommodate.

Introduction 3

The Interaction of Phenomenology, Epistemology, Metaphysics, and Science: Vision in Modern Philosophy Before we move on to consider the way the philosophy of perception developed as a distinct sub-discipline in the twentieth century, it would be instructive to sketch how these considerations – phenomenology, epistemology, metaphysics, and the science of the time – worked together to shape discussions of the nature of perception in the Modern philosophers. These discussions exert a significant influence on subsequent theorizing in the philosophy of perception in at least two ways. First, they show a tendency to focus on vision as the paradigmatic case of perception; second, they incorporate a particular way of conceiving of the relationship between perception and the world, and hence frame our conception of the concerns that philosophical theories of perception are expected to address. Prior to the scientific revolution, thinking about perception (like science in general) had been dominated by Aristotelian considerations. For Aristotle, although vision does involve a causal process, it is a process by which the sensible qualities of external objects – such as their colors – are transferred from the object to the perceiver’s mind (1984: Book II). On Aristotle’s hylomorphic view, objects themselves are composites of matter and immaterial form. When we perceive objects, we receive the form of the object without its matter. This process proceeds as follows: air is transparent as it contains a substance which, when fire (or something resembling fire) is present in it, can be “set in movement” – changed or affected – by color. So the first stage of this causal process is that the colored object affects the transparent medium. The now affected transparent medium can then act on the perceiver in turn, which on Aristotle’s view it does by altering – again, changing or affecting – the perceiver’s sense organ. The sense thereby “receives the sensible form of things without the matter” (674), similar to the way in which sealing wax can receive the design of a signet ring without receiving the metal that the ring is made of. In this way, the colors of objects, but not the objects themselves, are transmitted from the object to the medium and from the medium to our eyes, which for animals such as us constitutes our seeing the object. The Aristotelian view began to fall out of favor as the scientific revolution gathered steam during the early Modern period, however. In place of this hylomorphic physics, a more austere mechanistic physics was developed in which matter had only basic properties, such as size and shape, position and motion. On such a view, there is no scope for vision to involve sensible qualities such as color being transmitted to the perceiver, as objects that possess only basic properties do not actually have such qualities to be transmitted. So where Aristotle thought that what is transferred to the mind is the form of what we perceive, on the modern mechanistic view the way in which perceptible objects are perceived is by mechanical contact: by a process of things bumping and pushing, coming into contact with one another. Descartes thus explains the mechanisms of visual perception by analogy to the way a blind man navigates his environment by means of a stick. What matters here,

4 Introduction

Descartes suggests, is simply how one thing pushes against another – objects affect the stick by resisting its movements and the stick, in turn, pushes against the man’s hand: nothing needs to “issue from the bodies and pass along his stick to his hand” (1637/2001: 153). Sight, suggests Descartes, works in a similar way: when light is reflected from an object, this affects the light in a particular way (in Descartes’ view, by giving the particles that constitute it a unique “spin” that correlates with one of the different colors that we experience when we perceive objects). When these particles reach a perceiver’s eye, they push on the retina, and this particular “pushing” excites the nerves of the retina in a distinctive way, which in turn excites the brain, causing the perceiver to have certain sensations. But the motions which are thus excited in the brain by the nerves affect the soul or mind, which is intimately conjoined with the brain, in different ways, according to their own diversity. And the different affections of our mind, or thoughts, immediately following upon these motions, are called perceptions of the senses, or in common speech, sensations. (1644/1985: 280) So sensations do correlate with some feature of the nature of the object – different objects affect the brain in different ways according to their own diversity – but whilst these sensations do enable us to perceive the world, like the sensations the blind man receives from his stick, Descartes cautions us against simply assuming that they resemble the objects with which we began. So in Descartes, we find a philosophical theory of visual perception that is both constrained by features of the science at the time – in particular, the austere metaphysical picture of the nature of matter – yet is also shaped by the conceptual tools it makes available, such as the mechanistic view of causation as involving particles engaged in billiard ball-type collisions or interactions. When viewed through a philosophical lens, however – in particular, when we ask how well the theory wears the epistemological and phenomenological hats – this picture appears less than satisfactory. As far as the phenomenological hat is concerned, to the extent that the theory has any kind of story to tell about what it is like for us to perceive, it is essentially stipulative; according to the theory, it is the sensations that are ultimately caused by the perceptual process that account for what it is like to have conscious perceptual experiences. To this extent, then, the theory could be claimed to capture what it is like for us to perceive – it is a matter of the particular sensations that occur at the end of the causal chain. Yet we might have two concerns with such a view. First is that there is no non-stipulative explanation of why it is like this to have any given experience – this is just accounted for by the nature of the sensations that happen to occur when human minds are affected in a certain way; there is no deeper explanation of why this kind of experience corresponds with these aspects of the world. Second, when we talk about “what it is like” to perceive, it can be natural to think that part of capturing the phenomenology of perception would be to capture the sense that, in vision at least, when we experience our environment, our experiences

Introduction 5

appear to give us some kind of access to the nature of the external world. Yet, as Descartes makes clear, on this picture there is no reason to think that our experiences resemble the external world at all. We might also choose to express this underlying disquiet in epistemological terms: on the face of it, it appears as though one type of knowledge that we acquire from visual experience is knowledge of what the external world is really like. Yet when we look at this theory, we find that it is not clear that it can accommodate this. The story it has about how we get knowledge of the world on the basis of perception turns on the analogy to the blind man’s stick. In this case, we know that the sensations caused by the stick meeting a sidewalk kerb do not in any way resemble that kerb, yet despite this, the man can still acquire knowledge about the location and size of the kerb on their basis, which can enable him to successfully navigate his environment. In this way, the perceptual process thus understood can give us knowledge of the external world. However, this knowledge is not quite what we were looking for from our discussion of the phenomenological hat: one thing that we cannot know from our perceptual experiences, thus understood, is what the world is really like – as Descartes says, we cannot assume that the sensations we enjoy bear any resemblance to the external objects themselves. On this picture, then, our primary mode of conscious awareness in perception is awareness of sensations, which are understood as fundamentally distinct from, albeit caused by, external objects. To the extent that we can be said to be aware of external objects by having perceptual experiences, then our awareness of the world is indirect – we are aware of the world by being aware of the sensations that this world causes. This indirectness can also be framed as a challenge for epistemology. Our primary epistemic access is to the sensations; to the extent that we get knowledge about the world by enjoying these sensations, it is by inference. Just as the blind man infers that there is a curb in front of him by the sensations he receives through the stick (in conjunction with knowledge that these sensations correspond with curbs), we infer that there are objects present in our environment on the basis of the fact that we are having certain sensations, in conjunction with knowledge about the kinds of objects that these sensations correspond with. Locke develops a similar picture, although he more directly addresses the phenomenological question of explaining what it is like to have visual experiences. Where Descartes had sensations, Locke has Ideas, which have usually been understood as mental objects – the things in our minds that we immediately perceive – that are caused by and represent (by resembling, in part) the objects in the world. Again, influenced by the conceptual tools provided by the science of his day, Locke held a broadly “atomic” theory of ideas, according to which experience furnishes the mind with a store of simple atomic ideas, which could then be combined into more complex “molecules” by the activity of the mind. So although Locke’s view also appears to involve an indirect awareness of the world, and thus often meets similar epistemological criticisms, he does have a more straightforward phenomenological story to tell: the things that we are directly aware of – the things we pre-theoretically take to be external objects – are in fact ideas or mental objects. In some dimensions, such as size, shape, motion, and location – Locke’s primary

6 Introduction

qualities – the ideas resemble their causes; both ideas and material objects possess primary qualities. In other cases, however, our ideas fail to resemble anything in the objects at all: where qualities such as color, smell, taste, and so on are concerned, these are qualities that are literally possessed only by ideas. Although material objects are not literally colored, they do nonetheless have associated “secondary” qualities alongside their primary qualities, where their secondary qualities can be understood merely as powers to produce ideas (involving colors, smells, and tastes) in perceivers. So where the phenomenology of experience tells us that we perceive a world populated by objects that possess not only primary qualities but also colors, smells, and tastes, we are partly right and partly wrong. We are right that mindindependent external objects do possess both primary and secondary qualities, but are mistaken if we think that the secondary qualities they possess just are colors, smells, and tastes; in fact, they are merely the powers to produce ideas with those qualities. Locke characterizes the difference between primary and secondary qualities using the notion of resemblance: [T]he ideas of primary qualities of bodies, are resemblances of them, and their patterns do really exist in the bodies themselves; but the ideas produced in us by these secondary qualities have no resemblance of them at all. There is nothing like our ideas existing in the bodies themselves. (Locke 1690/1961: 51) So although mind-independent objects do possess both primary and secondary qualities, according to Locke, we are mistaken if we think the secondary qualities are like the colors, smells, and tastes we encounter in experience. Likewise, we are right that there are objects that we are immediately aware of; it is just that we are wrong in thinking that these are mind-independent external objects: they are in fact ideas. For this reason, this kind of picture is often referred to as a version of indirect realism: “realism”, because it accepts the existence of a realistically understood external world, yet “indirect” because it accepts that our phenomenological and epistemological access to this external world proceeds via more primitive forms of access: in this case, to ideas. However, these experiences can serve to give us knowledge of the external world because the ideas that we are aware of resemble the objects that populate the external world, at least in those qualities that are most important for action, such as size, shape, and location. In both Descartes and Locke then, the science of the time shaped both the austere metaphysics of the external world and the mechanistic, causal understanding of the perceptual process. The knock-on effects of these constraints then led the subsequent theories of perception to have difficulties in successfully wearing the phenomenological and epistemological hats. By framing the issues in this way, we can then see the theoretical developments of later Modern philosophers as their attempts to defuse this tension in different ways. Berkeley, for example, argued that these overall theoretical commitments are actually best coupled with a rejection of metaphysical realism. Berkeley argues first that, if it is true that we perceive only ideas, then from the straightforward-looking

Introduction 7

claim that we do in fact perceive ordinary external objects, we can derive the conclusion that ordinary external objects are in fact ideas. Of course, this argument doesn’t consider the possibility (touched on above) that the verb “perceives” is in fact ambiguous, having both a direct and an indirect sense. When the two occurrences of the verb are clarified in this way – that we directly perceive only ideas but that we thereby indirectly perceive external objects – Berkeley’s conclusion no longer follows. To bolster his argument, Berkeley therefore attacks Locke’s claim that ideas resemble mind-independent objects, at least where the primary qualities are concerned: “an idea can be like nothing but an idea” (1734/1982: 8). So to the extent that the categories we have for thinking about objects, including shape and size as well as color and texture, characterize ideas, we cannot assume that things that are not ideas can be characterized by these qualities. Moreover, when we reflect on these considerations, we can come to question how we could have ever come to the conception of traditional “external” objects in the first place. In this way, Berkeley attempts to undermine the metaphysical realist aspect of the indirect realist position, and in consequence, his approach has arguably an easier time wearing the phenomenological and epistemological hats. If we were simply mistaken about the metaphysical status of ordinary objects – we thought they were mind-independent external objects when in fact they are mind-dependent ideas – then what it is like to have a visual experience will be as it seems: it is like being directly aware of ordinary objects (it is just that they are not mind-independent). Moreover, as our awareness of these objects is not mediated, then our direct experience of ideas constitutes direct awareness of the world of ordinary objects. There are of course a range of objections to Berkeley’s “Idealist” view, but it serves as an interesting example of how the phenomenological, epistemological, metaphysical, and scientific considerations all interact with one another in the construction of a philosophical theory of perception. Hume, in contrast, retains Locke’s commitment to metaphysical realism alongside the broad outline of his theory of ideas. However, finding that Berkeley’s “ingenious” arguments “admit of no answer” (1748/2011: 136, fn.a), he is thus less optimistic about the ability of our awareness of ideas to put us in a position to gain knowledge of an external world. Hume says that “here philosophy finds herself quite embarrassed. … By what argument can it be proved, that the perceptions of the mind must be caused by external objects” (1748/2011: 134)? He notes that appealing to experience will not help in this case, as nothing that is in experience could speak to how those experiences are caused. He therefore concludes that, where this issue is concerned, “the profounder and more philosophical sceptics will always triumph” (1748/2011: 135). Hume therefore reluctantly seems to think that if we want to maintain metaphysical realism, then skepticism is unavoidable. Thomas Reid, however, tries to find a third way between skepticism and idealism. Finding both Locke’s attempt to account for perception of ordinary objects in terms of awareness of ideas and Berkeley’s attempt to identify ordinary objects with ideas unacceptable, he develops a theory on which the perception of ordinary objects is direct. Whilst Reid does accept that most of our perceptions are accompanied by sensations, he agrees with Berkeley that we cannot make any sense of the idea that

8 Introduction

these sensations resemble anything that is non-mental. Instead, he argues that due to the way we are constituted, our having sensations of certain kinds leads us to immediately and automatically conceive of the object that caused the sensations as having certain properties. He explains this using the example of smelling a rose: The object of my perception, in this case, is that quality in the rose which I discern by the sense of smell. Observing that the agreeable sensation is raised when the rose is near, and ceases when it is removed, I am led, by my nature, to conclude some quality to be in the rose, which is the cause of this sensation. This quality in the rose is the object perceived; and that act of my mind by which I have the conviction and belief of this quality, is what in this case I call perception. (1785/2011: 227) So for Reid, perceiving is not a matter of reasoning from sensations to the existence of the external world. Instead, he argues, perception is the act of mind by which we have a “conception, and an immediate natural conviction” (200) of an external object. So while sensations do play a causal role in this process, they do not play an epistemological role, which enables Reid to claim that his theory qualifies as a direct realism as there is nothing (no thing) – no sensations, no ideas – to mediate our perceptual access to the environment. As to how well Reid’s theory wears the phenomenological and epistemological hats, that is of course open to debate.

Conclusion This brief sketch of some of the key theories of visual perception in the Modern era plays a useful role in showing how the range of considerations we identified – phenomenology, epistemology, metaphysics, and the science of the time – interact in the construction of philosophical theories of perception. Following the science of the time, all the Moderns accepted a broadly mechanistic, causal background story for perception – that perceptual experiences, or at least the ideas/sensations we enjoy during perceptual experience, occur at the end of a causal chain that, assuming realism, starts with external objects. Yet each puts the puzzle pieces together differently. Of the four philosophers explored here, Descartes allows the science to drive his theorizing almost exclusively; in some ways, Descartes is more interested in drawing out the philosophical consequences of what the science tells us than developing an independent philosophical theory. As we saw, however, this leaves the subsequent theory struggling both to say anything interesting about the phenomenological aspects of perception and to explain how perception can play the epistemological role we take it to. Against this background, Locke develops a metaphysics of ideas to try to enable the theory to better wear the phenomenological and epistemological hats, but this comes with two downsides: first that it requires us to be radically mistaken about the direct objects of our awareness – we thought they were mindindependent material objects, when in fact they are mind-dependent ideas – and

Introduction 9

second, as Hume reluctantly conceded, this picture makes it difficult to see how perceptual experiences, thus understood, can give us knowledge of the external world. Berkeley then attempts to resolve these epistemological problems using a revisionist Idealist metaphysics, whereas Reid retains the original realist metaphysics but attempts to show how a theory of perception can wear the epistemological hat by engaging in a distinctively philosophical project that distinguishes the mental act of perceiving from the having of sensations, where that is understood as an activity which is involved in perceiving but is not itself constitutive of perceiving. This brief overview provides us with two things. First, it provides us with a sketch of the lay of the land that twentieth-century philosophers were faced with as they began to develop philosophical theories of perception in the early parts of the century. In the first part of this book, we will explore how the principles, preoccupations, and predilections that were inherited from the Moderns, including their tendency to focus on vision, have been taken up, amended, and sometimes discarded as philosophical theories of perception developed from the early twentieth century to the present day. In addition, it shows us how the wider context – particularly concerning dominant approaches in the science of the time – exerted a strong influence on the way in which philosophical theories of perception developed. As we shall see, this influence is as strong today as it was in the Modern period. In the second part of this book, we will therefore explore the contemporary interactions between empirical science and the philosophy of perception.

Questions •

•

•

Descartes draws an analogy between our perceptual situation and a blind man navigating his environment by means of a stick. How plausible do you find this analogy? Locke attempts to ensure that his theory can wear the phenomenological hat by positing ideas to bear the qualities that we are aware of in perception. Does this response successfully accommodate what it’s like to perceive? By denying that sensations have an inferential role to play in the perceptual process, Reid holds that his theory qualifies as a version of direct realism. How well does Reid’s theory wear the epistemological hat?

Further Reading Extracts from Aristotle and the Modern philosophers’ discussions of perception can be found in Volume I of Perception: Critical Concepts in Philosophy (Fish 2016).

PART I Philosophical Theories of Visual Perception

2

Sense Datum Theories

Overview This chapter begins by introducing two key principles by which philosophical theories of visual perception may be distinguished from one another: the Phenomenal Principle and the Common Factor Principle. The Phenomenal Principle states that if I am consciously aware of a property, then a bearer of that property must exist for me to be consciously aware of. The Common Factor Principle states that any indiscriminable perceptual experiences, including veridical perceptions, hallucinations, and illusions, have an underlying mental state in common. These principles, together with some other unexceptional premises, can be used to argue that in all visual experience, whether perceptual or otherwise, we sense nonphysical objects, which nowadays are usually known as sense data. This theory is outlined and difficulties for it are then discussed, including claims that it gets the phenomenology of visual experience wrong, that it cannot deliver a satisfactory epistemology, and that it is metaphysically problematic.

One of the interesting features of the development of philosophical theories of perception in the Modern philosophers was the explicit introduction of mental objects as the immediate objects of perception over and above the sensations that occur when we perceive. Why might we think that this introduction of a new metaphysical category is warranted?

The Phenomenal Principle One significant reason can be found in a well-known passage by H.H. Price, in which he contends that: When I see a tomato there is much that I can doubt. I can doubt whether it is a tomato I am seeing or a cleverly painted piece of wax. I can doubt whether there is any material thing there at all. Perhaps what I took for a tomato was

14  Philosophical Theories

really a reflection; perhaps I am even the victim of a hallucination. One thing however I cannot doubt: that there exists a red patch of a round and somewhat bulgy shape, standing out from a background of other colour patches, and having a certain visual depth, and that this whole field of colour is directly present to my consciousness. (Price 1932: 3) This principle – that if I have an experience as of something red, then there must be a red thing that I experience – was named the Phenomenal Principle by Howard Robinson, who formulates it explicitly as follows: (P) If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that quality. (Robinson 1994: 32) To clarify the terminology present here, “sensibly appears” is used to indicate that we are dealing with conscious awareness. To say that there sensibly appears to me to be something pink is to say that I am consciously aware of pinkness: that pinkness is phenomenally present to me. The Phenomenal Principle then states that in such a case, there must actually be something pink of which I am aware. It is important to note that the Phenomenal Principle has the form of a conditional (an if – then statement) with a phenomenological antecedent and a metaphysical consequent. It tells us that, in order for things to be a certain way for us, phenomenologically, certain things must exist. As these brief discussions indicate, the main motivation for endorsing the Phenomenal Principle derives from our own introspective knowledge of what it is like for us to have conscious experiences. You can test the strength of this motivation for yourself by closing this book and looking at it. The force behind the Phenomenal Principle is simply this: in order for your experience to be the way it is, pinkness and rectangularity have to actually be there for you to be aware of – there must be current instantiations of these properties to adequately explain what it is like for you to have this experience. The Phenomenal Principle codifies this by saying that whenever we have an instance of this kind of conscious awareness, then there must be something – some object – that the subject is aware of and that bears the properties that characterize what it is like for the subject. Are there any other arguments in favor of the Phenomenal Principle? Possibly; it might also be argued to be an implicit commitment of our linguistic practices. Take, for example, the phenomenon of afterimages. If you stare at a bright light for a while, you will usually find that when you close your eyes, you are aware of a bright spot in the center of your visual field that is roughly the same size and shape as the light you were staring at. When having such an experience, you might assent to the truth of the following statement: I am aware of a bright, circular patch. In assenting to the truth of such a statement, you appear to be committing yourself to the existence of a bright, circular patch that you are aware of. This patch would be

Sense Datum Theories  15

the kind of object that the Phenomenal Principle insists must be involved in every visual experience. So the defender of the Phenomenal Principle might also argue that the language we use in talking about our experiences incorporates a tacit commitment to the Phenomenal Principle. Yet the Phenomenal Principle alone cannot explain why we might feel the need to introduce mental objects as the immediate objects of perception. After all, when you look at this book and enjoy a pink, rectangular kind of experience, there is something pink and rectangular that you are aware of – the mind-independent book itself. Why not hold that this is the immediate object of the experience? To take us all the way to mental objects, we need something else.

Mental Objects and the Common Factor Principle To introduce the principle that bridges this gap, consider that different experiences, all of which fall under the general heading of visual experiences, can be more or less correct or successful. Tradition distinguishes three cases: •

•

•

Fully successful cases of perception – cases in which an object is seen and seen correctly or “as it is” – will be termed perception or sometimes veridical perception. When it comes to the associated verb, if we find a subject “seeing” or “perceiving”, it should be understood that we are dealing with a case of successful perception. In contrast, “illusion” refers to cases in which something is seen, but seen incorrectly or “as it is not”. So, for example, illusions includes cases in which a round object is seen to be oval, a blue object is seen to be green, or a tall object is seen to be short. Unfortunately, as there is no aesthetically acceptable verb form, when it is required we will have to talk about subject’s being under an illusion or suffering from an illusion. Finally, the term “hallucination” refers to cases in which it seems to the subject as though something is seen, but where in fact nothing is seen. Classic examples include Macbeth’s hallucination of a dagger and Hamlet’s hallucination of his father. Thankfully, we have an acceptable verb form here: hallucinate.

If we need a term that refers to an experience regardless of which of these three categories it fits into, we will use the term “perceptual (or visual) experience”. So where you find this term, it should be read as a generic term that includes perceptions, illusions, and hallucinations. We may also from time to time distinguish the “good cases”, i.e. the fully successful cases of veridical perception, from a general group of “bad cases” that includes both illusions and hallucinations. Now, when we focus on the bad cases of illusion and hallucination, the Phenomenal Principle can be used as a premise in an argument for the existence of mental objects. Consider hallucination first as it is the simplest case. When a subject hallucinates, there sensibly appears to that subject to be something that possesses at least one sensible quality. Take a hallucination of a pink elephant, for example. In such a case, there sensibly appears to the subject to be something that possesses the quality

16  Philosophical Theories

of pinkness. If we accept the Phenomenal Principle, then this entails that there is something of which the subject is aware that possesses the quality of pinkness. Yet as this is a case of hallucination, there is no suitable pink thing in the world – the pink thing of which the subject is aware is not, therefore, a worldly entity. In the case of illusion, the considerations are similar. Once again, there sensibly appears to the subject of illusion to be something that possesses a sensible quality. If a gray elephant looks pink to the subject, then once again there sensibly appears to the subject to be something that possesses the quality of pinkness. The Phenomenal Principle therefore insists that there is something of which the subject is aware which possesses the quality of pinkness. Yet the only worldly candidate to be the experienced object in this case – the elephant – does not have the property of pinkness, so cannot be the pink thing of which the subject is aware. These arguments give us the first stages of what are known as the argument from hallucination and argument from illusion, respectively. They move from an endorsement of the Phenomenal Principle, together with the plausible claim that hallucinations and illusions are cases in which there sensibly appears to the subject to be something that possesses a sensible quality, to the conclusion that the subject is aware of something that cannot be identified with an everyday worldly object. More formally, the arguments work as follows: Premise 1 (P): If there sensibly appears to a subject to be something which possesses a particular sensible quality, then there is something of which the subject is aware which does possess that quality. Premise 2(h): In hallucination, there sensibly appears to a subject to be something which possesses a particular sensible quality when there is no suitable worldly object at all. Premise 2(i): In illusion, there sensibly appears to a subject to be something which possesses a particular sensible quality when there is no suitable worldly object that possesses that quality. Conclusion (h): In hallucination, the something of which the subject is aware is not an everyday worldly object. Conclusion (i): In illusion, the something of which the subject is aware is not an everyday worldly object. If the subject is not aware of an everyday worldly object, what is the subject aware of? Well, whatever it is, it is pink. And as (we can safely stipulate) there is nothing pink in the subject’s brain, the object of awareness in these cases is typically taken to be nonphysical. Given that hallucinations seem to depend only on processes internal to the subject – processes in the brain, for example – the object of hallucinatory experience is also typically (although not exclusively) taken to be mind-dependent and “private” to its subject, in the sense that only the subject of the hallucination can be aware of that particular object. The standard contemporary answer to the question of what subjects are aware of in cases of illusion and hallucination is therefore this: subjects are aware of nonphysical, mental objects known as sense data (singular: sense datum).

Sense Datum Theories  17

This gives us the positive conclusion of the first stage of the arguments from illusion and hallucination (which, for simplicity’s sake, I will combine): Conclusion (+): In hallucination and illusion, subjects are aware of sense data. Thus far, however, we only have an argument for mental objects/sense data as the immediate objects of perception in the bad cases. We don’t yet have a generalized argument that mental objects are the immediate objects of perception in all cases. To get this, we need to deploy our second principle: the Common Factor Principle. Now, the core of any philosophical theory of perception is an account of the nature of the mental state or event that occurs when we perceive. With this in mind, consider the following three situations: a subject is seeing an elephant that has been painted pink; a subject is under the illusion that the very same elephant, unpainted, is pink (perhaps because of new, experimental lighting at the zoo); and a subject is hallucinating (or maybe dreaming) that they are at the zoo looking at a pink elephant. And assume that the experiences the subject has in these three cases cannot be told apart, that is, that they are indiscriminable or indistinguishable from one another. The Common Factor Principle says that in such indiscriminable cases of perception, illusion, and hallucination, the mental state or event that occurs is the same, regardless of which of these categories the visual experience falls into. In order to be clear, let me say a little more about the idea that perceptions, illusions, and hallucinations have a “mental state or event” in common. The reason we need to say more is because, on one level at least, these three experiences are importantly distinct – they are a perception, illusion, and hallucination in turn! Given this, what does it mean to say that the mental state or event in these three cases is the same? Let me explain by way of an analogy. Consider: two different sorts of burn, exactly alike in the type of physical injury they involve (call it type B), but differing with respect to what causes the injury; there are sunburns, in which B is caused by exposure to the sun, and scorches, in which B is caused by proximity to a source of heat. (Child 1994: 145) In such a case, although sunburns and scorches are different injuries, they nonetheless have a “physical injury” in common – a burn of type B. Likewise, the Common Factor Principle states that, although there is a sense in which perceptions, illusions, and hallucinations are different mental states, they nonetheless have a “mental state or event” in common. The latter claim is to be understood as analogous to the former. I will mark this by saying that, according to the Common Factor Principle, perception, illusion, and hallucination have an “underlying” mental state or event in common. With this behind us, we can formulate the Common Factor Principle as follows: (C) Phenomenologically indiscriminable perceptions, hallucinations, and illusions – perceptions, hallucinations, and illusions of a particular kind – have an underlying mental state in common.

18  Philosophical Theories

This principle can then be appended to the arguments from hallucination and illusion that have been developed thus far to create a general argument for the conclusion that the immediate objects of perception are mental. Recall that the positive conclusion of the first stage of the arguments from hallucination and illusion was that the immediate objects of hallucination and illusion are mental: Conclusion (+): In hallucination and illusion, subjects are aware of sense data. Another way of phrasing this would be to say that in cases of hallucination and illusion, the “underlying mental state” that subjects enjoy is the state of being aware of sense data of certain kinds. So long as we assume that there could be a hallucinatory/illusory counterpart of any possible veridical perception – that is, for any veridical perception, there could be a phenomenologically indiscriminable experience that was not itself a veridical perception – then, given the Common Factor Principle’s claim that phenomenologically indiscriminable perceptions, hallucinations, and illusions have an underlying mental state in common, we can infer that we must be aware of sense data in veridical perception too. So should we accept the Common Factor Principle? Well, there are a number of reasons why this principle might seem appealing. For instance, in the kind of case discussed above, the subject may be completely unable to distinguish between the experience they have when they perceive, when they hallucinate, and when they suffer from an illusion. If we think that our introspective capacities must be able to turn up a difference between two mental states or events if there is a difference there to be turned up, then the fact that we cannot discover a difference between perceptions, illusions, and hallucinations would show that there are no differences between them. Even if we do not hold such a strong view of introspection, the fact that the experiences in these different situations can be indiscriminable could at least be cited as evidence (albeit defeasible evidence) that the subject’s underlying mental state or event must be the same. We might also appeal to ways we find it natural to speak. Above, when I outlined the three pink elephant experiences – one veridical, one illusory, one a hallucination – I talked about them being indistinguishable from one another. However, it would have been very natural to express the commonality between these experiences by saying that the subject’s experience is the same in each case. Alternatively, consider a case in which we are having one of these pink elephant experiences, but do not know which one; that is, we do not know whether we are seeing a pink elephant, under the illusion that a gray elephant is pink, or simply hallucinating. In such a case we might naturally say that we are having the experience of seeming to see a pink elephant, where this is understood as something that could occur regardless of which of the three situations we are actually in. Another argument in favor of the Common Factor Principle, particularly with respect to its role in the arguments from illusion and hallucination, focuses on the implausibility of the alternatives. In considering what would be involved in

Sense Datum Theories  19

nevertheless insisting that, whilst non-veridical experiences involved mental objects, veridical experiences did not, Price contends: Is it not incredible that two entities so similar in all these qualities should really be so utterly different: that the one should be a real constituent of a material object, wholly independent of the observer’s mind and organism, while the other is merely the fleeting product of his cerebral processes? (1932: 31–2) So Price argues that the indiscriminability of perception and cases of hallucination and/or illusion give us reason to think that if one is an awareness of sense data, then so must the other be. Yet in response to this kind of consideration, Austin asks: “If I am told that a lemon is generically different from a piece of soap, do I expect that no piece of soap could look like a lemon? Why should I” (1962: 50)? However, there is a closely related way of developing this kind of consideration that cannot be resisted in this way. This turns on the thought that our experiences of the color of objects can change quite substantially depending upon the particular viewing conditions at the time. As examples of this, A.D. Smith mentions “the common phenomenon of looking at an article of clothing under the artificial lighting of a shop and discovering its ‘real’ colour in daylight” and “the way in which our awareness of the colours of objects changes as dawn gives way to the full light of morning, or as dusk descends” (2002: 27). If we make the plausible assumption that there is a unique “real” color that each object is, then in these transition cases, at least one of our perceptions of these objects would have to count as illusory. When we consider how familiar these cases are, we can see that transitions from experiences that qualify as perceptual to experiences that qualify as illusory are pervasive. Given this, it would seem implausible to hold that they involve distinct kinds of awareness – an awareness of sense data in the illusory cases and some other kind of awareness in the perceptual cases. A second reason connected to and reinforcing these considerations turns on evidence from psychology and neuroscience. We know from studying a range of phenomena that our ability to have veridical perceptions depends upon the right kinds of activity taking place in our brains. We also know that if this brain activity is altered in certain predictable ways, subjects can be made to have illusory experiences. We are also confident that brain activity alone can be sufficient for a subject to have a hallucination. If we accept the first phase of the arguments from illusion/hallucination, we therefore have strong reasons to think that in the case of hallucination, brain activity of the right kind is all that is needed to create a sense datum of which the subject is aware. What is more, the subject being aware of this sense datum suffices for the subject to have an experience that is indiscriminable from a perception. Given this, if the same kind of brain activity were to occur during a perceptual episode, then it should also be sufficient to create a sense datum of which the subject is aware. And, as before, the existence and awareness of this sense datum would suffice for the subject to have an experience that seemed perceptual. So, regardless of the ways in which perceptions, hallucinations, and illusions may differ, there is good reason to think that they share a common element – a common awareness of sense data.

20  Philosophical Theories

The Time-Lag Argument The standard arguments from illusion and hallucination both attempt to prove something of a deceptive case and then extend that conclusion to perceptual cases too. An alternative argument known as the time-lag argument does not involve two stages. Instead, it simply operates with the Phenomenal Principle together with the premise that in time-lag cases, there sensibly appears to a subject to be something which possesses a particular sensible quality when the only suitable worldly object either no longer exists or has changed its qualities. What makes this argument different from the others is that it attempts to include all cases – perceptual and non-perceptual – as time-lag cases. So an extra generalizing stage of the argument is not required. The defense of the claim that all cases of perception are time-lag cases turns on two related considerations. The first is that where distant objects are concerned, light from those objects takes a finite time to reach us, and that the objects may well have changed during that time. For instance, the distance to some stars is so large that by the time the light from those stars reaches our eyes, the stars themselves will have ceased to exist. Yet it is only when the light reaches our eyes that we have a visual experience of the star. So in such an experience, there sensibly appears to a subject to be something star-like when the only suitable worldly object no longer exists. Given the Phenomenal Principle, this entails that there must be something of which the subject is aware which possesses whatever qualities the “star” appears to have. As the star itself no longer exists, this provides another reason to think that the thing of which the subject is aware is a proxy star – a sense datum. Yet even in everyday cases, it takes a finite amount of time for light from an object to reach our eyes and be processed by our visual systems. Although in such cases the time-lag is so miniscule the object will not have had time to cease to exist or change its properties, we can contemplate the possibility that it might have done, and when we do, we see that we would have had the experience that we have regardless. The conclusion the argument aims to establish is therefore that in all cases of visual experience, we are aware of a sense datum.

Sense Datum Theory Phenomenal Principle Common Factor Principle

✔ ✔

As we have seen, the sense datum theory arises from acceptance of both the Phenomenal Principle and the Common Factor Principle. To enable us to see the relationships between different theories of perception with regard to their acceptance

Sense Datum Theories  21

or rejection of these principles, this will be represented as above. Together, these principles suggest that as every experience of certain qualities or properties requires us to be aware of bearers of those qualities or properties, and as we have to give the same analysis of good case experience as we do of bad case experience, then the property-bearers that we are aware of cannot be everyday material objects. It is only a small step from there, via the thought that there are no objects outside the mind that could play the required role, to the view that visual experience must, therefore, involve an awareness of mental objects – sense data. Given that the sense datum theory accepts the Common Factor Principle, the primary formulation of the theory will thus be an analysis of that common factor – that is, of a visual experience where, recall, this is neutral as to whether a case is perceptual, hallucinatory, or illusory. A sense datum analysis of seeing or veridical perception will then have to involve the analysis of the common factor, together with an analysis of what differentiates the veridical cases from the deceptive cases that involve the same common factor. We will explore this in due course; but for now, let’s explore the primary formulation of the sense datum theory as an analysis of visual experience: A subject S has a visual experience as of a property F if and only if: • S senses an F sense datum, D. To clarify this theory, we need to say something about both what sense data are and what the relation of sensing is.

The Metaphysics of Sense Data Sense data are defined as nonphysical objects of awareness, which are logically private to a single subject. In addition to this, the classical picture of sense data also insists that they “actually possess standard sensible qualities, for example, shape, colour, loudness, ‘feel’ of various sorts [but] possess no intrinsic intentionality” (Robinson 1994: 2). We will look at this idea of intentionality in more detail in the next chapter, but the key claim Robinson is making for present purposes is that whilst it might be the case that when we have a given visual experience, we take the sense data to be “of the world” as a matter of habit or custom – that is, we are accustomed to take the sense data to be worldly objects – they are not “of the world” in their intrinsic nature. In their intrinsic nature, sense data possess “only sensible qualities which do not refer beyond themselves” (1994: 2). In the present philosophical environment, the very fact that sense data are held to be nonphysical would be reason enough for them to be viewed with deep suspicion. But not only are they nonphysical, they are also a very strange kind of existent. Barnes (1965: 143–52) asks a number of pertinent questions, including the following: Does a particular sense datum persist while I blink or is it replaced by a new one? Do sense data change in size as I move closer/further away, or are they constantly replaced by new smaller/larger sense data? If one sense datum moves across a visual field, does the whole package of sense data (the visual field) change,

22  Philosophical Theories

or do the different component sense data change their relative positions? If sense data really exist, the thought goes, these questions must have answers. Whilst it is of course open for sense datum theorists to simply stipulate answers to these queries, Barnes worries about the fact that there seems to be no ground for choosing one answer over another. “It may be said that to answer these questions is not important”, he accepts. “I am inclined to agree that it is not; but the only reason I can see for this is that, [sense data] being wholly fictitious entities, we can attribute to them what qualities we please” (1965: 150). The thought here is that the fact that these metaphysical questions do not admit to any principled investigation suggests that the metaphysical realm in question does not really exist. Another metaphysical worry is brought to light in an objection that, according to Chisholm (1942: 368), is originally due to Gilbert Ryle: the “Problem of the Speckled Hen”. When you perceive a speckled hen, you are likely to see that it has a large number of speckles without being able to see precisely how many speckles it has. Whilst the hen, as an existing object, has to have a determinate number of speckles (given the metaphysical principle that to be is to be determinate), our perception of the hen is indeterminate as to its speckledness. The difficulty that this indeterminacy creates is that as sense data are what we are immediately aware of in perception and as our awareness is indeterminate, it seems to imply that the sense data will have to be indeterminate in nature. But this can be argued to contravene the metaphysical principle cited above. In response to this objection, sense datum theorists might stipulate that sensing something does not entail being aware of every feature of that thing (Robinson 1994: 193). Yet this doesn’t quite seem to solve the problem. It is one thing to allow that we might fail to notice some feature of a sense datum but quite another to claim that a sense datum could have features that we are unable to become aware of. Yet it seems that the speckled hen could be just such a case. Even if I try to count the hen’s speckles, I may be unable to. So if the sense datum is determinate as to its number of speckles, this must therefore be a feature of the sense datum that I am unable to become aware of. As Armstrong says, this “has the paradoxical consequence that objects specially postulated to do phenomenological justice to perception are now credited with characteristics that lie quite outside perceptual awareness” (1968: 220–1).

The Relation of Sensing When it comes to the relationship of “sensing”, a natural way to understand the notion is in a pseudo-perceptual way: as kind of like seeing sense data. But if this is how we are to understand the notion then, as Gilbert Ryle points out (1949/1990: 204–5), we would be left with a troublesome regress. If sensing an object, D, was a matter of perceiving D, and D was perceived in virtue of sensing some further object D* (as the sense datum theory of perception suggests), then we have simply ended up where we began. So the relationship of sensing needs to be understood in a non-perceptual way. Having said this, however, further analyses of this relationship have not been forthcoming. The relationship of sensing is left fundamental and unanalyzed. This is not necessarily a problem, though, as Wittgenstein says, all explanations must come to an end somewhere.

Sense Datum Theories  23

This gives us the sense datum take on the ontological structure of perception. What about the phenomenological side? In order to discuss this adequately, let me take a moment to introduce some important theoretical terminology. Philosophical theories of the conscious aspects of mental states and events often utilize the notion of phenomenal character. To provide a theory of the conscious aspects of an experience is then to provide a theory of that experience’s phenomenal character. The problem is that people disagree as to what phenomenal character is. On one influential reading, the defining feature of phenomenal character is that it is a property of an experience; thus Byrne suggests we should just stipulate that “the phenomenal character of an experience e is a property, specifically a property of e: that property that types e according to what it’s like to undergo e” (Byrne 2002: 9). Yet there are theorists who agree that experiences have phenomenal characters, but reject the claim that they are properties of those experiences. This is because those theorists take the defining feature of phenomenal character to be what it is that we are aware of when we perceive/introspect and they argue that there is no reason to suppose that, in such cases, we become aware of properties of experiences (e.g. Tye 2000). For this reason I find it helps to keep things clear if we distinguish between phenomenal character and presentational character. The former, phenomenal character, is understood according to Byrne’s definition – as the property of an experience that types that experience by what it is like to undergo it. The latter, presentational character, is understood as whatever it is that we become aware of when we perceive or introspect. As we shall see as we proceed, this distinction enables us to accommodate anything that anybody might wish to claim by denying that phenomenal characters are properties of experiences by treating whatever they say about phenomenal character as a claim about presentational character and then identifying an appropriate phenomenal character to fit. To see how this works, consider what the sense datum theorist would say we are directly aware of when we perceive. On this theory, we are directly aware of the sensible qualities of sense data. So the presentational character of an experience, according to the sense datum theory, will be constituted by the sensible properties of the sense data that the subject is aware of. Yet although these sensible qualities are properties of sense data, and sense data are constituents of an experience, it would be a fallacy of composition to assume that these properties are thereby properties of the experience itself. So these properties should not be identified with the phenomenal character of a sense datum experience. Yet there is an associated property that such an experience would have that would type it by what it is like to undergo it: it is the property the experience has of being a sensing of sense datum (or collection of sense data) D. The sense datum theorist can therefore identify this property with the phenomenal character of the experience.

Sense Datum Theory and the Two Hats The Phenomenological Hat As we have seen, the sense datum theory is motivated primarily by phenomenological considerations. In particular, it is motivated by the thought that you cannot do justice to what it is like to have an experience unless that experience actually involves an awareness of objects. However, it has been argued that the austere nature of sense

24  Philosophical Theories

data – as objects that only possess sensible qualities – makes it difficult for the theory to adequately capture the phenomenology of “real” experiences. This is Strawson: Suppose a non-philosophical observer gazing idly through a window. To him we address the request, ‘Give us a description of your current visual experience’ […] Uncautioned as to exactly what we want, he might reply in such terms as these: ‘I see the red light of the setting sun filtering through the black and thickly clustered branches of the elms; I see the dappled deer grazing in groups on the vivid green grass …’ and so on. (1979: 43) However, if the considerations presented in the arguments for sense data are accepted, it is possible that the subject might have an experience of this very kind even if there were no elms, no deer, no grass, etc. So we ask our observer to explain what his experience is like without committing himself to the existence of the (purportedly) perceived objects. If the sense datum theory is correct, then this ought to be possible by describing the sensible properties possessed by the immediate objects of his experience. Yet Strawson claims that a careful subject “does not start talking about lights and colours, patches and patterns. For he sees that to do so would be to falsify the character of the experience he actually enjoyed” (1979: 43). Although the subject could attempt to exhaustively describe this in terms of patches of color, this would, according to Strawson, falsify the character of my experience – it would fail to adequately capture what it is like to have that experience. A more specific phenomenological problem area for sense datum theory has concerned our experience of depth. The premise for this objection is that depth is a phenomenological feature of our visual experiences – that the experience of depth is, strictly speaking, a purely perceptual phenomenon. If this premise is accepted, then it seems that the sense datum theorist has three choices: either to claim that depth is a further sui generis sensible property that sense data can possess (a response Robinson (1994: 206) describes as “ad hoc and bogus”); to accept that the sense datum theory fails; or to accept that sense data are in fact three-dimensional and are literally at a distance from the perceiver. This option is taken up by sense datum theorists such as Jackson (1977: 102). Yet even given the strangeness of this claim, it is unclear that it can resolve the problem. Depth perception, as it is normally understood, is a feature of binocular vision. So even if sense data are in three-dimensions, there are still questions about how depth is sensed: is sensing binocular? Other sense datum theorists (such as Robinson: 206–7) therefore prefer to reject the premise, endorsing Berkeley’s famous claim that: distance of itself … cannot be seen. For distance being a line directed end-wise to the eye, it projects only one point in the fund of the eye. Which point remains invariably the same, whether the distance be longer or shorter. (1709/1910: 13) This response treats the experience of depth as not, strictly speaking, part of the phenomenal nature of the visual experience at all, but rather as a result of the cognitive states that naturally accompany the experience.

Sense Datum Theories  25

The Epistemological Hat As we might expect from the fact that the sense datum theory is motivated primarily by phenomenological considerations, it has been criticized for being unable to adequately wear the epistemological hat. Typically, the sense datum theorist will claim that we perceive external objects “in virtue of” the sensing of sense data. A realist version of sense datum theory, which endorses the mind-independence of the material world, is therefore a version of Indirect Realism. As we saw in the previous chapter, the suggestion that it is “indirect” indicates that everyday material objects are held to be perceived indirectly, in virtue of an awareness of sense data, which are the direct or immediate objects of experience. This picture has led to the objection that in claiming that all we are ever directly aware of are sense data and only through them aware of the external world, we “raise a veil” that separates us from the external world. Jonathan Bennett therefore calls this the “veil-of-perception doctrine” (1971: 69).

Mental Objects and the Concept of an External World Ryle suggests the problems run even deeper than this. He wonders how, given the commitments of the sense datum view, we could get to a point where we had the concept of a mind-independent world at all. He illustrates this concern with the following analogy: There is immured in a windowless cell a prisoner, who has lived there in solitary confinement since birth. All that comes to him from the outside wall is flickers of light thrown upon his cell-walls and tappings heard through the stones; yet from these observed flashes and tappings he becomes, or seems to become, apprised of unobserved football-matches, flower-gardens, and eclipses of the sun. How then does he learn the ciphers in which his signals are arranged, or even find out that there are such things as ciphers? How can he interpret the messages which he somehow deciphers, given that the vocabularies of those messages are the vocabularies of football and astronomy and not those of flickers and tappings? (1949/1990: 212)

We need to be careful here. Some of the appeal of this objection is sensationalist, playing on the image of some kind of cognitive “ray” being aimed at the world by our minds but being thwarted by sense data that “get in the way”. But most realist sense datum theorists would not deny that external objects can be objects of

26  Philosophical Theories

perceptual awareness, they would just hold that this awareness is mediated by sense data. As Dancy puts it, sense datum theory “does not have the consequence that external objects are unobservable; it purports simply to tell us something about what it is to observe them” (1985: 165). To see how this might work, consider the relationship between seeing objects and seeing their surfaces. Jackson (1977) argues that we do not directly “see” material objects, but merely their facing surfaces. We then see the objects in virtue of seeing their surfaces. On the face of it, this is a reasonable claim. But if we then view the relationship between sense data and objects on this model, it is no longer obvious that it involves a veil of perception. Nobody thinks that objects’ surfaces “hide” the objects themselves from us – rather, they “facilitate” or “mediate” this awareness. This is exactly the way that sense datum theorists view the relationship between our awareness of sense data and our awareness of (the facing surfaces of) objects. Yet even if the sense datum theorist can adequately rebut the objection that the theory precludes the awareness of external objects, a key epistemological question remains: How does perception, so understood, enable us to acquire knowledge of the external world? Traditionally, this kind of theory of perception has been associated with foundationalist theories of epistemology. Foundationalism begins from the observation that many of our empirical beliefs are justified by the relationships they stand in to other beliefs. So, for instance, my belief that is justified by my beliefs that and . However, my belief that London is more populous is only conditionally justified; it is only justified if my beliefs about the relevant populations are justified. So what justifies me in these beliefs? Well, the thought goes, other beliefs, such as beliefs about the results of recent censuses and so on. However, this suggests the threat of a worrisome regress. If every belief is only justified if another belief is justified, and if that belief is only justified if a yet further belief is justified, then it looks as though we will never be in a position to say that any of our beliefs actually is (unconditionally) justified. To avoid this, the Foundationalist claims that all of our empirical beliefs ultimately depend for their justification on certain foundational beliefs – beliefs that are justified in a special unconditional way. What are these foundational beliefs? Well, as Lewis put it, “empiricists generally are agreed that nonperceptual synthetic knowledge rests finally on knowledge which is perceptual, and so find the root problem in the nature of perception” (1952: 170). So the foundationalist’s foundations are our beliefs about our own sensory experience. Why these beliefs? Because: When I perceive a door, I may be deceived by a cleverly painted pattern on the wall, but the presentation which greets my eye is an indubitable fact of my experience… The given element is this incorrigible presentational element; the criticizable and dubitable element is the element of interpretation. (ibid)

Sense Datum Theories  27

The similarity between this claim and Price’s phenomenological defense of the Phenomenal Principle is striking. According to Lewis, what marks out beliefs about our own experiences is that, whilst there is much that can be doubted, there is a core belief which is indubitable or incorrigible. This explains why these beliefs are justified and hence apt to serve as our epistemological foundations. According to traditional foundationalist versions of sense datum theory, our beliefs about the external world are then justified by beliefs about our own experiences.

Incorrigible, Indubitable, Infallible To say that something is indubitable is to say that it cannot be doubted. To say that something is incorrigible is to say that it cannot be corrected. However, I’m not sure that either of these two notions are quite what the Foundationalist is looking for. To see why, note that there are two reasons why it may be impossible to doubt something or impossible for something to be corrected. In the first case, either because it could be false, but we’re not cognitively equipped to doubt it, or because it simply couldn’t be false (it is infallible). In the second case, either because it could be false, but you are in a far better position than me to know whether it is true or false (so although you might get it wrong, you’re guaranteed to be in a better position than me), or because it simply couldn’t be false (it is infallible). For the purposes of epistemology, it looks to be the second sense of indubitability/incorrigibility that is relevant, which suggests that the really important notion is that infallibility – of being unable to be false.

Over the years, many philosophers have argued that such a claim renders sense datum theory fundamentally unable to deliver a satisfactory epistemology. The worry is highlighted by the very arguments that aim to show that sense data are necessary. As the second stage of the arguments from illusion and hallucination makes clear, the sense data that we are aware of in a case of perception are exactly the same as those we would have been aware of in an indiscriminable case of illusion or hallucination. Given this, we can see that the sense datum theory is committed to the idea that an experience of a particular kind could occur in the absence of appropriate objects. Yet this has the consequence that we cannot move deductively from one to the other. Given the possibility of hallucinations, our having an experience of a certain kind does not entail anything about the world whatsoever. Induction is likewise problematic. An inductive justification would claim that we are justified in assuming that experiences of certain kinds are reliable indicators of the world being a certain way because we have previously established the regularity of these two events coming together. Yet in the present case, this is a nonstarter as we have

28  Philosophical Theories

no independent access to the world on which to establish that a certain kind of experience is correlated with the world being a certain way. At this point, purist sense datum theorists may turn to abduction; to suggesting that our belief that sense data correlate with the external world is justified as it is the best available explanation of certain evidence. In this context, the “evidence” is that experience is highly organized and predictable, both across senses and across time. The “best available explanation” of this evidence is then held to be that the experiences are reliably correlated with a highly organized and predictable external reality. To see how this might work, imagine a scientist observing (what we now know to be) alpha-particle tracks in a cloud chamber. At first, let us suppose the scientist just sees the tracks. Yet after careful experimentation, they realize that the presence of tracks is reliably correlated with the presence of certain materials in the chamber. The scientist therefore hypothesizes that these materials all emit a certain particle, which they call an alpha-particle. After a period of using the hypothesis, they become so familiar with the hypothesis that they start to think and talk directly about alpha-particles, bypassing the tracks in both thought and language. Indeed, the scientist may even get to a point where they no longer really “see” the tracks, in that they no longer perceptually attend to the tracks at all. Instead, they attend directly to the alpha-particles (we might say, they “see” the alpha-particles) in virtue of their more basic kind of awareness of the tracks. This is, I take it, pretty much how the abductive approach sees the role of sense data in visual experience. However, some theorists, including some sense datum theorists, have worried that treating our belief in the external world as a “theory” formulated to explain the “evidence” is implausible. For example, Price contends that “we do not invent [the theory that there is an external world]. We have already on other grounds formed a conception of the physical world” (1932: 89). Likewise, Armstrong says that “surely we are not prepared to degrade bodies into hypotheses? We want to say that our assurance of the existence of the physical world is far stronger than any assurance we could obtain by indirectly confirming a theory” (1968: 30). Uriah Kriegel puts the point as follows: [I]n casting knowledge of ordinary physical objects as based on something like inference to the best explanation, [the sense datum theory] offers the wrong model of such knowledge. On a more plausible model, we know that the table is brown by seeing that the table is brown and endorsing what we see. We do not know that the table is brown by first being perceptually aware of objects other than the table and then inferring that there must be a brown table if we are to be perceptually aware of those other objects. (2011b: 247) Moreover, Price goes on to suggest that it is not the “best explanation” anyway; he claims there are a “thousand and one” other hypotheses such as the hypothesis that God causes all of our experiences, most of which are “much simpler” (1932: 89).

Sense Datum Theories  29

Sense Data and Idealism As we saw when we discussed Berkeley in the opening chapter, if you (a) take the arguments in favor of sense data to be strong, whilst (b) feel the force of the epistemological considerations just listed, yet (c) want to allow for us to have knowledge of the external world, what should you do? Some theorists in this position have followed Berkeley and taken it to show that we should endorse idealism: the metaphysical thesis that the external world is not mind-independent after all, but mind-dependent. This enables “the physical world to come within the reach of direct perceptual awareness by taking it to be something which is logically created by facts about human sensory experience” (Foster 2000: 1). As Idealism is not a theory of perception, however, but rather a metaphysical theory about the nature of reality, I shall not discuss it further here.

Sense Datum Theory as a Theory of Seeing As we saw, the fact that the sense datum theory accepts the Common Factor Principle means that the primary formulation of the theory is an analysis of the visual experience, which is the common factor involved in all cases: perceptual, hallucinatory, or illusory. This means that the theory we have been investigating thus far does not tell us what is required for the subject to have a veridical experience, or more specifically that it does not explain the circumstances in which the episode of sensing an F sense datum constitutes a successful perceiving of that property. To get a sense datum analysis of seeing, then we need to ask what differentiates the veridical cases that involve a certain visual experience from the deceptive cases that involve the same common factor. In order to develop such a theory, we need a way of identifying, out of all of the occasions on which a subject has a visual experience as of a property F, which of those occasions are ones in which the subject actually sees property F. This is similar to the discussion from within epistemology about what distinguishes mere belief from knowledge proper. A component of an influential response to this question has been that at least part of what distinguishes knowledge from mere belief is that, to know something, your belief has to be true. On such a view, knowledge shares something with failed attempts to know – believing – but is a case of believing in which other conditions are met. We can imagine, the thought goes, that a belief might fortuitously happen to be true in situations where we would not want to say the subject has knowledge. Such analyses therefore hold that other conditions, such as the belief being justified, also need to be met. So cases of knowledge are a subset of cases of belief – the cases of belief in which the belief is also true (and justified, and …). If we try and adopt this tactic in the case of perception, we would treat veridical perception – seeing – as visual-experience-plus. That is to say, a seeing of property F would be a case in which subjects have a visual experience as of an F and some other conditions are met. The question then becomes, what is the nature of these

30  Philosophical Theories

“other conditions”? One suggestion is that S’s visual experience as of an F needs to be satisfied, where – at least for the sense datum theory – a visual experience will be satisfied so long as the external object or scene “matches” the subject’s experience, where “matching” is a matter of resemblance between sense data and scene. The analysis of seeing would therefore be: A subject S sees an F object, O, if and only if: • S has a visual experience as of an F object (= S senses an F sense datum, D), and • this visual experience is satisfied (= D resembles F object, O). An initial concern with the appeal to satisfaction concerns precisely what is required for an experience to be satisfied. For instance, consider a visual experience as of a distant airplane, and suppose my having that visual experience involves my sensing a speck-like sense datum. Does this match or resemble the world? After all, a speck does not resemble an airplane. Yet intuitively we would want to count this as a case of successfully seeing an airplane inasmuch as this is what airplanes look like from a distance. So there is work to do to give an account of what kinds of relationships between experience and object/scene qualify as “matching”. Nevertheless, even if this difficulty can be overcome, this analysis will not yet work. An appeal to satisfaction alone will not distinguish between those visual experiences that qualify as veridical and those that do not. Consider Grice’s example of the clock on the shelf. In this example, a subject has a visual experience as of a clock on a shelf, so the first condition of our analysis of seeing is met. What is more, there is a suitable clock on a suitable shelf in front of the subject, so the second condition is met. As both conditions are met, our current analysis would count this as a case in which S sees the clock on the shelf. However, as Grice goes on to say: it is logically conceivable that there should be some method by which an expert could make it look to X as if there were a clock on the shelf on occasions where the shelf was empty: there might be some apparatus by which X’s cortex could be suitably stimulated, or some technique analogous to post-hypnotic suggestion. If such treatment were applied to X on an occasion when there actually was a clock on the shelf, and if X’s impressions were found to continue unchanged when the clock was removed or its position altered, then I think we should be inclined to say that X did not see the clock that was before his eyes. (1961: 142) Grice goes on to say that: I think we should be inclined to say that X did not see the clock that was before his eyes just because we should regard the clock as playing no part in the origination of his impression. … There seems then a good case for allowing that [a causal] condition is necessary. (1961: 142)

Sense Datum Theories  31

In other words, Grice suggests that it is a necessary condition that for S to see the clock, the clock must be the cause of S’s experience. Yet as Grice notes, causation alone will also not provide a sufficient condition for seeing to take place. For one thing, there is the problem of picking out which component of the long causal chain is the component which is seen; an experience is causally dependent on numerous states of affairs – from states of our brains and eyes down to states of the electricity plant that generates the  electricity that lights up the room – none of which are seen. For another, there is an  issue with nonstandard experiences. For example, we might think that my hallucination was caused by my brain, but we do not want to say that in having a hallucination, I see my brain. To resolve this, a causal condition can be added to the satisfaction condition to get the following analysis of object perception (a comparable analysis of property perception is a little more complex): Subject S sees object O if and only if: • S has a visual experience, E, as of O, and • E is satisfied, and • E was caused by O. Yet this analysis has also been criticized. Consider the following counterexamples (Lewis 1980): The Brain Before the Eyes: I hallucinate at random (thus the experience is caused by my brain) and seem to see a brain before my eyes which perfectly matches (purely accidentally) my own brain. But my brain is before my eyes—it has been removed from my skull and all nerves (etc.) have been stretched somehow. In this case, the first condition of the analysis is met—I have a visual experience as of a brain before my eyes. So is the second condition: this experience is satisfied as there is a brain before my eyes. And so is the final condition: my having this experience was caused by the brain before my eyes. As all three conditions are met, our analysis would count this as a case in which I see the brain before my eyes. But that doesn’t seem right. The Light Meter: I am blind, but electrodes have been implanted in my brain in such a way that when a light meter mounted on my head receives light over a certain threshold level, they will cause me to have an experience of a certain sort of landscape. By chance, just such a landscape is before my eyes, and its illumination is enough to turn on the electrodes. Once again, the first condition of the analysis is met—I have a visual experience as of a certain kind of landscape. So is the second condition: this experience is satisfied as there is just such a landscape in front of me. And so is the final condition: my having this experience was caused by the landscape. As all three conditions are met, our analysis would count this as a case in which I see the landscape. But again, that doesn’t seem quite right. Lewis describes these scenarios as cases of veridical hallucination; they are cases that we intuitively want to describe as hallucinatory, but where the hallucination

32  Philosophical Theories

just happens to both be caused by and to match the world. Yet while we want to classify these cases as hallucinations, our analysis as it stands counts these situations as cases in which something is seen. This suggests that we need to augment our conditions to rule out such cases as counting as cases of seeing. Grice suggests that maybe what we need to do is to restrict the modes of causal mediation to the standard ones. Even if we can’t spell out what they are (yet), we can nevertheless leave “a blank space to be filled in by the specialist” (1961). With this modification made, our analysis of seeing would hold that: Subject S sees object O if and only if: • S has a visual experience, E, as of O, and • E is satisfied, and • E was caused by O in the normal visual way. The problem with this proposal, Lewis argues, is that it is too strong, and rules out possible cases of nonstandard seeing. For example, if it turned out that some people had visual systems which work on different principles to the rest of us, then their causal relation would be mediated in something other than “the normal visual way”, and hence they would not be said to see. Alternatively, and perhaps more plausibly, it would also appear to rule out the very possibility of prosthetic seeing. Suppose, for example, that scientists developed a prosthetic eye which, when used to replace a damaged eye, would function just like a normal eye. Although this operation would seem to be a way of restoring a subject’s sight, on this view, as the causal relation in such subjects would not be the normal visual causal relation, this operation would leave subjects unable to see. In both of these cases, Lewis suggests, we should not rule out the possibility that these subjects might really be said to see. So the proposed modification should be rejected as it is too strong. Lewis instead suggests that what is required is that the visual experience we have is counterfactually dependent upon the scene before the eyes. The underlying thought here is that if you really see an object/scene, then any changes in that object/scene would be reflected by changes in the visual experience. With this modification made, our analysis of seeing would hold that subject S sees object O if and only if: • • • •

S has a visual experience, E, as of O, and E is satisfied, and E was caused by O, and the phenomenal character of E is counterfactually dependent upon O. This would rule out the original counterexamples as follows:

•

•

The Brain before the Eyes: As I am hallucinating, I would have had an experience as of a brain before my eyes even if the brain had not been there. So the counterfactual dependence of experience on scene does not hold. The Light Meter: Again, it doesn’t matter what the scene in front of the subject is, so long as the illumination was adequate. You could change various aspects of the scene (so long as the overall illumination remained the same) without

Sense Datum Theories  33

thereby changing the nature of the visual experience. So the counterfactual dependence of experience on scene does not hold.

Lewis on Causation Strictly speaking, for Lewis, the last two clauses of this analysis are not entirely distinct. This is because Lewis endorses a counterfactual theory of causation, according to which, to say that A caused B is to say no more than that the existence of B is counterfactually dependent on the existence of A – that had A not existed, then B would not have existed (see his paper “Causation” in his (1986) Philosophical Papers volume II).

Yet the addition of the counterfactual dependence clause would allow for nonstandard/prosthetic seeing, because in each of these cases if these alternative systems really worked, then if you changed the scene, the experience would change, and hence we have a counterfactual dependence of experience on scene. However, there is a further counterexample from Michael Tye (1982) which involves two robots, Tom and Tim. Although I cannot see it, I am standing facing a mirror angled so I see an area to my right. Behind the mirror, and therefore hidden from my view, stands robot Tim. Away to my right, hidden behind a wall, stands Tim’s left-right inverted robot twin, Tom, who is facing the mirror. Robot Tom is therefore reflected in the mirror, such that the image I see is of a robot which looks just like Tim would if the mirror were to be removed. Now, Tom is wired up so that all his movements are caused by Tim’s movements – the only reason he is standing where he is, is because Tim is standing where he is. And any movements Tim makes are copied (but left-right inverted) – if Tim waves his left hand, then Tom waves his right hand, and the mirror image of Tom “waves its left hand”.

MIRROR

Figure 2.1

34  Philosophical Theories

If I stand at the appropriate place in this setup and look toward the mirror, I will have a visual experience as of a robot in front of me, so the first condition of Lewis’s revised analysis will be met. As is the second condition: my experience is satisfied as there is just such a robot in front of me – robot Tim. As is the third condition: my having this very experience is caused by the robot that is in front of me (Tim), as it is Tim’s being where he is that causes Tom to be where he is. And Lewis’s final condition is also met. The phenomenological features of my experience are counterfactually dependent upon Tim – if he moves, Tom will move and my experience will change accordingly. As all four conditions are met, Lewis’s revised analysis would count this as a case in which I see robot Tim. Yet although all the conditions of this analysis of seeing are met, our intuitions remain that it is Tom, not Tim, which we see. We will pick this discussion up again as we explore other theories of perception; but for now, note that any theory that endorses the Common Factor Principle and proposes to provide a primary analysis of visual experience, understood as a general category that includes all successful and unsuccessful cases, faces the requirement to provide an additional analysis of what then distinguishes the successful cases from the unsuccessful ones. As we have seen, this is not obviously straightforward and, in particular, merely appealing to a causal connection between the visual experience and the object or scene it is an experience of faces some thorny problems.

Questions • • •

How compelling are the arguments in favor of sense data? Where are their weaknesses? How compelling are the metaphysical objections to mental objects? Should they make us avoid such objects in our theories of perception? What makes it the case that, on a given occasion, a visual experience of a certain kind qualifies as a case of veridical perception?

Further Reading Important classic readings on sense datum theory, including papers or extracts from Russell, Moore, Price, Ayer, Austin, Barnes, Jackson and Grice, can all be found in Volume I of Perception: Critical Concepts in Philosophy (Fish 2016). These readings also contain Chapter 8 from Howard Robinson’s (1994) Perception, which remains the classic contemporary defense of sense data. Although Robinson eventually (and tentatively) favors an anti-realist version of the theory, the majority of the book is an argument for the philosophy of perception claim that perception involves sense data and takes no stand on this metaphysical question.

3

The Representational Principle and Intentional Theories

Overview In this chapter we introduce a third key principle known as the Representational Principle. We then explore classic theories that endorse this principle – one as a development of the sense datum theory and one as an alternative to it – before moving on to consider a range of contemporary theories that treat visual experiences as a kind of propositional attitude in which subjects take a distinctively perceptual attitude toward an intentional content. We discuss a variety of different kinds of intentionalist theory, including representationalism, which claims that the phenomenology of a visual experience is determined by its representational content. We then go on to discuss a number of important issues for intentionalists, focusing in particular on the nature of perceptual content.

In the previous chapter, we saw that one of the most significant concerns with the sense datum theory was that it struggled to explain how we can move from experiences to beliefs about the world, given that on the sense datum theory, empirical knowledge would have to be the result of some kind of inference from experiences to world. Yet as we saw, an adequate account of how these inferences get started is not easy to come by. Indeed, the best account of this seems to involve treating our conviction that there is a stable, mind-independent external world as some kind of deeply held hypothesis – deeply held because it best explains why we have the sense experiences that we do. In addition, we saw that it is questionable whether treating the world-involving aspects of experience as purely cognitive effects of visual experiences rather than perceptual phenomena themselves provides a phenomenologically apt picture of what actually happens when we have visual experiences. As Price says, “we simply jump straight from the awareness of A to the thought of B, without any preliminary wondering or considering of evidence, indeed without any rational process whatever” (1932: 140–1).

36  Philosophical Theories

In this light, one off-shoot from the sense datum theory also endorsed a further principle – the Representational Principle – which states that all visual experiences are intentional or representational.

The Representational Principle When theorists claim that a visual experience is intentional, they are attempting to draw our attention to a particular feature of visual experiences: that an experience is about something in the world, something other than itself or “beyond itself”. This raises the question of how visual experiences come to have this intriguing property. A common contemporary understanding of how visual experiences have the property of being “about” something is to see them as representing that the world is a certain way.

Brentano on Intentionality This is not to say that one couldn’t make a case for a different interpretation of intentionality. The terminology itself derives from Brentano (1874), who characterizes intentionality in three ways: as involving “the intentional (or mental) inexistence of an object”, “reference to a content”, and “direction toward an object” (88). One could, therefore, contend that a theory of perception counts perceptual states as intentional so long as it can count them as being somehow directed toward an object. However, the contemporary use of intentionality is most closely connected with the “reference to a content” characterization, and hence to the notion that visual experiences contrive to be about things in virtue of being representational. We shall therefore restrict our understanding of intentionality in this way. To see how these concepts go together, consider a map of London and a postcard of St Paul’s Cathedral. Both the map and the postcard tell us about St. Paul’s. In virtue of this, both objects can be said to represent St. Paul’s or simply to be representations. However, as representations of St Paul’s, there are two key differences between the map and the postcard. •

•

On the one hand, there is a difference in how they represent the cathedral. The map represents it symbolically and linguistically, with a cross symbol next to the words, “St. Paul’s”, whereas the postcard represents it pictorially, with a photographic image of the building. Partly due to these differences in how postcards and maps represent, there is also a difference in what these two objects tell us about the cathedral. The map tells us where the cathedral is in the overall geography of London – it’s on the North bank of the Thames, just opposite the Millennium Bridge – but not

The Representational Principle  37

much at all about how the cathedral looks. The postcard, on the other hand, tells us much more about the look of the cathedral – it tells us about the shapes and colors of its signature dome and clock towers as well as other things about its surroundings – but gives us less information about its location relative to other London landmarks. By focusing on what these different representations tell us about St Paul’s, we can introduce an important distinction between representational vehicles and representational content. The vehicle of representation is the thing that is doing the telling – in the cases just discussed, the picture on the postcard and the icons on the map are the vehicles as they are doing the telling. The content of the representation is what the representation is saying – where the map says that St. Paul’s is on the North bank of the Thames, the postcard tells us that it has a large domed vault. Here are some examples of philosophers making the claim that visual experiences as well as maps and pictures are representational/have contents: A perceptual experience represents the world as being a certain way. (Peacocke 1992: 61) [P]erceptual states represent to the subject how her environment and body are. The content of visual experiences is how the world is represented to be. (Martin 1994: 464) What does it mean to say that properties are represented in visual experience? The notion of representation is tied to the idea that experiences have contents … I call this thesis the Content View. (Siegel 2010: 4) As can be seen in the quote from Martin, when we talk about the content of a visual experience, we are talking about what the experience tells us about the world/ the information it conveys to us. Because a representation tells us something in this way, it is possible for it to mislead us: for it to tell us something that is incorrect. Just as a misprinted map might wrongly inform us that St Paul’s is on the South bank of the Thames, a visual experience might mislead us as to the state of the world. In virtue of having a content – in virtue of telling us something about the world – a visual experience can therefore misinform us: it can tell us that things are a certain way when they are not. In this way, both maps and experiences (as well as pictures, stories, etc.) are the kinds of things that can be accurate (or inaccurate). This is a somewhat broader notion than truth, both because accuracy can come in degrees – a picture can be more or less accurate, but a statement cannot be more or less true – and because (partly for this reason) it seems awkward to describe things like pictures as true (Crane 2009: 458). Accuracy is similar to truth, however, inasmuch as when we determine whether or not something that can be accurate is accurate, we have to

38  Philosophical Theories

evaluate it in relation to something else – whatever it is that it is purportedly accurate of. In the case of the picture of St Paul’s, this would be the cathedral itself – if the picture is accurate, it matches the cathedral it is (or at least purports to be) a picture of. In the case of experiences, when we ask whether an experience is accurate or veridical, we assess it against a situation: “in the simplest case [we are] assessing them with respect to the situation in which they are had” (Siegel 2010: 31). If we specify how the situation would have to be for the experience to be accurate, we are simultaneously specifying the experience’s accuracy conditions (similar notions include correctness conditions (Crane 2009: 457)) and “conditions of satisfaction” (Chalmers 2004: 341)) and how it represents the world to be. This specification of an experience’s accuracy conditions also gives us the representational content of the experience. This means that it is important to distinguish this relatively technical use of the word “content” from its more everyday cousin. At present, for example, the contents of my pocket are my wallet, keys, a cellphone, and lint. But the fact that my pocket has contents in this sense doesn’t mean that the contents of my pocket – my wallet, say – has accuracy conditions or is potentially accurate or inaccurate. So we need to be aware that there is a fairly innocent spatial understanding of the term “content” and that sometimes when people talk about the “content of perception”, they may use the term in this way to refer simply to what is in our experience/what is perceived. When it is used in this way, it is used as a way of referring to the object of perception – to what is perceived; it is not (necessarily) an endorsement of the Representational Principle. To enable us to refer back to it, let us specify a formal statement of the Representational Principle as follows: (R) All visual experiences are representational. As we have already noted, our key principles are motivated, in part, by our firstperson understanding of what is involved in being a perceiver. So what aspects of that understanding motivate the Representational Principle? First, when we consider what it is like for us to enjoy visual experiences, it seems clear that perception is world-directed in some important way. In particular, perception plays a critical role in enabling us to find out about and navigate our environment. How could it do this if it were not the case that it carried information about that environment? Endorsing the Representational Principle can be seen as an attempt to capture this key feature of perception. Second, there is our talk about experiences. We often speak of people seeing that the sky is blue, the sea is green, and so on. This might be taken to indicate that our everyday talk includes a tacit commitment to visual experiences having world-involving contents as the Representational Principle suggests. Third, there is the observation that much psychology also treats visual experiences as representational. So to claim otherwise – to deny the Representational Principle – might be seen to be, in some sense, anti-science (this will be discussed in more detail in Chapter 5).

The Representational Principle  39

Representation and Cognitive Psychology It is no accident that representationalist thinking in the philosophy of perception began to become prominent in the second half of the twentieth century, at around the same time as the cognitive revolution within psychology, which also has a central place for notions such as representation, was taking shape (e.g. Chomsky 1959; Neisser 1967). And although philosophical representationalism is defended in part by appeal to important philosophical considerations, such as the need to explain the conscious character of perceiving and how experiences justify beliefs, there are deep connections between cognitive psychology and philosophers’ endorsement of the Representational Principle. For example, the development by empirical scientists of theories of vision that tried to explain how the brain computes a representation of its environment (such as Marr 1982) not only influenced, but was also often cited in support of philosophical theories that deployed notions such as representation and computation.

Representational Variants of Sense Datum Theory Phenomenal Principle Common Factor Principle Representational Principle

✔ ✔ ✔

Given the problems that the pure version of the sense datum theory had with explaining how visual experiences are world-involving, it may be appealing to endorse the Representational Principle in addition to the Phenomenal Principle and the Common Factor Principle. The austere sense datum view is that, although we naturally take our experiences to be of a world, this is a purely cognitive rather than perceptual matter – a matter of the way we interpret those experiences rather than a feature of the experiences themselves. By also endorsing the Representational Principle, the thought would be, the visual experiences would then tell us about the world as part of their very nature. In this case, we would not need to develop and endorse a theory about the world – a visual experience would, in a sense, wear the world on its sleeve. This was developed in two ways in the early twentieth century. The first of these – the sensory core theory – augmented classical sense datum theory with the Representational Principle by taking visual experiences to be comprised of both a phenomenal component, involving the direct awareness of sense data, and an additional representational component, which tells us about the world. This view has echoes of that of Thomas Reid, which we met briefly in the opening

40  Philosophical Theories

chapter. Price is a notable advocate of this kind of theory. In discussing the case of someone hallucinating a pink rat, he says that the correct analysis of the situation is that the man is both “acquainted with a pink sense datum” and “takes for granted the existence of a rat” (1932: 147). These two states of mind “arise together … all in one moment. The two modes of ‘presence to the mind’, utterly different though they are, can only be distinguished by subsequent analysis” (1932: 141). On a two-component sense datum theory, normal visual experience involves a subject both (i) sensing or being acquainted with sense data and (ii) “taking for granted” that an appropriate material object exists. To give a full account of visual experience, then the notion of something’s being “taken for granted” must be further explained. In discussing this, Price makes it clear that: what is taken for granted is … that so and so is the case—that a material thing exists here and now, that it has a surface of such and such a sort, that it is grass, etc.—in short, what is taken for granted is a set of propositions. (1932: 166) The second component, then – the non-sensory component – is a representational component. It involves the subject’s representing the world to be a certain way – the way specified by the relevant “set of propositions” – and taking it for granted that the world is as represented. Like the sense datum theory proper, however, the sensory core theory will also hold that the presentational character of the experience will be given by the sensible properties of D and the phenomenal character of the experience will thereby be its property of being a sensing of D. This has been argued to give the sensory core an advantage over the pure sense datum theory when it comes to the phenomenological hat. We noted that the classical sense datum theory has been criticized over whether its focus on “lights and colours, patches and patterns” can adequately capture the rich phenomenology of visual experiences. The sensory core approach has an answer to this objection: to allow that the cognitive component of a visual experience can affect what the subject is visually aware of. Consider a familiar example: a subject perceiving a circular coin that is tilted on its side. Although there is a sense in which the coin presents an elliptical face to the perceiver – its silhouette would be elliptical, for example – subjects may not notice this, taking the coin to be circular. The classical sense datum theorist explains this by holding that, although the subject of this experience is only strictly visually aware of ellipticality (in virtue of sensing an elliptical sense datum), she does not notice this ellipticality due to her familiarity with such cases, meaning that on the basis of this experience, she interprets the coin to be circular. According to the sensory core theorist, however, the subject is consciously aware of ellipticality in one sense (by sensing an elliptical sense datum) whilst also being consciously aware of circularity in another (because the subject represents the coin as being circular and takes it for granted that this representation is correct).

The Representational Principle  41

Perception and Cognition This approach appeals to the important distinction between perception and cognition. This distinction is present in psychology, and tells us that our awareness of certain properties, such as color, is strictly experiential/phenomenal, whereas our awareness of other properties, such as depth perhaps, occurs in a different, cognitive, way. Interestingly, although this distinction has a long-standing philosophical pedigree (it has its roots in Aristotle), it is often elided by our everyday talk about perception. This raises an interesting question: of those properties that we say can be seen, which actually do belong to perception proper (are bedrock phenomenal), and which are rather a feature of our cognitive response to our experiences? This is a question that we shall return to in Chapter 6. This modification of the classical sense datum theory would therefore hold that classical sense datum theorists, as understood in the previous chapter, were half right. They were right, the thought goes, in seeing that an awareness of mental objects is required to accommodate the Phenomenal Principle. They were also right in treating the sense data themselves as not being representational. Where they went wrong, however, was in thinking that full-fledged visual experience was nothing more than an act of sensing sense data of various kinds. According to this line of thought, where the sense datum theorists took themselves to have given a full account of visual experience, they had in fact given only an account of the “sensory core” of visual experience. However, as sensory core theorists allow that we are aware of both the properties that the sense datum possesses (in a sensory way) and the properties that are represented (in a more cognitive way), they face a dilemma. Either they allow that the representational component is somehow based (in an epistemically relevant way) upon the sensory component or hold that it is distinct. If they take the former option and hold that the representational component is based upon the sensory component, then this seems to fail to yield a significant epistemological advance over the pure version of sense datum theory. If they take the latter option, however, then they need to either allow that the interpretative component arises simultaneously but independently of the phenomenal component or that this component is caused by/automatically prompted by the phenomenal component in some non-epistemically relevant way. Whichever way they go, this not only leaves the sensory core view owing some kind of story of how and why the sensing and the taking for granted are always (or at least usually) “appropriate” to one another, it also seems to conflict with the intuition that when we form a belief about the world on the basis of perception, we do so because of the way things seem (phenomenologically) to us. The second way in which the Representational Principle has been added to the core commitments of the sense datum theory was developed by Roderick Firth (1949/1965),

42  Philosophical Theories

in what he called the percept theory. What distinguishes the percept theory from the sensory core theory is the denial of the idea that our perceptual awareness of “sensible qualities” such as color and shape proceeds by way of sensing sense data, and that our perceptual awareness of other qualities such as depth (perhaps), clumsiness, reptilianness, and felineness proceeds by way of a distinct process of interpretation. According to Firth, we are conscious of all these different qualities in the same way. If Firth is right, then two possibilities present themselves: either we are aware of all these qualities in the kind of way that the classical sense datum theorist takes us to be aware of shapes and colors, or we are aware of these qualities in some other way. The first option would require that we are aware of properties such as clumsiness, reptilianness, and felineness by being aware of objects that actually possess those properties. Yet as Robinson points out, mental objects cannot literally be clumsy, reptilian, or feline (1994: 29). This leaves us with the second option: if we can be aware of these properties by being aware of sense data, and if our awareness of all qualities proceeds in the same way, then this is not by way of being aware of something that actually possesses those qualities. Instead, Firth’s view seems to be that while the mental object does not actually have these properties, it represents itself as having them. So sense data are not really clumsy, or reptilian, or feline, they just represent themselves as being clumsy, reptilian, or feline. As mental objects also represent themselves as being mind-independent, Firth calls them “ostensible physical objects”. In sum, the percept theory claims that there are mental objects in visual experience, which, unlike sense data classically conceived, have both sensible and representational properties. In normal experience, however, we are not aware of their intrinsic sensible properties at all. All our attention is directed at the world, by way of being aware of what these mental objects represent. When it comes to the explanation of the tilted coin, the percept theorist would hold that, although the subject is aware of an elliptical sense datum, they are in fact consciously aware only of circularity. This is because, despite actually being elliptical, the mental object represents itself as circular and it is the way it represents itself as being rather than the way it is that determines the subject’s phenomenology. The subject can, however, become aware of the ellipticality of the mental object by engaging in a process of perceptual reduction. The concern with this approach is that, when we look closely, we see that the percept theory actually embodies a rejection of the Phenomenal Principle. That principle, recall, says that: (P) If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that quality. As we have seen, the percept theorist claims that there are more sensible qualities – qualities that things can sensibly appear to possess – than the sense datum theorist can allow. So the percept theorist will certainly accept that there are many cases in which the antecedent of the Phenomenal Principle is met without the consequent thereby being true. To see this, take a particular case in which there

The Representational Principle  43

sensibly appears to a subject to be something which possesses the (now) sensible quality of clumsiness. The percept theorist does not agree that, in such a case, there is something of which the subject is aware which actually possesses that property. Rather, the claim is that there is something of which the subject is aware which represents itself as possessing that property. So the percept theory does not endorse the Phenomenal Principle, even though it retains the commitment to mental objects. But if all the heavy lifting is being done by the Representational Principle and not the Phenomenal Principle, then it might seem preferable to try and make this work without the extra ontological commitment to mental objects.

Belief Acquisition Theory Phenomenal Principle Common Factor Principle Representational Principle

✘ ✔ ✔

Above, we looked at Price’s sensory core version of the sense datum theory, on which visual experience not only involves a subject sensing sense data, but also a subject’s taking for granted that such-and-such is the case. We might wonder, therefore, whether we could manage with this “taking for granted” alone, and thereby not appeal to the problematic mental objects. This is, broadly speaking, the approach to visual experience taken by David Armstrong, George Pitcher, and the belief acquisition theorists. As Armstrong is keen to point out, such a view would mesh well with evolutionary thinking about the function of perception. When it comes to visual experience, what is important about perception, from an evolutionary standpoint, is the nature of the information about the world it provides us with. As Armstrong puts it: It is clear that the biological function of perception is to give the organism information about the current state of its own body and its physical environment, information that will assist the organism in the conduct of life. (1968: 209) In other words, the very reason we have perceptual systems is to enable us to gain knowledge about the environment in which we have to live and prosper (ignoring, for the present, perception of the organism’s bodily states). According to a highly influential philosophical theory of knowledge, at least part of what it is to know that x is to believe that x and for x to be the case (i.e. to be true). This is a most important clue to the nature of perception. It leads us to the view that perception is nothing but the acquiring of true or false beliefs concerning the current state of the organism’s body and environment. … Veridical perception is the acquiring of true beliefs, sensory illusion the acquiring of false beliefs. (1968: 209)

44  Philosophical Theories

As is indicated by this quote, the belief acquisition theorist accepts the Common Factor Principle (in earlier work, Armstrong calls the parallel between veridical and non-veridical cases “the parallel on which the Argument from Illusion rightly insists” (1961: 83)). Given this, the belief acquisition theory will hold that phenomenologically indiscriminable visual experiences have an underlying mental state in common. An interesting question is whether or not the belief acquisition theorist accepts or rejects the Representational Principle. In one sense, the event of perceiving is the event of acquiring beliefs, and that event does not possess intentionality. However, the things that are acquired – the beliefs themselves – do possess intentionality. So there is a very clear sense in which perception can be said to be intentional on a belief acquisition theory. It is just that, as Armstrong puts it, “the intentionality of perception reduces to the intentionality of the beliefs acquired” (1968: 211). What about the Phenomenal Principle? That principle states that: (P) If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that quality. If we accept that the antecedent of the Phenomenal Principle can be met in both good and bad cases – that is, accept that in cases of illusion and hallucination there sensibly appears to be something which possesses a particular sensible quality – then the claim that perception is “nothing but” the acquisition of beliefs seems to entail a rejection of the Phenomenal Principle. Hallucinations would be cases in which there sensibly appears to be something that possesses a particular sensible quality when in fact there is not. In such cases, the “non-physical object of immediate apprehension is simply a ghost generated by my belief that I am seeing something” (1961: 84). Given Armstrong’s claim that perception and illusion (and by extension, hallucination) involve the acquisition of beliefs, together with the claim that the same mental state occurs in indiscriminable cases of perception and hallucination, we can provide a first-pass analysis of the belief acquisition theory’s claim about the neutral category of visual experience as follows: A subject S has a visual experience as of a property F if and only if: • S acquires the belief that something is F. In addressing this theory, it is important to be clear that the key theoretical claim is not that perception is believing, but rather that it is the acquiring of belief. Normally, to say that a subject has a certain belief doesn’t entail that there is any occurrent mental event – anything going on in the subject’s mind at that time. We continue to believe certain things (e.g. that Paris is in France) even when we are not actually thinking that thought, for example, when thinking about

The Representational Principle  45

something else or when asleep. But perceiving something is an occurrent mental phenomenon. So it is important to be aware that this analysis does not hold that perception is believing (which may be a nonconscious, dispositional state), but that perception is the acquiring of beliefs. The acquiring of a particular belief is something which happens at a particular time and hence is plausibly an occurrent mental phenomenon.

Belief Acquisition and Unchanging Scenes Pitcher (1971) raises a concern with the term “acquisition”, inasmuch as it implies obtaining something which one did not previously have. As we can continue to perceive an unchanging scene (I see the cup on the table at t1, I still see it at t2, etc.), it would seem strange to say that we acquire the belief that the cup is on the table at t2 as we already believe that from perceiving it there at t1. To avoid this implication, Pitcher prefers to use the locution “causally-receives” as we can keep causally receiving a belief even if we already have it. With this point made, however, I will stick with “acquire”. In developing his theory of perception, Armstrong is particularly keen to enable his theory to adequately wear the epistemological hat. In particular, in doing away with mental objects as direct objects of perception, the theory can claim to have drawn aside the veil of perception that was claimed to separate us from the world. Moreover, as the beliefs we acquire in perception are already about the world, there is no longer any problem in explaining we might justifiably move from our experiences or beliefs about our experiences to beliefs about the world. Of course, if we assume that our empirical beliefs are justified and that these beliefs are either identical to or justified by some of the beliefs we acquire when we perceive, then there still remains the question of how the beliefs we acquire through perception are themselves justified. In response to this, Armstrong, endorses a theory of justification he calls the “thermometer” view: Suppose that ‘p’ is true and A believes that p, but his belief is not supported by any reasons. … [What justifies such a belief?] My suggestion is that there must be a law-like connection between the state of affairs Bap and the state of affairs that makes ‘p’ true such that, given Bap, it must be the case that p. (1973: 166) The thermometer analogy functions as follows. So long as a thermometer is working properly, there is a law-like connection between the temperature of the environment and the temperature reading on the thermometer. Likewise, so long as our belief-forming mechanisms (our perceptual mechanisms) are working correctly, then there will be a law-like connection between the external environment and our beliefs about it.

46  Philosophical Theories

Sense Data and Armstrong’s Theory of Justification In the present context, it is worth noting that there seems no reason that a sense datum theorist could not also endorse this view of justification. If external objects cause sense data, which, in turn, cause beliefs, then so long as there is a law-like connection between the beliefs and the state of the world, these beliefs would be justified on Armstrong’s account. This would potentially give such theories all the purported epistemological advantages of belief acquisition theory. If these theorists don’t endorse such theories of justification, it is because they want to hold onto the ideas that somehow our empirical beliefs are justified by the sensory aspects of our visual experiences. Yet as we shall see, this is just the feature of visual experiences that belief acquisition theorists are accused of overlooking.

As the belief acquisition theory endorses the Common Factor Principle, it, like the sense datum theory, owes a further account of what marks out those visual experiences that constitute successful cases of seeing. When it comes to supplementing this basic analysis of visual experience into an analysis of the success state of seeing, the belief acquisition theory has a relatively natural explanation of what it is for a visual experience to be satisfied. As is indicated in the quote from Armstrong above, if this belief is true, then the subject will thereby be said to see that something is F; if it is false, the subject will be under an illusion (or hallucination). So where the sense datum theory attempted to explain satisfaction in terms of a resemblance between sense data and objects, the belief acquisition theory explains it in terms of the truth of the acquired beliefs. This approach to justification would also enable the belief acquisition theorist to adopt something very like Lewis’s counterfactual dependence condition as a way of ruling out cases of veridical hallucination while allowing for nonstandard/prosthetic seeing (save for the fact that, where Lewis has it that the phenomenal character of the experience is what is counterfactually dependent on the scene, the beliefacquisition theorist would have the belief’s contents be counterfactually dependent). Indeed, Armstrong’s suggestion that there must be a law-like connection between the environment and our beliefs about it would explain why those beliefs are counterfactually dependent. In both the brain before the eyes case and the light meter case, there is not the kind of law-like connection that Armstrong requires, and hence the belief contents do not change in the way they would need to for counterfactual dependence to hold. In both of these cases, if the world were not in exactly the way the subject believes it to be, the subject would still have believed that it was. However, it does so in such a way that it would not rule out nonstandard cases of seeing – so long as the relationship between the external environment and our beliefs about it remained law-like, it would constitute a method of seeing. Can it also rule out Tye’s Tim and Tom case that we met in the previous chapter? This would depend on how widely we are willing to understand the notion of a law-like

The Representational Principle  47

connection: if there is a law-like connection between my visual experiences and robot Tim in the set-up Tye describes – and there would seem to be – then it would not rule out these cases; but outside of this tightly constrained scenario, the law-like connection would break down. Probably the most significant objection that has been raised for the Belief Acquisition Theory is that it has been argued to leave something out: it seems there are occasions in which we experience something to be a certain way without believing it to be that way. If experiences are nothing more than the acquisition of belief, then there doesn’t appear to be room for this kind of disconnect. Examples of this include what we might call cases of known illusions. Take the famous Müller-Lyer illusion.

Figure 3.1 The left line looks longer than the right line, but at least for those who are familiar with the illusion, we don’t end up believing that the left line is longer than the right line. Why not? Because we know that, in cases like this, the lines are in fact the same length despite the fact that they don’t look to be. But then, if perception is simply the acquisition of belief, when we see the Müller-Lyer lines, we should acquire the belief that the lines are the same length. So the theory predicts that, in such cases, we should end up knowingly holding two inconsistent beliefs – that the lines both are and are not the same length – which does not seem possible. The traditional response to this, offered by both Armstrong and Pitcher, is to treat these as special cases. In typical cases, when we perceive that p, we simply acquire the belief that p; in situations where we have reasons to think that experience can mislead, however, we do not simply acquire the belief in question. However, there remains a strong inclination to acquire the belief; it is just that this inclination is suppressed by our knowledge of misleading situations such as these. Given this response, perception is not, strictly speaking, the acquisition of a belief, but the acquisition of an inclination to believe that, in normal cases, is immediately acted upon, but in special cases can be suppressed. This approach offers to avoid the problematic outcome that the subject holds contradictory beliefs by saying that in the illusory cases, the subject does not acquire the belief that p, merely the inclination to believe that p. Yet if by “inclination to believe”, the belief acquisition theorist is really talking about an inclination, then having an inclination to believe is not believing, just as my having an inclination to go to the gym is not the same as my actually going to the gym (more’s the pity). Given this, it is difficult to see why being inclined to believe that I see

48  Philosophical Theories

a cat on the mat should make me say that it seems to me as though I see a cat on the mat. At different points, however, Armstrong defines an inclination to believe in different ways. In Perception and the Physical World, for example, he defines it as “a thought held back from being a belief by other, contradictory, beliefs” (1961: 86). This definition raises the possibility that an “inclination to believe” in fact names a particular kind of mental state – like a thought or belief but, because of the presence of contradictory beliefs, with a restricted functional role – rather than a disposition to form a belief. So on both of these responses, the argument is blocked by denying that the mental state we acquire in perception is really a belief, but is in fact something else: an inclination to believe or a restricted belief-like state. As Kathrin Glüer points out, however, there is another opportunity available to the defender of a belief acquisition theory. This is to block the argument not by denying that the mental state we acquire in perception is really a belief, but by denying that experiential beliefs and regular beliefs have the same content. On this proposal, we should construe perceptual experience as a kind of belief, just one that ascribes phenomenal properties – properties such as looking red or looking round – to objects and not (as regular beliefs do) sensible properties such as being red or being round. If a perceptual experience involves acquiring the belief that the Müller-Lyer lines look different lengths, as opposed to are different lengths, then there is no inconsistency in holding this belief alongside the standing belief that the lines are, nevertheless, the same length. Moreover: on the assumption that experience is a kind of belief, having both of these beliefs nicely captures what intuitively is going on in a case of known illusion. A belief about how things look is precisely what is independent of background belief to the effect that things in fact are different from what they look to be in the situation in question. (2009: 313–4)

Intentional Theories Phenomenal Principle Common Factor Principle Representational Principle

✘ ✔ ✔

Far and away the most popular contemporary way of building a theory with the Representational Principle at its heart, however, has been to resolve this problem in the kind of way Armstrong hints at – by treating perceptual experience as a sui generis state with representational content. Some theorists contend that this initial move – to a view on which all perceptual experiences have content – is relatively neutral as regards what else might be true of perceptual experiences. For example, Susanna Schellenberg (2011: 719–20) argues that so long as a subject of perceptual experience is aware of the world, then the world will seem a certain way to them and that, given this, we can therefore articulate how the world seems to them. This approach therefore allows us to “read

The Representational Principle  49

off” a content simply from the way the world seems to the subject (the passage from Strawson we saw in the previous chapter is an example of this process). To show that this is really a content rather than just a description of how the world seems to the subject, Schellenberg goes on to argue that this description does supply an accuracy condition. So long as the world seems a certain way to the subject, then either the world is the way it seems to be or it is not. If the world is the way it seems to be, then the content will be accurate – it will describe the world correctly; if it is not, then the content will be inaccurate. So this process not only articulates a proposition to serve as the content of the experience, but also provides the experience with accuracy conditions. Susanna Siegel (2010: 45) also presents a similar argument that moves from the premise that visual perceptual experiences present clusters of properties as being instantiated to the conclusion that these experiences have contents. As both Schellenberg and Siegel point out, this argument does not serve to establish anything fundamental about the nature of perceptual experience. In particular, it does not establish what Siegel calls the Strong Content View. On this view, it is not only the case that visual experiences have contents, but that they “consist fundamentally in the subject’s bearing a propositional attitude toward the contents of her experience” (2010: 73). This more substantial view treats visual experience not merely as a sui generis state with representational content, but as a sui generis propositional attitude. Propositional attitudes are mental states – the paradigmatic example is that of belief – that can be analyzed into three components: a subject who takes an attitude toward a proposition or content. For Susanna to believe that New York cabs are yellow is thus for her to take the attitude of belief toward the propositional content . The sui generis representational theory claims that visual experiences are similar to beliefs, in that they involve the subject taking attitudes toward contents that represent the world as being a certain way. So just as Susanna can believe that , she can also see that . This theory – that perceptual experiences are a sui generis kind of propositional attitude – is known as intentionalism: I [take] as ‘intentionalist’ … the theory which treats perception as a kind of propositional attitude, akin to belief”. (Crane 1998: 233)

An intentional theory of perception claims that visual experiences have an intentional content that represents the world as being some way. This is to see experiences as akin to propositional attitudes such as beliefs. (Martin 2004: 745) The key difference between perceptual experiences and beliefs, however, lies in the different attitudes subjects take toward a content. So while the intentionalist approach sees important similarities between perceptions and beliefs, unlike the belief acquisition theory, they do not attempt to simply reduce the one to (the

50  Philosophical Theories

acquisition of) the other. Instead, intentionalist theories retain the distinction between the perceptual and conceptual components – they allow that visual experiences are something over and above beliefs and the acquisition of beliefs. On this view, in perception, subjects perceive that such-and-such is the case, and that this state can then explain the subject’s forming the subsequent belief that such-andsuch is the case.

The Perceptual Attitude Intentionalist theorists have often created their own names for the special perceptual attitude: Byrne, for example, calls the attitude “exing” (2009: 437) and thus holds that when Susanna sees that New York cabs are yellow, she officially exes that . Pautz (2010: 258) suggests we use the term “sensorily entertains”: Susanna sensorily entertains that . However, having noted that the literature contains different ways of expressing this key perceptual attitude, for present purposes I propose to simply use the familiar “perceives”.

As we saw when we introduced the Representational Principle, in virtue of having a content – in virtue of telling us something about the world – a representation can misinform us: it can tell us that things are a certain way when they are not. As the majority of intentional theorists, at least those that we will be considering in this chapter, accept the Common Factor Principle, this enables the intentionalist to echo an important claim about belief in the perceptual cases. This is the idea that the difference between true and false belief is a matter of the truth or falsity of the belief’s content. Applied to the perceptual case, the difference between the good cases of veridical perception and the bad cases of illusion or hallucination will lie in the accuracy or inaccuracy of the perceptual content. In a case when a subject veridically perceives a yellow lemon in front of them, then the intentionalist will claim that the subject is having a visual experience that involves the propositional content that there is a lemon in front of them, and that this content is accurate. If the subject hallucinates a yellow lemon, however, the intentionalist will claim that the subject is having a visual experience with the same representational content, it is just that in this case the content is inaccurate. And just as my believing that such-and-such is the case says something about how I take the world to be, whilst not entailing that the world is that way, my perceiving (exing / sensorily entertaining) that suchand-such is the case also does not entail that the world is that way – I could be in the same intentional state whether perceiving or hallucinating. When it comes to explaining what it is for an experience to be satisfied then, the intentionalist can offer a response much like the belief acquisition theorist: for a visual experience to be satisfied and hence (potentially) constitute a case of seeing is for its content to be accurate.

The Representational Principle  51

Higher-Order Theories and Representationalism So this is the basic intentionalist thesis. As we saw with versions of the sense datum theory that attempted to add representation to their basic analysis of the phenomenology of perceptual experience, however, it is possible to endorse the view that all perceptual experiences involve a sui generis propositional attitude with most any theory of the phenomenal aspects of perception. This approach would deliver what is often called a “weak intentionalism”. Yet there is a way of developing the theory that attempts to explain the phenomenal aspects of perceptual experiences in terms of their intentional aspects. There are two standard approaches here. According to higher-order versions of intentionalism, we have lots of first-order sensory representations (representations about the world produced by the senses), but these are not automatically conscious. They become conscious when the subject has the right kind of higher-order state about that first-order state. Exactly how this is spelled out differs according to different theorists. For instance: •

•

Lycan (1996) argues that the higher-order state is analogous to a first-order visual experience. So a first-order sensory representation R will be conscious when and only when an internal “scanner” is producing higher-order representations of R. (This position is therefore known as a Higher-Order Perception (HOP) or Higher-Order Experience (HOE) theory.) Rosenthal (1990) argues that the higher-order state is a cognitive state such as a thought or belief. On this view, a first-order sensory representation R will be conscious when and only when the subject has a higher-order mental state that is about R. (This position is therefore known as a Higher-Order Belief (HOB) or Higher-Order Thought (HOT) theory.)

So a higher-order version of intentionalism will proposes the following analysis of visual experiences: A subject S has a visual experience as of a property F if and only if: • S is in a first-order state R that represents that F is instantiated, and • R is scanned/the subject of a HOT. The currently more popular form of intentionalism is a first-order form, which – somewhat confusingly – is often called representationalism. Typically, representationalists do not hold that entertaining a certain content suffices for the presence of phenomenology. The standard approach is to hold that a first-order sensory representation R is conscious when and only when it (i) has the right kind of content and (ii) plays the right kind of functional role. Thus Tye, for example, holds that visual experiences are conscious when they have PANIC – Poised, Abstract, Nonconceptual Intentional Content. On this view, the “right kind” of content is content that is both Abstract (does not refer to particular objects)

52  Philosophical Theories

and Nonconceptual (does not require the subject to possess the concepts used to specify the content); the “right kind” of functional role is that the mental state be Poised: poised to impact upon the subject’s cognitive processes.

Pure Representationalism David Chalmers highlights the possibility of a position that does hold that content alone will fix phenomenology – he calls it “pure representationalism” – but suggests that in practice, most representationalists do not endorse this view on the grounds that representational contents can be represented unconsciously (2004: 158–9). However, this kind of view has begun to attract defenders in recent years, including David Bourget (2010, 2017).

Representationalism therefore proposes the following analysis of visual experiences. A subject S has a visual experience as of a property F if and only if: • S is in a first-order state R that represents that F is instantiated and meets other further conditions. As both the Higher-Order and First-Order variants of intentionalism attempt to explain consciousness in terms of content, they have to endorse a claim that I will call the Mirroring Thesis: (M) Difference in phenomenology ↔ Difference in representational content In English, the Mirroring Thesis states that there will be a difference in phenomenology, or what it is like to have an experience, if and only if there is a difference in the experience’s representational content. As (M) is a biconditional, it is important to remember that it should be read both ways. We can call these (M)LR (for reading from left to right) and (M)RL (from right to left) in turn: • •

(M)LR: That any difference in phenomenology necessitates a difference in representational content, and (M)RL: That any difference in representational content necessitates a difference in phenomenology.

The Mirroring Thesis is thus what enables strong versions of intentionalism to avoid adding something else into the picture in order to ensure that the theory can wear the phenomenological hat. So long as differences in phenomenology follow differences in content in this way, then the possibility for grounding phenomenology

The Representational Principle  53

in content, or even identifying the phenomenal aspects of experience with (a certain type of) content, is made available. On such a view, when we specify the key notion of phenomenal character, it will be expressed in representational terms. For example, it may be the experience’s property of bearing a distinctively perceptual attitude to thus-and-such a representational content, or its property of having thusand-such a representational content and being scanned. The key idea is that in grounding the experience’s phenomenology in its content, the specification of the property of the experience that types it by what it is like to undergo it will make reference to the experience’s representational properties. Before we move on to explore some important theories concerning the nature of perceptual contents, let’s briefly sum up where we are now. The general intentionalist paradigm holds that perceptual states are sui generis representational states. These states are belief-like in as much as they involve bearing an attitude toward a content, but differ in holding that perception involves a distinctively perceptual attitude. The question then arises of how the theory explains the phenomenal features of perception. The weak intentionalist holds that some other theory of phenomenology needs to be added to the theory; the strong intentionalist holds that the resources are already present within the theory. Pure representationalists hold that nothing else is required, and that any state involving this perceptual attitude toward a content will be conscious. Higher-order theorists hold that first-order perceptual states become conscious when they become the subject of higher-order perception or higher-order thought, respectively. General representationalists hold that the first-order states are conscious in virtue of playing a particular functional role. The following diagram gives a visual structure of the different theories of perception that fall under the general heading of intentionalism. INTENTIONALISM

Strong Intentionalism

Weak Intentionalism

Higher Order

First Order

Representationalism

Pure Representationalism

Higher Order Thought

Higher Order Perception

Figure 3.2

Theories of Perceptual Content When it comes to discussions of intentionalism, deciding on which intentionalist theory of perception is only half the job, you also need to decide on a theory of perceptual content. A discussion of some of the key choice points for this follow.

54  Philosophical Theories

The Nature of Contents There are many different theories of propositions/contents. Here we will discuss three influential ones. On each of these theories, our linguistic specification of a content expresses a proposition that is identified with the content. The differences between the theories turns on their different conceptions of what propositions are: Possible World Contents: According to the possible world theory of content, our linguistic specification of a content—that “the cat is black” for example— serves to cleave possible worlds in two: those in which the proposition is true and those in which it is false. Possible worlds theories of content identify content with that set of possible worlds in which it is true. They are thus metaphysically committed to possible worlds. Singular (Russellian/Millian) Contents: An alternative theory of content is the singular content theory. The linguistic specification of a content contains singular terms, predicates, connectives, and so on. On the singular content theory, the propositions actually have the referents of the singular terms/predicates as constituents. For instance, the proposition expressed by “the cat is black” is constituted by the cat itself and the property of blackness. Fregean Contents: Like the singular content theory, the Fregean theory identifies contents with propositions. Yet rather than claiming that objects and properties are constituents of propositions, the Fregean holds that propositions are constituted by modes of presentation of objects and properties. Metaphysically, of course, the question is: what are modes of presentation? On one view, modes of presentation are properties. For instance, consider the well-known example of Hesperus and Phosphorus—the names for the evening star and the morning star (both Venus), respectively. Each name corresponds to a different mode of presentation of the planet Venus—the first via the property being the object visible in the evening sky, the latter via the property being the object visible in the morning sky. On this view, when we say that contents are constituted by modes of presentation, we mean that they are constituted by properties of this kind. How should we choose between these alternatives? Well, one consideration turns on what we might want to say about our experiences of qualitatively identical yet numerically distinct objects. A singular content theorist will claim that experiences of distinct but indiscriminable objects will differ in content: an experience of red ball A will have the content whilst an experience of qualitatively identical red ball B will represent that . Yet, as Martin Davies argues: if two objects are genuinely indistinguishable for a subject, then a perceptual experience of the one has the same content as a perceptual experience of the other. … [P]erceptual content is a matter of how the world seems to the experiencer … where there is no phenomenological difference for a subject, then there is no difference in content. (1991: 26)

The Representational Principle  55

If perceptual contents are, in this sense, conveyed to their subject, then this line of argument suggests that any theories of content that make singular reference to the object of the experience should be rejected (see also McGinn 1982). If the attribution of content to an experience is indeed intended to capture how things seem to the subject, then theories that attribute different contents to these experiences would imply that how things seem to the subject of these two experiences would be different, which conflicts with the assumption that they are indiscriminable. This is why Michael Tye, in spelling out his theory of PANIC, insists that perceptual contents must be abstract. However, Matthew Soteriou (2000: 179–80) argues that an abstract content theory that utilizes only existential contents of the form is open to the same kind of objection that has plagued those theories that attempt to explain what it is for a perceptual experience to be veridical in terms of causation. He asks us to consider a subject wearing displacing glasses that shift the apparent location of objects rightwards, such that objects to the left of the subject look directly in front, objects directly in front of the subject look off to the right, and so on. Now imagine we have red ball A placed to the subject’s left such that the action of the displacing glasses makes it look as though there is a red ball in front of the subject. As things stand, the content of this experience is false. Yet we can make it true by adding indiscriminable red ball B directly in front of the subject. Now there is a red ball in front of the subject, just as the experience represents there to be. Given this, it looks as though the content comes out true and the subject’s experience qualifies as a case of perception. Yet intuitively there is something wrong with this experience. One way of trying to accommodate this would be to include a self-referential component into the content of the experience, such that the experience not only represents that , but also that (Searle 1983: 123). Given that the second self-referential component of the content is not true – it is B that is in front of me, but A that is causing the experience of a red ball appearing to be in front of me – this approach would count the experience as non-perceptual. However, this approach has been criticized for getting the phenomenology wrong. The attribution of content to a visual experience is supposed to explain how things seem to us, and it is just not the case that it seems to us as though there is a causal relation between the object and our experience. In order to account for the fact that the experience of the red ball fails to be a case of perception, Soteriou therefore suggests that we should take the content of this experience to be singular, not abstract. He follows Burge (1991) and includes a demonstrative element in the content of the experience. On this view, when I see a red ball, the content of this experience would be that . So, rather than the content of the experience being , which is true, it should be , which is false as it is actually B in front of me. This would also appear to rule out the possibility of veridical hallucination as well as Tye’s Tim and Tom case; in this case, although Tim is the robot in front of me, Tom would be the object that was a constituent of the experience’s content. So the content of my experience would

56  Philosophical Theories

be that , but as it is Tim that is actually in front of me, this content would not be satisfied and the case would not constitute one of perception. Thus, the thought goes, by accepting that perceptual contents are singular for objects, we can use this singularity to explain the difference between important problem cases. Michael Tye also argues that this view can be developed to accommodate indiscriminable cases of hallucination. He suggests that where a perception of a china frog has the china frog itself as a constituent of its content, the content of an indistinguishable hallucination should have a gap in place of the frog. “The two contents thus have a common structure”, Tye suggests. This structure may be conceived of as having a slot in it for an object. In the case of the first content, the slot is filled by the china frog. In the case of the second content, the slot is empty. (2009: 546) This content schema is what fixes the phenomenal character of the experience, and explains what perceptions of distinct but indistinguishable objects as well as indistinguishable hallucinations of such objects have in common. These considerations provide reasons to think that perceptual contents should contain the referents of the singular terms as constituents, but what about the referents of the predicates? As we saw above, on the original Russellian view, the proposition expressed by “the ball is red” would be constituted both by the ball itself and by the property of redness. Yet this has been argued to lead to problematic consequences when it comes to our experiences of color. As we saw in Chapter 2, our color experiences of one and the same surface can vary significantly across time and changes in viewing conditions (A.D. Smith 2002: 27). With this in mind, suppose we have a number of experiences of one and the same red ball. On the singular content view, we would take all of these experiences to have contents that are part-constituted by the ball’s particular property of redness. Yet we might think that as conditions change (or are manipulated by us to make a philosophical point), the ball could be made to look different colors at different times. A singular content that is constituted by the same redness property itself would thus seem to struggle to account for this. This can be used to argue in favor of a Fregean approach, which takes the contents of our experiences of the ball to be constituted by different modes of presentation of properties such as redness. The natural response for the singular content theorist to make is to say that in those cases in which the ball does not look the color it really is, we are therefore under an illusion – we are misrepresenting the color of its surface. Suppose that the lighting is arranged in such a way that the red ball looks pink. In such a case the singular content theorist would not hold that the content of this experience is constituted by the property of redness. Not only would this be false to the phenomenology, it would also have the consequence that the experience qualified as veridical – because the surface really is red. Instead, the singular content theorist would claim that the content is part-constituted by the property of pinkness, and

The Representational Principle  57

argue that this is both true to the phenomenology and to our intuitions that the subject’s experience is misleading. However, when we consider the extent to which our everyday color experiences do vary, this would suggest that many – probably even most – of our color experiences would actually be counted as illusions (A.D. Smith 2002: 28). Susanna Schellenberg argues that these two approaches can be reconciled with a Fregean version of the gappy content approach. On this view: the token content of a perception of object o1 that instantiates property P1 can be expressed in the following way: (contentp) where MOPr1 (o1) is an object-related de re mode of presentation of o1 and MOPr2(P1) is a property-related de re mode of presentation of an instance of P1. (2013: 728) As with Tye’s view, in cases of subjectively indistinguishable hallucinations, these modes of presentation will be gappy. Thus the content of the equivalent hallucination can be given as follows: (contenth) < MOPr1 (_), MOPr2(_)> Yet as the content is constituted by potentially gappy modes of presentation, this view can both accommodate the idea that veridical perception presents us with particular objects, whilst also accommodating the idea that the way these particulars are presented can vary across time and viewing conditions.

Internalism/Externalism? A further important question about the nature of perceptual contents concerns the question of whether the contents of a particular subject’s visual experiences are fixed solely by features of the subject’s brain/central nervous system/body, or whether the environment that the subject is in also plays a part. •

•

Internalists about perceptual content hold that once you have fixed the physically internal states of the subject, you thereby fix the contents of their visual experiences. This has the consequence that any two physically identical subjects will be in states with identical contents, regardless of any differences in the physical or social environments in which they live. Externalists about perceptual content, however, hold that external relations are also partly responsible for fixing contents. According to externalists, therefore, physically identical subjects may therefore have experiences with different contents.

58  Philosophical Theories

Internalism and Externalism The terms “Internalism” and “Externalism” are used in many areas of philosophy, so we should try to make sure we are clear about what we are being internalist/externalist about. The terminology stems from the Twin Earth arguments in Hilary Putnam’s (1975) “The Meaning of ‘Meaning’” (which argues that natural kind terms get their meaning from the physical environment) and Tyler Burge’s (1979) “Individualism and the Mental” (which argues that some terms get their meaning from features of our social environment). In this context, internalism and externalism about perceptual content should be distinguished from internalism and externalism about phenomenology. Internalism about phenomenology claims that internal physical duplicates will share phenomenology; externalism, that features external to the brain and body can determine phenomenology. Having said this, internalism about perceptual content is a natural partner of internalism about phenomenology – if phenomenology fixes content and we are internalist about phenomenology, then content will also be fixed by the internal states of the subject. However, externalism about phenomenology is often paired with strong intentionalism given that (as we shall see) a number of plausible naturalistic theories of content are externalist, meaning that changes in environment can lead to changes in content. If it is also held that content fixes phenomenology, then changes in environment will thereby yield changes in phenomenology. Tyler Burge defends externalism in the case of perceptual content by considering a twin earth-style argument (1986/2007). He asks us to imagine two distinct entities – a small shadow (an O) and a similarly sized and shaped crack (a C). Although these objects are both very small and their discrimination does not carry any evolutionary advantages or disadvantages, it turns out that Os are very common in the environment, whereas Cs are very rare. Burge then asks us to imagine a subject, P, who has grown up being confronted with Os. He contends that “perceptual representations are formed and obtain their content through regular interactions with the environment. They represent what, in some complex sense of ‘normally’, they normally stem from and are applied to” (203). Given this, the perceptual state that P enters when he encounters an O will be veridical – it will represent that an O is present. Now imagine that P, unbeknownst to him, encounters a C. According to Burge, our theory ought to treat this as a misrepresentation – P misrepresents a C as an O. Now Burge asks us to imagine a counterfactual case. Suppose that the optical laws are different and that in P’s environment there are no longer any visible Os. Suppose however that P’s history is exactly as it was in the original environment, it is just that wherever P saw an O in the original case, he now sees a C. As the normal causes of P’s perceptual states have changed, Burge contends that they will now be representing Cs, not Os.

The Representational Principle  59

Finally, suppose that where P saw a C in the original environment, he also sees a C in the counterfactual environment. Given their identical histories, we need not suppose that there are any physical differences between P in the original environment and P in the counterfactual environment, nor need we suppose that there are any differences between the light that impacts upon P in the two scenarios. Yet despite this, where P misrepresented the C as an O in the original environment, in the counterfactual environment, P correctly represents the C as a C. Despite all the physical similarities, there is a difference in their perceptual contents – the contents are determined, in part, by the nature of the environment. To counter this argument, internalists have argued that Burge’s characterizations of the perceptual contents are too fine-grained – that P’s experiences do not distinguish between Os and Cs. This is the kind of response taken by Gabriel Segal, who argues that P’s experiences actually have a less precise content and merely represents the presence of a thin, dark mark that could be either a shadow or a crack (1989). How we resolve this depends upon whether we place more importance on the causes or effects of our experiences in fixing on assignments of content. If we focus on the fact, as Burge does, that they are experiences that under normal circumstances correlate with distinct entities, then this will seem to license the attribution of distinct contents. Yet if we follow Segal and focus on the facts that things will seem the same to P and hence P will behave in the same way regardless of whether or not the object is a crack or a shadow, then this can seem to license the attribution of a common content. Perhaps both of these considerations might be taken to show that visual experiences actually have two kinds of content – a “narrow” internalist content and a “wide” externalist content.

Conceptual/Nonconceptual? As we have seen, intentionalist theories of perception take perception to be a sui generis but belief-like propositional attitude. Yet we normally think that in order to have a certain belief, we need to possess the concepts required to specify the content of that belief. How could we believe that whales were mammals, for example, unless we had the concepts of whale and mammal. Yet if this requirement is also placed upon our ability to entertain perceptual contents, then this will run in to problems explaining the apparent ability to perceive possessed by apparently concept-less creatures such as infants and animals (Evans 1982; Bermúdez 1995; Peacocke 2001). The argument proceeds from the starting point that there are good reasons to think that animals (and young children) can have experiences of their worlds that are similar to ours (with due note taken of the different levels of discriminatory prowess of different animals). Yet, as such unsophisticated perceivers would lack the concepts required to specify the contents of their experiences, this suggests that entertaining the contents of these experiences cannot require the subject to possess the concepts used to specify the contents. And to the extent that mature human perceptual capacities are built on foundations that we share with such creatures, it would therefore seem that our experiences should also not require us to possess the concepts used to specify the perceptual contents.

60  Philosophical Theories

A related concern is that, on the face of it, perception provides us with far more information and at a far greater level of detail than we could possibly capture with the concepts that we possess. As Tye points out (1995: 139), even if we had command of every color name in the color dictionary – all 4,000 of them – our conceptual capacities would still be woefully short when it came to trying to capture the precise shades of color we can discriminate in experience. So any conceptually specified content would lack the requisite richness required to capture the detail we are aware of in perception. In order to adequately specify the rich detail that we can be aware of in perception, then perceptual contents cannot be constrained by the concepts that the subject of experience possesses.

Rich Perception and Demonstrative Concepts The claim that we lack the concepts to capture the rich detail in our experience can be challenged. John McDowell argues that: in the throes of an experience of the kind that putatively transcends one’s conceptual powers—an experience that ex hypothesi affords a suitable sample—one can give linguistic expression to a concept that is exactly as fine-grained as the experience, by uttering a phrase like “that shade”, in which the demonstrative exploits the presence of the sample (McDowell 1994: 56–7) In other words, although we cannot capture the rich detail of experience using concepts such as “red”, “yellow”, “burnt sienna”, and so on, we can conceptually capture this detail by using demonstrative concepts that pick out the very detail that is supposed to be problematic. Yet concerns have been expressed with this response. The conceptualist seems to need to argue that the perceiver must have this demonstrative concept in order to be able to have the experience. But isn’t that upside down? Doesn’t the perceiver have an experience with a certain feature and thereby come to be able to demonstrate that feature? But if this is so, then we need to be able to make sense of the idea that the experience can have this feature prior to the perceiver’s deploying a demonstrative concept. But to explain this, don’t we need nonconceptual content (or some such)?

To avoid these objections, a number of intentionalists (including representationalists) reject the claim that perceptual content, like belief content, has to be conceptual in nature. If we no longer insist that all forms of representational content must be conceptual, the door is then open to introduce a nonconceptual form of content. Essentially, the suggestion is that there are lots of mental states which represent the world around the subject to be a certain way. Sometimes, to be in one of these states requires that we possess the concepts which are used to specify the content; if this

The Representational Principle  61

is the case, as it seems to be for belief, then we say that such a state has conceptual content. On other occasions, however, we don’t have to possess the concepts in order to entertain the content. In these cases, we can say that the state has a nonconceptual content. As Crane defines the notion: For any state with content, S, S has a nonconceptual content, P, iff a subject X’s being in S does not entail that X possesses the concepts that canonically characterise P. (1992: 143) Importantly, the claim is not that a given state might have conceptual content (if we happen to have the concepts) or nonconceptual content (if we don’t). Instead, what makes a state conceptual or nonconceptual is whether one’s having the concepts is a necessary condition of one’s being in that state. So a given state might still be nonconceptual even if we have the concepts used to specify the content, so long as we could still have been in that state had we not had the concepts.

Nonconceptual Content and Concept Learning Another argument for the claim that experiences have nonconceptual contents turns on the explanation of concept learning. Adina Roskies (2008) points out that unless we are willing to endorse the view that color concepts are innate, then we must hold that such concepts are learned. She then argues that the natural explanation of how we acquire color concepts would be that we learn them from having color experiences. Yet if the conceptualist is right, and having red experiences involves the concept of RED, then a vicious circle threatens whereby we cannot acquire color concepts without color experiences, but cannot have color experiences without color concepts. To break out of this circle, Roskies argues, we should accept that the color experiences must therefore have nonconceptual content.

How Do Experiences Get Their Contents? In this chapter, we have considered different intentionalist theories of perception as well as different theories of perceptual content. Before we move to seeing how well these theories wear our two hats, one further issue needs to be raised: how do mental states such as experiences acquire their contents? As representational theories explain what it is for a state to have phenomenology by appealing to the content it has, an account of what is for a state to have a content of a certain kind is needed in order to thereby explain what it is for an experience to have phenomenology of a certain kind. This means that to be complete, such theories must address the question of how experiences get their contents: In virtue of what does a given visual experience have the particular content it does?

62  Philosophical Theories

A theory of content that answers this question is often known as a psychosemantics. As we saw when discussing Burge above, intentionalists have often endorsed some kind of reductive, naturalistic psychosemantics. This is because this approach offers to provide a two-stage naturalistic explanation of consciousness. The first stage involves the explanation of consciousness in terms of content or representation, as exemplified by the strong intentionalist theories discussed above. Then, if we can provide a naturalistically acceptable account of what is involved in a state’s having a particular content, we will have thereby provided a naturalistically acceptable account of what is involved in a state’s being conscious. Although there are many naturalistic theories of content in the literature, including the “asymmetric dependence” theory of content (Fodor 1992), “success semantics” (Whyte 1990), “conceptual role semantics” (Harman 1987), and various hybrid accounts, strong intentionalists have tended to endorse variants of a teleosemantic theory of content (see, e.g., Dretske 1995: 15; Tye 1995: 153; Lycan 1996: 75). Although there are a number of different variants, the core of teleosemantic theories is their focus, in attributing contents to mental states, on what the biological function of these states are. Despite this, there is no reason why (at least so far as I can see) that strong intentionalism could not be combined with any of these theories of content, and indeed it would be an interesting question which of these theories fits best with strong intentionalism.

Tye’s Theory of Content To provide a little more detail for those who are interested, I will briefly outline the approach taken by the representationalist Michael Tye, who combines a teleosemantic approach with a dash of causal covariation. Tye contends that “experiences represent various features by causally correlating with, or tracking, those features under optimal conditions” (2000: 64). So, in order for mental state M to represent F, M must covary with Fness under optimal conditions. When are conditions optimal? “In the case of evolved creatures, [optimal] conditions for vision involve the various components of the visual system operating as they were designed to do in the sort of external environment in which they were designed to operate” (2000: 138). Thus, suppose that an organism evolved in environment E and that in E, M reliably covaries with F. Now suppose that the organism’s sensory mechanisms were selected for because they provided it with information about the environment in which it evolved (E). According to Tye’s theory, M will therefore represent F.

With this in mind, what are the key considerations for a strong intentionalist when it comes to deciding which psychosemantic theory to endorse? Given the nature of the strong intentionalist claim, probably the most important consideration is that, as the strong intentionalist claims that what it is like to have an experience

The Representational Principle  63

is explained by its representational content, an adequate theory of content must ascribe contents to experience that are as rich and detailed as that experience’s phenomenology. However, it has been argued that no bottom-up account of this kind could possibly do the job required of it: the “flat and one-dimensional appeal to specific experientially entertained contents simply served up to the subject by the system lacks the richness, variety, and depth required to capture the manifold looks of the worldly objects that we perceive” (Brewer 2017: 220; for a longer discussion, see Travis 2004). Related to this, it must be able to explain and predict the kinds of content/phenomenology covariation that are stipulated by the Mirroring Thesis. The Mirroring Thesis, recall, stated that any difference in phenomenology necessitates a difference in representational content. With this in mind, consider two experiences, e and e*, which differ in phenomenology. According to the Mirroring Thesis, these two experiences must therefore differ in content. But because the strong intentionalist insists that content is the primary notion, the difference in content has to explain the difference in phenomenology rather than the other way around. So an adequate theory of content must predict changes in representational content in all and only those situations in which what it is like to have the experience would change. How well the theories of content outlined above can explain these features is an important topic for further research.

Schellenberg’s Capacitism Another important consideration is that the metaphysical theory of content must tie up in an appropriate way with the chosen psychosemantic theory. For this reason, as representationalists are often also singular content theorists, it is more natural to see teleosemantic theories of content delivering objects into contents than modes of presentation. However, Susanna Schellenberg’s Capacitism (2018) offers a way to have a naturalistic Fregean theory. Schellenberg’s view begins from perceptual capacities – capacities to detect and discriminate mind-independent particulars – yet contends that in employing a perceptual capacity, we single out the particular in a certain way. Perceptual capacities are thus “the mental counterpart of Fregean modes of presentation” (2018: 52), and employing perceptual capacities thereby yields an experience with a Fregean perceptual content. Moreover, Schellenberg argues that capacities are fallible (2018: 44), meaning that it is possible to attempt to use a capacity to pick out a particular of a certain kind, yet fail to so do either because the relevant particular is not of that kind or because there is no particular there at all. In the former case, this would constitute an illusion, where a mode of presentation presents an object as being a certain way when it is not; the latter case would of course constitute a hallucination, where the lack of an appropriate object yields a gappy mode of presentation.

64  Philosophical Theories

Representationalism and the Two Hats As there are a number of different intentional theories of perception, I will focus the present discussions on the question of how well the most popular of these theories – representationalism – manages to wear our two hats.

The Phenomenological Hat A leading motivation for representationalism has been that it can adequately capture the phenomenology of experience without requiring an appeal to anything metaphysically suspect, such as mental objects and their properties. Reflections on what has become known as the “transparency” of visual experience suggest that this approach is not only metaphysically advantageous, but also better accounts for the phenomenology of visual experience. Gilbert Harman explains this by drawing a parallel between experiences and paintings: In the case of a painting Eloise can be aware of those features of the painting that are responsible for its being a painting of a unicorn. … But in the case of her visual experience of a tree, … she is not aware of, as it were, the mental paint by virtue of which her experience is an experience of seeing a tree. She is aware only of the intentional or relational features of her experience. (1990: 39) Although both paintings and experiences have representational content, Harman suggests, the way we become aware of what is represented in the two cases is importantly different. When we look at a painting of a tree, we can either focus on what the painting represents – the tree – or we can focus on the distribution of paint on the canvas by virtue of which the painting represents a tree; in the case of an experience, however, there is no parallel to the second of these activities: we do not find any “mental paint” which would play a comparable role: When you see a tree, you do not experience any features as intrinsic features of your experience. Look at a tree and try and turn your attention to intrinsic features of your experience. I predict that the only features there to turn your attention to will be features of the presented tree. (1990: 39) This is supposed to be a phenomenological observation: it tells us that when we introspect our visual experiences, all we find are the features of the objects we see or seem to see. To this extent, then we can describe the experience as “transparent” – we see through the experience itself to become aware of the objects of experience. How does the transparency observation support representationalism? Well, transparency tells us that the properties we are aware of in visual experience appear to be properties of presented objects. There is therefore strong phenomenological evidence that the presentational character of a visual experience is constituted by properties of objects. Perhaps, then, it is constituted by the very properties of the

The Representational Principle  65

mind-independent objects that we see. However, Tye argues that the possibility of indiscriminable hallucinations shows that this cannot be the case. In such cases, we can entertain a presentational character of this kind without their being any appropriate mind-independent objects. So although presentational character is constituted by properties of objects, they cannot simply be identified with properties of the objects that we see. Moreover, Tye contends that they cannot be properties of mind-dependent objects – sense data – for the “whole host of familiar reasons” (2000: 46) we met in the previous chapter. Representationalists, however, have an answer. The properties that constitute presentational character are indeed properties of objects. But they are not necessarily properties that are currently instantiated by objects; instead, they are properties that the experience represents as being instantiated. In this way, suggests the representationalist, the phenomenon of transparency can be accommodated without having to find objects to actually bear the relevant properties. On this view, the presentational character of an experience is constituted by the everyday mind-independent properties. The phenomenal character of the experience – the property of the experience that types the experience by what it is like to undergo it – is the experience’s property of representing that these properties are instantiated, or alternatively, the experience’s property of having thus-and-such a representational content.

Phenomenal Character as Identical to Content Tye actually claims that when we introspect, we are aware of “aspects of the content of the experience” (1992). If an experience represents that the Pacific Ocean is blue, say, we can say that blueness is a component of the content of the experience and it is this that we are aware of. Because of this, Tye then officially goes on to identify the experience’s phenomenal character with its representational content (its PANIC). But we must remember that Tye is here using the term “phenomenal character” in the way we have been using the term “presentational character”. As we are using the term, phenomenal characters are properties of experiences and, even though experiences have contents, the contents they have are not properties of those experiences. Everything that Tye wants to say by identifying phenomenal character with PANIC, however, is captured in the presentation here.

The representationalist thus rejects the Phenomenal Principle. That principle claims that if there sensibly appears to a subject to be something which possesses a particular sensible quality, then there is something of which the subject is aware which does possess that quality. According to the representationalist, there sensibly appearing to a subject to be something which possesses a particular property is a matter of the experience’s representing that something possesses that property. For this to be the case, nothing actually need possess that property at all.

66  Philosophical Theories

Some opponents of representationalism have argued that representation is not enough to do the job of making properties present in experience in the way required: Suppose that you and I are having an argument about the battle of Gallipoli. Our attention may be focused entirely on the situation represented; we are not thinking at all about the words we are using, or how we are talking about the people in question. In this case it’s quite natural to say that the words are ‘diaphanous’, our attention falls through them and onto the things we are thinking about. But ‘transparency’ in this sense is not enough to make the situation at Gallipoli ‘present’ to us in the way in which colors are present in color experience. The sense in which redness is ‘there’ in an experience of red can’t be explained in terms merely of the transparency of visual representation. (Campbell 2020: 407; see also Papineau 2016: 335–9) Essentially, this line of argument appears to simply reassert the Phenomenal Principle in the teeth of its rejection by the representationalist. For this reason, it is not clear how compelling this line of attack should be. After all, for representationalists, the representational relation in the case of visual experience is intended to be sui generis. So there is an obvious way in which they can maintain that while the way of representing objects in speech does not suffice to make those objects “present” to us, the way of representing redness in visual experience is special, in no small part because it does make properties “present” to us, even in cases when there is no instance of that property in the immediate environment. Perhaps the most significant objection to representationalism – at least in its most popular guises – turns on the role played in the account by functionalist considerations. As we have seen, representationalists account for the phenomenology of visual experience by appeal to the subject being in a state that represents that there is an object at such-and-such a location. This is then held to explain why it is, for the subject, as if there is such an object. Yet if representation of property F alone suffices for the subject of this representation to be consciously aware of Fness, then property F could not be represented unconsciously (Kriegel 2002; see also Chalmers 2004). But this can seem implausible in the light of pathological phenomena such as blindsight. Blindsight subjects typically have localized damage to an area of V1, the primary visual cortex, which results in a scotoma – a blind spot – in part of the visual field. If a stimulus is presented to the subject in their blind spot, they claim to be unable to see it. However, as Larry Weiskrantz and his colleagues discovered in the 1970s (Weiskrantz et al. 1974; Weiskrantz 1986), if subjects are instructed to guess whether a stimulus is present in that area or to guess its orientation/direction of motion, some subjects perform significantly better than chance. Indeed, some subjects can even perform better than chance when asked to guess which of an X or an O is presented. This suggests that despite subjects lacking conscious experience of the area behind the blind spot, information from that area is nevertheless getting through to the subject in such a way that it can be retrieved in the right circumstances. This suggests the possibility of nonconscious representation of visual information.

The Representational Principle  67

If this is indeed possible, then it would seem there must be something else that distinguishes conscious representations of Fness from nonconscious representations of Fness. And indeed, this is borne out when we consider specific representationalist theories. Only the pure representationalists maintain that representation alone suffices for consciousness, but as we shall see in the next chapter pure representationalism does not typically share the reductive aspirations of the other strong intentionalists (Bourget 2010). The higher-order theorists, however, require that to be conscious, a first-order representation must be either scanned or the subject of a higher-order cognitive state. Similarly, as we noted when we were discussing the different elements of Michael Tye’s first-order PANIC theory, representationalists will typically hold that, to be conscious, a representational state also has to play the right kind of functional role. This means that, in all of these cases, the subject’s being in a state that represents that there is an object at such-and-such a location cannot, by itself, explain why it is, for the subject, phenomenologically as though there is such an object. This raises the question of why these additional considerations – playing the right functional role; being scanned by an internal scanner; being the subject of a higher-order cognitive state – have the ability to change a nonconscious state into a conscious state. As a way of making this concern vivid, we might wonder why, given that Fness can be represented nonconsciously, an (in itself) unconscious representation becomes conscious just in virtue of playing a particular functional role? Couldn’t there be a case in which this representation plays this functional role without thereby making its subject conscious? If there could, then this suggests that this account of why there is something it is like to enjoy a visual experience may be lacking.

The Epistemological Hat According to conceptualist versions of intentionalism – versions that hold that the contents of visual experiences are conceptual – empirical beliefs arise simply from endorsing the contents of visual experiences. This allows for the possibility that, on such a view, empirical beliefs can be rationally, and not just causally, grounded by visual experiences. Moreover, the regress problems do not arise for such a view – when we justify a belief by pointing to another belief, we can always ask what justifies that further belief. (It is in response to this challenge that Armstrong, as we saw, adopted an externalist conception of justification.) Yet here the situation is different. When we respond to the challenge, “Why do you believe that p”? with the answer, “Because I perceive that p”, we are not pointing to another belief as justifying the initial belief but an experience. And the question of what justifies an experience sounds odd – experiences are not the sorts of things that can be justified: we just have them, that is all. So as long as perceiving that p is adequate justification for believing that p, which intuitively it seems to be, we look to have an adequate epistemology. This account of the justification of empirical beliefs has been challenged – it has been argued that, as we could have the visual experience in which we bear the perceptual relation to the content that p when p was false (in the case of hallucination,

68  Philosophical Theories

say), having that visual experience is not adequate justification for believing that p – but a more pressing concern is the fact that, as things stand, this approach is not available to the majority of intentionalists, at least, the majority of strong intentionalists, as they tend to reject conceptualism in favor of nonconceptualism about perceptual contents. Whilst this move does have the advantage of (purportedly) getting the phenomenology right, it comes at an epistemological cost: we can no longer simply talk about the experience’s content simply being endorsed in belief. To have access to this account of empirical justification, then the nonconceptualist needs to say something about the relationship between the nonconceptual contents of visual experiences and the conceptual contents of beliefs, such that the former can be an epistemically suitable ground for the latter. One response by nonconceptualists has been to say that belief contents and nonconceptual perceptual contents do stand in rational relations to one another. For example, if a nonconceptual content specifies a way of filling out space around the subject, then some beliefs about the layout of space will be inconsistent with this content, or entailed by it, or made probable by it, and so on (Heck 2000). Another response has been to hold that visual experiences and beliefs have the same whole contents: when a perception that p causes a belief that p, the whole contents of these two states are of the same type—p. … belief conceptualises the content of perception. So treating the transition from perception to belief in terms of whole contents allows us to explain how perceptions have contents that can be the contents of beliefs. (Crane 1992: 155) Conceptualists, however, are unlikely to be swayed. Even if perceptual contents can stand in rational relations to the contents of beliefs, they will contend that visual experiences with nonconceptual contents cannot be genuine reasons for a subject – reasons that figure as such from the subject’s point of view – unless they have conceptual content. As Brewer argues: even though being in [a state with nonconceptual content] may make it advisable, relative to a certain external end or need, for [a subject] to make the judgement or hold the belief in question, it cannot provide her own reason for doing so. (2005: 219)

Questions • • •

How plausible is the claim that what it is like to have an experience is just a matter of the experience’s representational content? Can a nonconceptual content theorist provide an adequate account of the epistemological role of perception? Can any naturalistic theory of content successfully predict the kind of content variations required by the mirroring thesis?

The Representational Principle  69

Further Reading For a recent defense of a broadly belief acquisition theory of perception, see Kathrin Glüer’s 2008 “In Defence of a Doxastic Account of Experience”. Susanna Siegel defends the claim that visual experiences have contents in her 2010 The Contents of Visual Experience. A comprehensive overview of the different approaches an intentionalist might take can be found in David Chalmers’s 2004 paper “The Representational Character of Experience”. A good place to start thinking about naturalistic theories of content is Barry Loewer’s (1997) essay “A Guide to Naturalizing Semantics”. For the transparency motivation, see Gilbert Harman (1990) “The Intrinsic Quality of Experience”. For a defense of the mirroring thesis, see Alex Byrne’s (2001) “Intentionalism Defended”. For higher-order versions of intentionalism, see Bill Lycan (1996) Consciousness and Experience and David Rosenthal (1990) “A Theory of Consciousness”. Pure representationalism is developed and defended by David Bourget in his (2010) “Consciousness is underived intentionality” and (2017) “Why Are Some Phenomenal Experiences ‘Vivid’ and Others ‘Faint’? Representationalism, Imagery, and Cognitive Phenomenology”. Versions of representationalism are developed in Dretske’s (2003) paper “Experience as Representation” and Michael Tye’s books Ten Problems of Consciousness (1995) and Consciousness, Color and Content (2000). In her 2018 book, The Unity of Perception: Content, Consciousness, Evidence, Susanna Schellenberg develops a detailed Fregean version of intentionalism, which grounds modes of presentation in perceptual capacities. For recent arguments that representationalism cannot adequately capture perceptual phenomenology can be found in Bill Brewer (2017) “The Object View of Perception” and David Papineau (2016) “Against representationalism (about conscious sensory experience)”.

4

Adverbialism and Qualia Theories

Overview The original adverbial theory retains the Common Factor Principle but rejects both the Phenomenal Principle and the Representational Principle. Adverbialists reject the need to appeal to the objects of acts of sensing to explain what it is like for the subject, appealing instead to ways of sensing. The formulation of adverbialism is discussed, and we meet the important many property problem. The connections between adverbialism and the more recent qualia theory from the philosophy of mind are then explored, before we turn to the phenomenal intentionality version of the qualia theory. This approach departs from traditional qualia theory by also endorsing the Representational Principle, along with the Mirroring Thesis we met in the previous chapter. However, phenomenal intentionality reverses the representationalist’s order of explanation: instead of attempting to ground a perceptual experience’s phenomenology in its representational content, the phenomenal intentionalist argues that we should instead ground its representational content in its qualia-based phenomenology.

In the previous chapter, we explored theories that are structured around the Representational Principle and attempted to explain all aspects of visual experience in terms of representation. In these discussions, we didn’t spend much time talking about the metaphysical commitments of intentional theories. Given that these theories didn’t explicitly introduce an unusual kind of entity, like a sense datum, it may seem that there is no real need to look too hard at the attendant metaphysics. However, Uriah Kriegel (2008: 92–4) has argued that things are not quite this straightforward. Focusing on the case of thought, he argues that contents should not be conceived of as that which one thinks of/about on the grounds that this would misrepresent the nature of thought: we don’t think about propositions, we think about the world by being in a state with a propositional content. Instead, contents are that which one thinks with – in other words, the content functions so as to direct the thought or experience onto an appropriate object. Thinking that the tree outside of

72  Philosophical Theories

my window is in bloom is thus a way of being related to the tree itself: the tree is the intentional object of the intentional state. So far, so good. Yet if we accept the Common Factor Principle, then we think that a state of this very kind could be present in cases of hallucination. This might suggest that if intentionality is constitutively relational, then hallucinations would also have intentional objects (2008: 82). He argues, however, that such a view would be metaphysically problematic, as none of the available options – either insisting that the intentional objects of deceptive states exist (just as concrete mental objects such as sense data, abstract objects, or concrete possibilia) or allowing that we can be related to non-existents – are plausible (2007: 310–2). For this reason, Kriegel argues that being in an intentional state cannot constitutively require the subject to bear a relation to an intentional object: Perhaps representation often involves a relation to the represented, but it never does so constitutively. That is, it is never the case that a representation represents in virtue of bearing a relation to the represented, so that the destruction of the relation would destroy the representation. There is of course a wellknown precedent in modern philosophy to a nonrelational treatment of an apparently relational, indeed apparently intentional, phenomenon. This is the adverbial account of sense perception. (Kriegel 2007: 312–3) Essentially, Kriegel’s contention is that despite appearances, intentional states cannot be constitutively relational and that in order to see how an apparently relational mental state can be treated non-relationally, we should look to adverbialism in the philosophy of perception as a model.

Adverbialism Phenomenal Principle Common Factor Principle Representational Principle

✘ ✔ ✘

Much like the belief acquisition theory, adverbialism arose in the mid-twentieth century as a reaction to the contention that the sense datum theory faces problems because of its introduction of mental objects, which leads us into epistemological and metaphysical difficulty. And much like the belief acquisition theory, it locates the problem in the sense datum theorist’s acceptance of the Phenomenal Principle. This principle, recall, claims that: (P) If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that quality. Barnes argues that this principle is based on a fallacy: “That something appears pink to me is not a valid reason for concluding either that that thing is

Adverbialism and Qualia Theories  73

pink or that there is some other thing which is pink” (1965: 153). As Robinson also notes when discussing the arguments in favor of mental objects, our language is stocked with idioms, such as “seems”, “appears”, and “looks”, which we can confidently use to assert for instance that a non-pink elephant seems pink (1994: 40). This kind of objection can be augmented by showing how statements that appear to commit us to the existence of certain objects can be reinterpreted so as to avoid such commitments. A highly influential approach to this project, an approach that, in turn, generated an alternative theory of perception, focuses on the possibility of providing adverbial translations of such sentences. To explain: note that many familiar sentences of English relate a subject to an adjectivally modified object. Thus: 1 Sofia had a heavy bag. 2 Santiago wore a broad hat. For each of these sentences, the subject (Sofia/Santiago) bears a certain relationship (having/wearing) to a certain object (the heavy bag/the broad hat). If sentences (1) and (2) are true, then we appear to be committed to the existence of those objects – if it is true that Sofia has a heavy bag, then the heavy bag that Sofia has must exist. However, the following sentences have a similar surface structure: 3 Sara gave an energetic performance. 4 Ali wore a broad smile. Sentences (3) and (4) also seem to relate subjects to objects in the very same way. Given this, we might think that if sentences such as (3) and (4) are true, then we must also be committed to the existence of certain strange objects – performances and smiles – which can both be adjectivally modified and to which we can bear relationships of giving and wearing. However, we don’t have to see things in this way. Although (3) and (4) have the subject : relationship : adjective : noun form, they can both be translated into a subject : verb : adverb form whilst retaining their sense. This translation has three stages. First, we map across the subject. Then we convert the noun from the original sentence into a verb (thus, in the case of (3) and (4), “performance” becomes “performed” and “smile” becomes “smiled”). Finally, we convert the adjective that modified the noun into an adverb that modifies the verb, thus “energetic” translates as “energetically” and “broad” as “broadly”. For (3) and (4), this yields the following translated sentences: (3trans) Sara performed energetically. (4trans) Ali smiled broadly. What is notable here is that (3trans) and (4trans) both make sense and intuitively mean the same as the pre-translated versions (3) and (4).

74  Philosophical Theories

Now let us try this with (1) and (2). Once the three-stage translation process is completed, we get the following sentences: (1trans) Sofia bagged heavily. (2trans) Santiago hatted broadly. Unlike the previous case, (1trans) and (2trans) are not sensible sentences of English and therefore could not mean the same as the coherent sentences (1) and (2). The moral of the story is that, whilst some sentences that have the subject : relationship : adjective : noun surface form can be translated into subject : verb : adverb form without losing their sense or changing their meaning, others such as (1) and (2) cannot. Where we cannot perform these translations, then if the sentences are true, the thought goes, we really are ontologically committed to the existence of the objects referred to in those sentences. However, where we can perform the translations, then even true sentences do not really commit us to the existence of their objects; they are just loose ways of talking. In the case of sentences such as (3) and (4), what we are really committed to are not objects such as performances and smiles which can be adjectivally modified, but rather events of performing and smiling which can be modified adverbially. In this light, consider the kind of sentence that, it was suggested, carries an implicit commitment to the existence of mental objects: 5 Junior had a pink sense datum. 6 Riya had a triangular afterimage. Then take these statements and push them through the translation to get: (5trans) Junior sensed pinkly; (6trans) Riya afterimaged (= sensed) triangularly. Adverbialists then contend that (5trans) and (6trans) mean the same as the original sentences. This is then argued to show that, contrary to first impressions, (5) and (6) do not commit us to the existence of problematic mental objects after all. Instead, like (3) and (4), they only commit us to events: events of subjects sensing in certain ways. These considerations have been taken to show the way to an alternative theory of perception that does not require mental objects. An adverbial theory of perception contends that visual experiences are not episodes of sensing sense data, but are rather episodes of sensing in particular ways. As Ducasse, one of the original adverbialists, put it: To sense blue is then to sense bluely […] Sensing blue, that is to say, is a species of sensing—a specific variety of the sort of activity generically called “sensing”. (1942: 232–3)

Adverbialism and Qualia Theories  75

The Metaphysics of Adverbialism In the main text, I outline an events-based metaphysics of adverbialism. This takes an adverbial statement, such as “Mia senses redly”, as committing us to the existence of an event whose particular nature as an event is described by the adverbial modifiers (Davidson 1980; Chirimuuta 2015: 140). This analysis reads “Mia senses” as committing us to the existence of an event of sensing of which Mia is the subject and “redly” as ascribing the property of being redly (whatever that may be) to the event. An alternative metaphysics, however, is outlined by Michael Tye, who suggests we treat “Mia senses redly” as a subject-predicate statement. According to this analysis, “Mia” refers to the subject and “senses redly” is a predicate, which functions to ascribe the property to the subject, Mia, of sensing redly (Tye 1984). Adverbialism, as a theory of perception, is often sorely misunderstood. There is something in the translation argument above that can make the theory look like a verbal dodge, not a substantive theory of perception at all. As Cornman, himself a defender of adverbialism, says, “merely devising an artificial terminology in which the problems [faced by sense data] do not arise neither solves nor dissolves the problems” (1971: 188). But as we shall see, adverbialism offers much more than an artificial terminology. In getting away from the sense of artificiality, an important thing to note is that, although the adverbs that are appealed to in these particular translations are nonstandard, the underlying idea is actually quite familiar. Imagine that instead of seeing a red triangle, you see a tree through heat haze. In such a case, we do not see a blurry tree – the tree doesn’t become blurry – rather we see the tree blurrily. The object we see doesn’t change – the way we see the object changes. In many ways, this observation is the core of an adverbialist theory of perception. For example, when I see a red object, according to the adverbialist, I don’t sense a red mental entity, but rather sense the material entity in a particular way: redly. Despite rejecting the Phenomenal Principle in this way, adverbialists do nonetheless typically agree with sense datum theorists (and intentionalists) in endorsing the Common Factor Principle, which claims that: (C) Phenomenologically indiscriminable veridical perceptions, hallucinations and illusions have an underlying mental state in common. If a subject sees a stick half-immersed in water, then according to the adverbialist, they don’t sense a bent stick, but rather sense a straight stick bently. If I see a white object under red light and thereby suffer from the illusion that I see something red, I once again sense redly; but in this case, a white object is sensed redly. And finally, a hallucination of a red object will also involve my sensing redly; it is just that in this case, nothing is thereby sensed. I still sense in a certain way – redly – but no object is thereby sensed.

76  Philosophical Theories

According to adverbialism then, what it is like for me is a matter, not of the object sensed (as there may not even be an object) but rather of the way I sense. For this reason, adverbialism embodies a rejection of the Phenomenal Principle. That principle says that: (P) If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that quality. In both the illusory and hallucinatory cases just described, we have seen that there can sensibly appear to a subject to be something which possesses the sensible quality of redness without its being the case that there is something of which the subject is aware which possesses that quality. In the case of illusion, the only object of which the subject is aware is white, not red, and in the case of hallucination, there is no object of which the subject is aware at all. Instead, in both of these cases, its sensibly appearing to the subject as though something is red is accounted for, not by what the subject senses, but rather by the way the subject senses: in both cases, the subject senses redly. In all of the different cases – veridical, hallucinatory, or illusory – the subject is simply in a distinctive kind of experiential state. So the adverbialist proposes to analyze visual experiences as follows: A subject S has a visual experience as of a property F, if and only if: • S senses in a particular way (F-ly). As with the previous theories, this is an analysis of the neutral category of visual experience. For such an episode of sensing to qualify as a perception of Fness, other criteria will need to be met. The issues the adverbialist faces here are similar to those faced by both the sense datum theorist and the intentionalist, save for an additional difficulty: where the sense datum theorist can explain what it is for an experience to “match” the world in terms of resemblance, and the intentionalist can explain this in terms of the experience’s content being true, the adverbialist doesn’t have such a straightforward option available. So in addition to needing a way to distinguish true veridical perceptions from hallucinations that merely happen to match the world, the adverbialist also needs an account of what it is for an experience to match the world in the first place.

The Many Property Problem Let us now turn to some objections to adverbialism. Thus far, we have been discussing simple adverbial statements of the form “Mia senses redly”. Yet, as Jackson points out: Our statements about afterimages are not just to the effect that an image is red, or square, or whatever; they are also to the effect that an image is red

Adverbialism and Qualia Theories  77

and square and … [The problem for] the adverbial theory turns on this point that an afterimage has many properties, and will be referred to as the manyproperty problem. (1975: 129) As this objection turns on the fact that (putative) sense data can have more than one property, this objection has become known as the many property problem. Let us imagine then that we are looking to provide an adverbial analysis of the following statement: Mia is experiencing a red, square afterimage (Mia is sensing a sense datum which is both red and square). How should we analyze this?

Sensing F-ly and G-ly The most obvious response would simply be to say that to sense a red, square sense datum is to sense redly and squarely. This would give us the following conjunctive principle of analysis: “X senses an F, G, etc. sense datum” is to be analyzed as “X senses F-ly and G-ly and etc.” So, “Mia is experiencing a red, square afterimage” would be analyzed as “Mia is sensing redly and squarely”. This analysis has the advantage of explaining the entailment from “I have a red, square afterimage” to “I have a red afterimage” and “I have a square afterimage”. But it faces the following problem. There is an important difference between the following two total afterimage experiences: 1 Simultaneously having a red, square afterimage and a green, round afterimage. 2 Simultaneously having a red, round afterimage and a green, square afterimage. According to the principle of analysis, the adverbial translation of these would be: (1trans) Sensing redly and squarely and greenly and roundly. (2trans) Sensing redly and roundly and greenly and squarely. But these are equivalent. Hence the conjunctive formulation doesn’t seem able to account for the differences between these total experiences. If you’re not sure about this, consider the following sentences and their putative analyses: (3) Mia has a red, round afterimage and a green, square afterimage. (3trans) Mia senses redly and roundly and greenly and squarely.

78  Philosophical Theories

(4) Mia has a red, square afterimage. (4trans) Mia senses redly and squarely. But whilst (3trans) entails (4trans), (3) does not entail (4). The fact that the entailments go awry suggests that the translations are not in fact equivalent.

Sensing F-lyG-ly An alternative response to the many property problem is suggested by the ambiguity of the following: (5) Hugo spoke impressively quickly. There is an interpretation of this which reads as conjunctive (Hugo spoke impressively and he did so quickly) – both adverbs modify the verb. However, an alternative interpretation reads impressively as modifying not the verb “spoke” but the adverb “quickly” (it was the speed of Hugo’s speaking which was impressive rather than the speaking itself). The thought would be that our principle of analysis would not have to be conjunctive (where both adverbs modify the verb), but instead accumulative, where each additional adverb modifies the previous one. Thus: “X senses an F, G, etc. sense datum” is to be analyzed as “X senses F-ly G-ly etc.” So when we say that Mia had a red, square afterimage, we should analyze this as follows: Mia sensed redly squarely. But rather than take both adverbs as modifying “sensed” (i.e. as being modes or ways of sensing), we take one adverb to modify the verb and the other to modify the first adverb. This leads to the first problem – how are we going to non-arbitrarily decide which adverb modifies the verb and which modifies the first adverb? There seems to be no principled reason to answer this question in a particular way. But even if we do, we run into another problem. Let us say that we decide that squarely modifies sensing and redly modifies squarely. But now, “redly” means something different in “Mia sensed redly” and “Mia sensed redly squarely”, whilst “red” means the same in “Mia had a red afterimage” and “Mia had a red, square afterimage”. This suggests once more that the analyses are not equivalent. Another way of seeing this is to look at the entailments on this approach. “Mia had a red, square afterimage” entails that Mia had a red afterimage, but “Mia sensed redly squarely” does not entail that Mia sensed redly. To see this, note that “Hugo spoke impressively quickly” does not entail that Hugo spoke impressively – the speech might have been impressively quick, but rubbish.

Adverbialism and Qualia Theories  79

Sensing FG-ly The final way Jackson suggests we might try and solve the many property problem is as follows: “X senses an F, G, etc. sense datum” is to be analyzed as “X senses FG(etc.)ly”. Where FG-ly names a simple (fundamental) mode of sensing, not a complex, derivative mode of sensing which is built up out of semantically significant components. So, the claim that “Mia has a red, square afterimage” should instead be analyzed as “Mia sensed red-square-ly”. In response to this, Jackson objects that: On this view, someone who remarks on the common feature in having a red, square afterimage and having a red, round afterimage [must be interpreted as] making a plain mistake—the first is sensing red-square-ly, the second sensing red-round-ly, which are different, and that’s that. But far from being a plain mistake, this remark looks like an evident truth. (1975: 133) A further problem is that afterimages can also have other properties – fuzziness and sharpness for example. But if I have a red, square, fuzzy afterimage, then this must be analyzed as sensing red-square-fuzzy-ly; so, sensing red-square-ly isn’t, as it happens, basic. But Jackson suggests that if the basic level of adverbial analysis has n components (i.e. is sensing F1-F2-…-Fn-ly), he will be able to point out another component at level n + 1 so that the adverbialist “cannot ever give even a single example of a basic mode of sensing, and thus cannot ever complete even one of his adverbial analyses”.

Sensing of-a-F-G-ly However, Tye (1975) follows Sellars (1975) in complaining that Jackson has missed the point. According to Tye and Sellars, the correct interpretation of the view is arrived at by the following grammatical transformations: A. Jian has a sensation of a red square, becomes B. Jian has an of-a-red-square sensation. Note that the phrase “of-a-red-square” in (B) is functioning as an adjective modifying the noun “sensation”. We then do our adverbial transformation to get: C. Jian senses of-a-red-square-ly or, as it is sometimes put, D. Jian senses in an of-a-red-square manner.

80  Philosophical Theories

Tye goes on to suggest that on this approach, “manners of sensing are conceived of as resembling and differing from one another in ways systematically analogous to the ways in which their ‘corresponding’ physical objects, i.e. their normal causes, resemble and differ from one another” (1975: 139). Thus, “red” and “square” in “an of-a-red-square manner” have a sense which is “derivative from and analogous to” (140) their sense when applied to normal physical objects. Thus the relationships which seem to hold between sensings can be accounted for as parasitic on the relationships which hold between physical objects—it is the resemblances between red, round and red, square objects which explain the resemblances between red, round and red, square sensings. This analysis is therefore claimed to rebut Jackson’s concern that we will never have a basic mode of sensing by providing a sense in which sensing in “an of-a-red-square manner” does include “red” and “square” as components.

The Many Relations Problem In later work, Tye (1984) develops a more complex adverbial semantics that he claims solves the many property problem. Treating adverbial statements as subject-predicate statements, Tye introduces some new operators to augment standard first-order predicate calculus to create a structured predicate theory. So “Mia senses” would be cast as Sj, “Mia senses redly” as R(Sj), and “Mia senses circularly” as C(Sj). “If [Mia] hallucinates a red, circular object, then redness and circularity … are combined together or united” (1984: 217). Tye attempts to capture this by introducing a new two-place operator “Coin” (for “coincidence”) and suggests this can be written RCoinC(Sj). So the kind of cases that give rise to the many property problem – cases such as “Mia has a red, round afterimage and a green, square afterimage” – would be regimented as RCoinC(Sj) & GCoinSq(Sj), which can be distinguished from Mia has a green, round afterimage and a red, square afterimage. In a recent paper, Alexander Dinges has argued that the many property problem resurfaces at the level of relations between apparent objects, as seen in cases such as Mia “has a red afterimage that is brighter than a green one and to the left of a yellow one” (2015: 235), and that Tye’s structured predicate theory cannot be extended to accommodate such cases. In this light, Justin D’Ambrosio (2019) attempts to revitalize the events analysis of adverbialism by developing an events-based semantics for adverbial statements that, he argues, can solve both the many property and many relations problems.

Adverbialism and the Two Hats The Phenomenological Hat Inasmuch as adverbialism is initially presented as a response to the sense datum theory, we might think that it suffers from the same problems when it comes to getting

Adverbialism and Qualia Theories  81

the phenomenology right. For example, if sense datum talk of lights and colors, patterns and patches cannot capture what it is like for a subject, then it would appear that neither could talk about sensing light-ly, color-ly, pattern-ly, and patch-ly. But this objection would miss the point of the appeal to the adverbial transformations. The reason adverbialists highlight the availability of these transformations is as a means to show that whatever can be said in terms of sense data can also be said without committing ourselves to such things. Unlike the sense datum theorist, the adverbialist does not have to be committed to the claim that the phenomenology of any experience can be captured so simply. To consider what the adverbialist might have to say about the phenomenology of a more complex experience, consider what the Sellars/Tye approach would have to say about Strawson’s example (from Chapter 2) of seeing the red light of the setting sun filtering through the black and thickly clustered branches of the elms, seeing the dappled deer grazing in groups on the vivid green grass, and so on. Their translations would take this report and transform it into the following: (CStrawson) Subject S senses of-the-red-light-of-the-setting-sun-filtering-throughthe-black-and-thickly-clustered-branches-of-the-elms-and-the-dappled-deergrazing-in-groups-on-the-vivid-green-grass-ly, or, as it is sometimes put, (DStrawson) Subject S senses in an of-the-red-light-of-the-setting-sun-filteringthrough-the-black-and-thickly-clustered-branches-of-the-elms-and-thedappled-deer-grazing-in-groups-on-the-vivid-green-grass manner. This seems to be a pretty accurate report of what it is like for subject S to have this experience and, note, a report that can be true regardless of whether or not the subject is perceiving or hallucinating. However, when considering the success of this kind of move, we must bear in mind that these statements are supposed to be understood as claiming that the subject senses in the kind of way that is normally caused by the red light of the setting sun filtering through the black and thickly clustered branches of the elms and the dappled deer grazing in groups on the vivid green grass. Once this is clarified, we might wonder whether this approach actually offers an explanation of why this state has the phenomenology it does or whether it simply takes this for granted. In particular, recall that the sense datum theorist would insist that to do justice to the phenomenology of this experience – which could of course occur in the absence of red light, vivid green grass, and the like – we must explain how it is that we are consciously aware of redness and vivid greenness. In response to this, the sense datum theorist, by way of the Phenomenal Principle, contends that it is because we are aware of objects that instantiate these properties. What can the adverbialist say in place of this explanation? Given that the kind of sensing that is normally caused by the red light of the setting sun (etc.) could occur in its absence, what explains our conscious awareness of redness and vivid greenness? It seems that the adverbialist has no explanation to offer. Instead, they seem to have to hold that it is just a brute fact about the kind of sensing that is caused by the red light of the setting sun (etc.) that it has the phenomenology it does.

82  Philosophical Theories

The Epistemological Hat On the face of it, adverbialism looks to have one up on the sense datum theory, epistemologically speaking. This is because it doesn’t have intermediate objects of awareness to “get in the way” of real, external objects. As we saw, the sense datum theory claims that when we have veridical experiences of objects in the world, we are aware of these objects indirectly, in virtue of being aware of different objects: sense data. The adverbialist, however, does not have to make such a claim. According to adverbialism, when we experience objects in the world, we are aware of those very objects by sensing in a particular way. So adverbialism can allow that we are aware of material objects directly, and hence demand the title of direct realism. Nevertheless, a pure adverbialist theory, just like a pure sense datum theory, both endorses the Common Factor Principle and rejects the Representational Principle. That is, a pure adverbialist theory insists that there is nothing more to a particular visual experience than sensing in the right kind of way, and maintains that this kind of experience could occur even if the world were not as it seems to be. But as was the case with the classical sense datum theory, it can be difficult to see how having such an experience – an experience that could occur without the world playing ball – could put us in a position to know anything about the world. If we stick with a foundationalist epistemology, then the question of how we move from a justified belief about our own sensory states to a belief about the world crops up for the adverbialist, just as it did for the sense datum theorist. However, the fact that the adverbialist’s picture of experience is not modeled on a picture of object awareness can make a move away from a foundationalist epistemology seem more appropriate. Given the sense datum theorist’s picture of experience, it can seem almost mandatory to try and justify my belief that there is a red object in my environment by appeal to my belief that I am sensing a red sense datum. As the adverbialist does not model experiential awareness on object awareness in this way, the adverbialist could say that we become aware of external objects by sensing in a certain way, but that our beliefs about such objects need not be grounded in our beliefs about those episodes of sensing. Instead, the adverbialist could argue that whilst our experiences do cause our beliefs about the world, they do not also justify those beliefs. This non-foundationalist claim could then be supplemented by an appeal to an alternative theory of belief justification of the kind discussed in the previous chapter. Of course, this kind of view also faces its own challenges. One initial concern might be that it appears to conflict with a very powerful intuition: that when I see the vivid green grass and come to believe that the grass is green, my belief about the color of the grass is not only caused by but is also justified by my experience. Butchvarov also argues that the mere fact that object: x is causally related to S’s sensing in a certain way can no more reasonably be described as S’s being conscious of x than the fact that the presence of carbon monoxide in the air is causally related to S’s having a headache can be described as S’s being conscious of carbon monoxide. (1980: 273)

Adverbialism and Qualia Theories  83

In other words, Butchvarov argues that S’s sensing in a particular way, even in the best possible cases in which this is appropriately caused by an object, cannot ever amount to being conscious of that object. This casts doubt on the claim that adverbialism is a theory that enables “direct” awareness of the external world.

Color Adverbialism In recent years, adverbialism – or at least a limited form of adverbialism – has begun to make a bit of a comeback, driven in part by developments in the philosophy of color (Chirimuuta 2015; Gert 2017). As we will explore in more detail in Chapter 7, philosophical theories of color are shaped by a number of factors, including how they attempt to capture or reject a range of intuitions and empirical discoveries about the nature of color, and our experiences of it. For instance, Mazviita Chirimuuta reviews the current state of color science and concludes that this shows that the perceived “color of a particular surface will be a function of its relations to other surfaces”, which, in turn, implies that “there is no one true color that an object has; colors are not assumed to belong to individual objects” (2015: 74). Having said that, she recognizes that: it would be impossible to use color for the identification, recognition, and memorization of objects [as we do], without some fairly stable color percept to be associated with an individual object. This constancy need not be absolute in the sense that there is no apparent difference in hue as the lighting changes. Rather, there needs to be some sort of label that stays relatively fixed from one viewing occasion to the next. The gross perceptual categorizations named by our basic color terms do this remarkably well. To know that lemons are yellow is enough of a guide for finding lemons in different situations, even when subtle changes in lighting, shade, and the peel of the fruit cause noticeable changes in perceived hue. (2015: 89) In Gert’s version of the view, objects such as lemons and bananas can be said to have “rough colors” (2017: 87) – the kind of generic color that is described by a color word such as “yellow” – but he agrees with Chirimuuta in denying that objects are any particular fine-grained shade of yellow. Yet we are aware of such fine-grained colors in our experience of objects such as lemons and bananas. To accommodate this, Chirimuuta emphasizes that perception is an active interaction with the world, and locates fine-grained colors not as properties of perceived objects, but as properties of our perceptual interactions with stimuli. As these perceptual interactions are relations between perceivers and stimuli, colors are thus both “ways stimuli appear to certain kinds of individuals” and “ways that individuals perceive certain kinds of stimuli” (2015: 142). As this phrasing suggests, the idea that objects are perceived in different ways aligns well with an adverbialist approach.

84  Philosophical Theories

As the adverbial aspects of this picture are limited to the colors we experience, this view can deal with the many property problem in the following way. As only color experience is to be treated adverbially, and colors are ways of experiencing objects/stimuli, in the case in which one sees both a red, round object and a green, square object, this view can connect the right colors with the right objects by noting that, in such a situation, it is the round object that would be seen redly and the square object that would be seen greenly. In this way, the ways of experiencing – the colors – are bound to the objects in a way that enables the view to avoid the many property problem for cases of successful perception (Gert 2017: 108; Chirimuuta 2015: 153fn). However, it does still face the problem Jackson identifies at the bottom of the box below: that it is not clear how this approach would resolve the many property problem for hallucination (although see Chirimuuta 2015: 155–6 for discussion). Moreover, we might also wonder whether the very aspect of the picture that enables this response to the many property problem – that the adverbial aspects of the view are limited to color experience – will cause problems for attempts to extend color adverbialism to a more general theory of the conscious aspects of perception (Buccella 2017).

The Complement Objection Jackson notes that normally it is “not possible for a person at a given time to V both F-ly and non-F-ly. I can sing badly easily enough, but I cannot sing both well and badly at the same time” (1977: 69). However, where sensing is concerned, it seems that we can sense both F-ly and non-F-ly at the same time – “I may have a red and a green afterimage at the same time” (ibid). Jackson considers the response that: though one cannot V both F-ly and non-F-ly at a given time, one can V F-ly with respect to A and non-F-ly with respect to B. For instance I can, during a concerto, listen happily to the strings and unhappily to the piano. (ibid) Yet he argues that whilst we may be able to understand this in the case of perception (I am sensing redly with respect to the round object and greenly with respect to the square object), it will not work for hallucinations, given that there are no appropriate physical things to be sensing F-ly with respect to: “It is hard to see what these could be other than the mental objects of the act-object [i.e. sense datum] theory” (ibid.).

Qualia Theory Phenomenal Principle Common Factor Principle Representational Principle

✘ ✔ ✘

Adverbialism and Qualia Theories  85

One prominent theory of phenomenal features that has been argued to have close links to adverbialism (e.g. Crane 2001; Papineau 2014) is qualia theory. On this view, experiences have certain intrinsic, consciously accessible properties that are responsible for the experience’s phenomenology and can be accessed by introspection: when you attend to … your experience of the redness of an apple, you are aware of an intrinsic quality of your experience, where an intrinsic quality is a quality something has in itself, apart from its relations to other things. (Harman 1990: 33) Such intrinsic, introspectable properties are what we are calling qualia. Crane argues that this connects to adverbialism as follows: consider once again the sense datum statement, “Mia senses a red sense datum”. As we have seen, this statement ascribes a sensible quality – phenomenal or sensible redness – to a mental object and states that Mia stands in the sensing relation to this object. In this way, the sense datum theory offers an explanation of why it is like this for Mia – Mia experiences red because Mia is aware of a sense datum that instantiates phenomenal redness. Given this, a natural reading of the adverbial equivalent “Mia senses redly” takes it to both identify a sensory event and ascribe a property to that event, a property which is marked by “redly”. If we follow this similarity through, the most natural interpretation would take this property to also be a phenomenal or qualititative property that explains why it is like this for Mia. On this natural reading of adverbialism, then the adverbialist, just like the qualia theorist, “holds the qualities sensed in experience to be modifications of experience itself” (Crane 2001: 177). When exploring the literature surrounding qualia, it is easy to get perplexed by the fact that the term has been used in a variety of different ways that is confusing, even for philosophers. For instance, there is an extremely benign usage that treats an experience having qualia as equivalent to its having phenomenal character or there being something it is like to have the experience (e.g. Chalmers 1996: 4); on such a definition, almost all parties would accept that there are qualia. At the other end of the spectrum, there are extremely precise definitions, such as Dennett’s characterization of qualia as “ineffable, intrinsic, private, directly apprehensible properties of experience” (1988: 639). On this definition, most parties  – Dennett himself included – would simply deny that qualia exist. For present purposes, I am using an intermediate definition, according to which qualia are intrinsic, introspectable properties that determine an experience’s phenomenal character. This definition treats qualia as a theory of phenomenal character, in the same way that representationalism is a (competing) theory of phenomenal character, rather than as a synonym for phenomenal character. For present purposes, we shall also assume that qualia themselves are nonrepresentational, as this is a part of the traditional qualia theorist’s position (although, as we shall see later, this claim is not accepted by all who endorse qualia in their theorizing about perception).

86  Philosophical Theories

The Connection between Adverbialism and Qualia Theory It is worth taking a moment to note that the attempt to connect adverbialism and qualia theory is not without its difficulties. For example, as we discussed above, an events analysis of adverbial statements is not the only player in the game; there is also the interpretation that analyzes them as subject-predicate statements. On such a view, the adverbs would not serve to modify events/experiences as Crane suggests, but to identify the properties instantiated by the perceiver. However, to the extent that qualia theorists might be inclined to endorse the view that qualia are fixed by the intrinsic properties of perceivers, this may not prove too problematic. Another concern with too closely linking adverbialism and qualia theory is raised by Christopher Hill, who argues that an adverbialist understanding of qualia cannot be reconciled with the view that qualia are properties that we can directly apprehend in experience. “[W]hile adverbialism allows that qualia exist,” Hill notes: it maintains that they only exist as forms of perceptual awareness – as ways of being perceptually aware of non-qualitative phenomena. That is to say, adverbialism maintains that qualia are adverbial qualifications of an underlying generic relation of perceptual awareness, a relation that agents bear to external objects. (Hill 2014: 207) However, to the extent that we agree that experiences are transparent, the claim that an adverbialist understanding of qualia cannot explain our ability to become aware of qualia may seem more of a blessing than a curse.

Qualia theory has been much discussed in the philosophy of mind more broadly, and it – or something much like it – is also present in a number of theories of perception as a way of explaining and accounting for the phenomenological features of perceptual experience. As an example, this is Ned Block: Much of recent philosophy of perception is oriented towards accounting for the phenomenal character of perception—what it is like to perceive—in a non-mentalistic way—that is, without appealing to mental objects or mental qualities. In opposition to such views, I claim that the phenomenal character of perception of a red round object cannot be explained by or reduced to direct awareness of the object, its redness and roundness—or representation of such objects and qualities. (2010: 23)

Adverbialism and Qualia Theories  87

Here, Block rejects the strong intentionalist view of the previous chapter: that phenomenal character is or is determined by an experience’s representational content. However, Block, like many other qualia theorists, does not wish to reject the Representational Principle itself – that is, to claim that experiences are not representational – but just the additional Mirroring Thesis, which is the heart of the attempt to explain phenomenology in terms of intentional content. “Don’t get me wrong,” he says: I think that sensations – almost always – perhaps even always – have representational content in addition to their phenomenal character … What I deny is that representational content is all there is to phenomenal character. I insist that phenomenal character outruns representational content. We can call this view “phenomenism”. Phenomenists believe in qualia. (Block 1996: 20) Qualia theory is thus often a component in what, in the previous chapter, we called “weak intentionalism”. However, as qualia themselves are nonrepresentational, at least on the traditional understanding of qualia, I treat qualia theory itself as incorporating a rejection of the Representational Principle. Even between theorists who would broadly fall into this camp, however, there are subtle differences on precisely how qualia and representational features contribute to perceptual experiences. In denying that representational content is all there is to phenomenal character, Block appears to suggest that representational content does contribute to, even if not fully explain, phenomenal character. Similarly, Christopher Peacocke defines the “sensational properties” of an experience as those properties of an experience that contribute to phenomenal character, yet are not “possess[ed] in virtue of features of the way the experience represents the world as being (its representational content)” (2008: 7), before insisting that any genuine perceptual experience must have some sensational properties. David Papineau, however, while having “no doubt that in the actual world all sensory experiences are representations” (2014: 20), insists that these representational features do not play a role in determining the phenomenal aspects of experiences. Using the terminology from Harman we met in the previous chapter, Papineau argues not only that experiences have “mental paint” properties, but that as far as the phenomenal aspects of the experiences are concerned, this is all they have: “our conscious experience is all paint, and any representational or represented features are quite external to our consciousness” (2016: 334). So while all three theorists accept that, as things are, most or all experiences are representational, they do differ over the question of whether or not the representational features of experience can contribute to (even if not fully explain) the phenomenal features of experience. This difference will also lead different theorists to have different views on the presentational and thus phenomenal characters of experiences. If we define qualia as intrinsic qualities of experience that we can become aware of when we attend to our experiences, then a pure qualia theory such as Papineau’s would therefore hold that all of the qualities sensed in experience – the constituents of the experience’s

88  Philosophical Theories

presentational character – would be properties of the experience: qualia. Moreover, as it is these properties that type the experience by what it is like to undergo it, on a pure qualia theory, both the phenomenal character and the presentational character would coincide. On a weak intentionalism that allowed both qualia and representational features to contribute to phenomenology, however, we would need a hybrid view. Where the representational features contribute to the phenomenology of an experience, the constituents of its presentational character would involve the properties that the experience represents as being instantiated alongside any properties of the experience – any qualia – that also make a phenomenological contribution. As an experience’s phenomenal character is the property of the experience that types the experience by what it is like to undergo it, this would also have a hybrid nature, involving both the experience’s qualia and its properties of representing that thusand-such properties are instantiated.

The Many Property Problem for Qualia Theory Interestingly, Michael Tye argues that our original many property problem also returns to haunt the qualia theory. Consider two experiences: one of a red square alongside a green triangle; the other of a red triangle beside a green square. As Tye points out, “each experience has the same color and shape qualia, Qr, Qs, Qg, and Qt” (2007: 310). Without some way of binding the right color qualia together with the right shape qualia, then a qualia theory will be lacking a key resource for expressing the difference between these two experiences. This connects to the binding problem that will be discussed in Chapter 9.

To try and show why qualia are indispensable in theories of perceptual experience, recent qualia theorists have attacked the attempt to explain phenomenal character wholly in terms of representational content. There are three main points of attack in the literature. The first attempts to identify cases of perception in which it is plausible to claim that experiences don’t represent anything at all; the second and third attack the two directions of the core Mirroring Thesis. As this thesis claims that there will be a difference in phenomenology or what it is like to have an experience if and only if there is a difference in the experience’s representational content, the qualia theorist typically proceeds by outlining ways in which phenomenal character and representational content are claimed to vary independently of one another. If successful, these attacks would show at the very least that not every aspect of an experience’s phenomenal character can be accommodated by appeal to its representational content, and hence that something else needs to be added in to the picture in order to account for the now-unexplained aspects of phenomenal character; qualia can then be offered as prime contenders to play this role. One of the reasons the transparency thesis can seem so plausible is that, in the case of vision, it can be difficult to think of any conscious visual experiences that

Adverbialism and Qualia Theories  89

do not serve to give us information about our environment. Nevertheless, some candidates have been identified. For example, Block (1996) discusses phosphene experiences – the strange sensations of points or rings of color or light that can be created by pressure on the eyeball or by direct electrical stimulation of the visual system – and afterimages – the strange visual “hangovers” we get, typically after seeing a bright light. In both cases, Block contends that it is a stretch to try and account for these cases in terms of representational content. Where afterimages are concerned, he contends they: look … illusory. … [W]hen one has an afterimage one has no tendency to think thereby that anything is really red, and so the introspective foundation for the theoretical claim that the afterimage represents something in the world as red is weak. (1996: 32) Likewise for phosphene experiences, he wonders, “What would the world have to be like for a phosphene experience to be veridical”? (1996: 35). As noted above, the arguments for qualia that focus on undermining the Mirroring Thesis typically fall into two camps: one that attacks the left-right (LR) reading of the Mirroring Thesis by arguing that phenomenology can differ while representational content remains the same, and one that attacks the right-left (RL) reading, contending that representational content can differ while phenomenology remains unchanged. Let’s explore these in turn. Recall that the left-right reading of the Mirroring Thesis, (M)LR, claims that any difference in phenomenology necessitates a difference in representational content. To challenge this, qualia theorists would need to present examples in which content remains the same while phenomenology differs. Some of the many examples that have been offered in this vein include the following: •

•

•

Binocular versus Monocular Experiences (Peacocke 1983): Look at the room you are in. This experience represents the layout of the room. Now close one eye. Due to using only one eye instead of two, this experience will be subtly phenomenally different. Yet it will still represent the layout of the room to be the same. Blurred Vision (Boghossian and Velleman 1989): If you remove your glasses (should you wear them – a similar effect can be achieved by unfocussing your eyes), you will probably find that your experience goes a little blurry. But the slight blurriness of your experience doesn’t seem to correspond to any changes in what your experience represents – after all, you don’t represent the objects to be blurry any more after removing your glasses than you did before. Ambiguous Figures (Macpherson 2006): Consider a regular diamond (a square rotated through 45 degrees around the center). It is possible to see this shape either as an upright diamond or as a rotated square, and there seems to be a phenomenal difference when we switch from seeing the shape as a diamond to

90  Philosophical Theories

•

•

seeing it as a square. Yet these two experiences don’t tell us anything different about the world, so don’t appear to differ in their contents. Shifting Attention (Chalmers 2004; Block 2010): Imagine looking at two red points of light against a black background and shifting attention from one light to the other. There is arguably a phenomenal difference between the  experience of attending to one and attending to the other, but it is not clear whether there is a representational difference between the two experiences. Inverted Spectrum (Block 1999): Suppose the phenomenological color you see when you look at blue things is the same as the color I see when I look at yellow things, but that what our experiences represent is fixed by the environment they give us information about. Now consider you and I looking at the blue sky. Intuitively our experiences differ in phenomenology but have the same representational content. This is known as the Inverted Spectrum hypothesis.

Representationalist Defenses of the Mirroring Thesis Of course, representationalists have developed responses to all of these kinds of scenario, and we cannot explore them all here. Instead, let’s consider a more general approach developed by Alex Byrne (2001). He asks us to imagine an idealized subject enjoying two consecutive experiences – e and e* – that differ in their phenomenology. Byrne argues that as there is a difference between the phenomenology of these experiences, then if an ideal subject were to experience e and e* in succession, the difference in phenomenology would mean that when the change takes place, the subject (being ideal) will notice this. As the subject notices that something has changed, it seems to follow that the way the way the world is presented to be has changed, which, in turn, suggests that the two experiences tell us different things about the world. But this is just to say that experiences e and e* represent the world to be different/have different contents. This argument allows us to conclude that if the phenomenology of an experience differs, then so will its content – the world will have to be slightly different in order for the experience to be accurate – which gives us (M)LR. Moreover, if two experiences differ in what they represent, they must therefore present the world to be slightly different  – in Byrne’s terms, the way things seem to the subject must be different – which is just to say that they differ in their phenomenology. This also gives us (M)RL. In this way, Byrne’s argument attempts to undercut the general principle that two experiences could differ in their phenomenology yet share the same representational content.

Adverbialism and Qualia Theories  91

(M)RL, however, claims that any difference in representational content necessitates a difference in phenomenology. As an objection to this claim, Ned Block (1990) develops a clever variant of the Inverted Spectrum hypothesis. Suppose that you are taken (in your sleep) to a new planet – Inverted Earth – in which objects have the inverse of their colors here on Earth. However, to ensure you don’t notice the difference, you are fitted with a pair of color inverting lenses. So when you wake up and see the yellow sky, you have an experience of blue, just as you did on Earth; likewise when you see the red grass, you have an experience of green. Block contends that over time, the representational contents of your experiences will change to fit your new surroundings (so to this extent, the success of this case requires adoption of an appropriate psychosemantics). But now the (blue) experience you have when you look at the sky will have the content , whereas on Earth it had the content . If so, then we have a case in which experiences with the same phenomenology (experiences of a clear sky on Earth and Inverted Earth) differ in representational content.

Qualia Theory and the Two Hats As we noted above, the scenarios we have just outlined – if successful – would serve to show that thinking in terms of representation alone leaves something out. But why should we think qualia – intrinsic, introspectable properties of experiences – are what needs to be added? Consider again the case of blurred vision introduced briefly above. As we saw, there is a strong sense in cases such as this that there is a significant and noticeable phenomenal difference between the regular and blurry experiences. Recall our discussions in the previous chapter concerning the claim that experience is transparent. There we saw Harman predicting that when you look at a tree, if you attempt to “turn your attention to intrinsic features of your experience … the only features there to turn your attention to will be features of the presented tree” (1990: 39). In this light, however, consider the case of viewing a tree and then taking off your glasses: you will still be aware of the tree, sure, but the blurriness that now characterizes your experience? Contra Harman that doesn’t obviously look to be a feature of the presented tree at all – the tree itself hasn’t started to look blurry; it’s just that our experience of the tree has become blurry. For this reason, Crane claims that this seems to be “a straightforward case of where one can be ‘directly’ aware of an aspect of one’s experience which is not an aspect of the objects of experience” (2006: 130). Moreover, this observation might suggest that a Byrne-inspired attempt to defuse this objection by identifying representational differences between the two experiences – such as Tye’s claim that blurry experiences fail to represent the sharp boundaries that objects really have (2003a: 20) or Allen’s suggestion that such experiences overrepresent the object’s boundaries at multiple places at once (2013: 267) – somewhat misses the point. As Amy Kind contends: “It still seems to us, when we are having a blurry experience, that we can focus on the blurriness itself, rather than on just what the blurriness is blurriness of ” (2008: 290).

92  Philosophical Theories

Kind goes on to argue that many of the other examples outlined above also undermine transparency, even if the representationalist does succeed in formulating a unique content that these experiences could possess. As Block argued above, if you press lightly on your eyeballs with the heels of your hands so as to induce a phosphene experience and then attend to the lights and colors, it just doesn’t seem to be the case that the features you are attending to are indeed features of some apparent external object. Likewise in the case of afterimages; if you stare at a green dot and then look at a plain white wall, you should find that you can now attend to the redness of the afterimage. Yet when you do, the redness you attend to does not seem to be a property of the wall or any other external object. In such cases, Kind suggests, the best explanation is that you are attending to properties of your experience: qualia. She also goes on to argue that reflection on these cases can undermine the transparency intuition, even in the kind of everyday cases that Harman describes: Look at a tree, focus on your experience, and then close your eyes and image the tree. Focus in on the greenness in your imaged experience. Now reopen your eyes, so that you’re looking at the tree. I predict that you will find features there, other than features of the presented tree, on which to train your attention. In particular, you can continue to attend to the greenness that you were attending to while your eyes were closed. (Kind 2008: 295–6) In this kind of way, the qualia theorist can argue that even if experiences have representational content, detailed phenomenological examination nonetheless suggests that this is not all there is to the phenomenology of such experiences. We can also be aware of the experiences themselves and some of their properties. So in addition to representational content, we also need to endorse qualia.

The Phenomenological Hat Given that qualia are introduced as intrinsic, consciously accessible properties that are responsible for an experience’s phenomenology, it may seem unnecessary to ask for more details about how a theory that postulates qualia wears the phenomenological hat. Indeed, many early arguments for qualia argued that they were required in our theories of experience due to the fact that other theories – functionalist and identity theories – missed out the crucial phenomenal features of conscious experiences, and qualia were then introduced to fill this gap. To this extent, then we should just accept that theories that appeal to qualia can wear the phenomenological hat. However, there is still an interesting question to be asked concerning what these properties are, such that they play this role. There are two main approaches in this space: the first springs from an antecedent commitment to physicalism or materialism. If we accept that qualia exist, but are committed to the view that nothing exists other than the physical or material, then the only option seems to be to hold that qualia are identical to certain physical properties and that we can identify

Adverbialism and Qualia Theories  93

which physical properties empirically. The concern with views of this sort is that they seem to leave a crucial question unanswered: we know qualia from the first person perspective and we are told that these intimately known properties are actually neural properties of some sort (there are numerous candidates in the literature); yet no matter what neural properties are held to be identical to qualia, we might wonder whether this is enough; qualia are introduced as the properties that are responsible for an experience’s phenomenology, but how could physical properties play this role? One feature of this general concern is expressed in the thought that any proposed identity claim of this kind will always seem somewhat brute: why that property? Why not a different one? Valerie Gray Hardcastle attempts to assuage this concern by drawing a parallel to other such a posteriori identity claims: A water-mysterian wonders why water has this peculiar property [of being wet]. She inquires and you give an explanation of the molecular composition of water and a brief story about the connection between micro-chemical properties and macro-phenomena. Ah, she says, I am a materialist, so I am convinced that you have properly correlated water with its underlying molecular composition. I also have no reason to doubt that your story about the macro-effects of chemical properties to be wrong. But I still am not satisfied, for you have left off in your explanation what I find to be most puzzling. Why is water H2O? Why couldn’t it be XYZ? Why couldn’t it have some other radically different chemical story behind it? I can imagine a possible world in which water has all the macro-properties that it has now, but is not composed of H2O.... [To] those who insist that they can imagine honest-to-goodness water not being H2O, what can one say? I think nothing. Water-mysterians are antecedently convinced of the mysteriousness of water and no amount of scientific data is going to change that perspective. (1996: 10) The appropriateness of this parallel can be challenged (see Chalmers 1997), but even if it were successful, it doesn’t succeed in dispelling the general mysteriousness concerning how these special introspectable phenomenal properties could in fact be neural properties of some kind. The second main theory of qualia, then, rejects the antecedent commitment to physicalism or materialism. David Chalmers argues that given the problems facing attempts to reduce qualia to physical properties, “a theory of consciousness should take experience as fundamental” (1995: 14). If Chalmers is right, to fully account for the phenomenological aspects of experiences, we have to add something new to our ontology in addition to the fundamental components of our physical ontology. The biggest concern with this approach, however, is that it threatens to render qualia epiphenomenal – that is, as being causally irrelevant to the things that happen in the physical world. Given that Chalmers does not want to reject the principle that everything that happens within the physical world can be fully explained from within the physical world, his view faces a problem with the fact that events in

94  Philosophical Theories

the physical world – such as the fact that he has written numerous books and articles about consciousness – can therefore be fully explained without appeal to the very properties of his experiences that are supposed to explain their being conscious: “it seems that consciousness is explanatorily irrelevant to our claims and judgments about consciousness. This result I call the paradox of phenomenal judgment” (1996: 177). What these brief discussions are intended to show is that, even though qualia are simply defined as those properties that are responsible for an experience’s phenomenology, the theory can still face broadly phenomenological challenges. In particular, once the theory says more about what kinds of properties qualia are, then we can see how challenges arise. If qualia are identified with physical properties, then we might wonder how these phenomenal properties that we know through experience could possibly be identical to physical properties, and this perplexity is then captured by the thought that such a hypothesis should do something to show why these physical properties are identical to qualia; but this seems to be something that simple identity claims cannot do. However, if qualia are held to be nonphysical properties, then even though we do not face the concern about how these properties could be qualia, we do face the problem of accounting for the apparent fact that qualia play a critical role in explaining our consciousness-related behavior.

The Epistemological Hat To the extent that qualia are used to supplement a weak intentionalism, the epistemological situation of the theory will be similar to that of the intentionalist theories we discussed in the previous chapter. Where a pure qualia theory is concerned, however, the epistemological situation of the view might look to be more like that of sense datum theory than that of adverbialism. To see why, consider Papineau’s claim that: our conscious sensory properties, the ones we are aware of when we introspect, are intrinsic properties of us, and metaphysically quite distinct from the properties of objects that successful sensory experience enables us to perceive. The ‘blueness’ that I know to be present when I introspect my sense experience is a property of me, not of the object out there. (2014: 26) This explanation looks very similar to that provided by the sense datum theory, just without the introduction of mental objects to bear the properties that we are aware of when we introspect. However, Papineau continues: Of course there is a perfectly good sense in which I perceive the external blueness of my shirt itself in the good case. But I so perceive that external blueness in virtue of having a different property of conscious sensory ‘blueness’. … I don’t at all mind saying that I am ‘aware of’ the shirt’s external colour in the good case. Of course I am. But in that sense I am not ‘aware of’ my conscious ‘blueness’. Indeed in a normal situation, where I am not introspecting, I deny

Adverbialism and Qualia Theories  95

that there is any sense at all in which I am aware of my conscious ‘blueness’. It is a conscious property all right, and I can turn my mind to it by introspecting, but when I am not introspecting, as I normally am not, I have no state of mind which is about that ‘blueness’. I just consciously have the ‘blueness’.” (2014: 26) This development suggests that my awareness of real-world blueness does not require an epistemically prior awareness of conscious “blueness” (although it does require me to “have” the property of conscious “blueness”). To this extent, the qualia theory looks less like sense datum theory minus mental objects and more like adverbialism – we become aware of real-world blueness by sensing in an appropriate way, where sensing in an appropriate way involves having the property of conscious “blueness”. So the epistemic position of the qualia theorist looks similar to that of the adverbialist we discussed earlier.

Phenomenal Intentionality Phenomenal Principle Common Factor Principle Representational Principle

✘ ✔ ✔

Sometimes, in outlining his pure qualia view, Papineau suggests that it sees “the consciously constituted experience as the vehicle of representation, rather than the content” (2016: 332). This suggests the possibility of a theory of visual experience that ties phenomenology to representational content in the kind of way the representationalist does, but which approaches the issues from the opposite direction: a view where the representational content of an experience is fixed by its phenomenology – by the qualia it possesses – rather than having its phenomenology fixed by its representational content. To explore this approach in more detail, let’s take the long way around, as this will also clarify the view’s connections to its adverbial ancestors. As we saw in the introduction to the chapter, Uriah Kriegel argues that to explain the apparent relationality of intentionality in a way that does not tangle us up in metaphysical problems, we should reject the claim that intentionality is constitutively relational and replace it with an adverbialist way of understanding intentionality: Although the surface grammar of “you are thinking of Bigfoot” casts the thinking as a relation between you and Bigfoot, the sentence can be paraphrased into “you are thinking Bigfootly,” or perhaps more naturally, “you are thinking Bigfoot-wise.” The latter casts the thinking as a non-relational property of yours. (Kriegel 2007: 314) As it stands, this view expands the application of adverbialism from the limited case of perception to the case of intentionality more generally. As far as the case of perception itself goes, however, this is just the traditional adverbialist picture (save

96  Philosophical Theories

for the fact that unlike the traditional adverbialist, Kriegel’s adverbialism allows for adverbial perceptual states to be representational). Importantly, it maintains the adverbialist’s rejection of a relational understanding of perception: although the surface grammar of “you are perceiving Bigfoot” casts the perceiving as a relation between you and Bigfoot, the sentence can be paraphrased into “you are perceiving Bigfootly” or perhaps more naturally into “you are perceiving Bigfoot-wise.” Recognizing this, Kriegel argues that the many property problem can be overcome by endorsing Jackson’s FG-ly response, along with the recognition that thoughts can be structured by a determinable-determinate relation. Consider the properties being a strawberry and being a berry: “the property of being a berry is a determinable of which the property of being a strawberry is a determinate” (Kriegel 2008: 89). If so, then anything that instantiates the first property (the determinate) thereby instantiates the second (the determinable) –anything that has the property of being a strawberry thereby has the property of being a berry. This is an instance of a general rule: something that instantiates a determinate property thereby instantiates all the determinables under which the determinate falls. What this enables us to do is to accommodate the inference that when we kick a strawberry, we therefore kick a berry. With this structure in mind, Kriegel argues that the property thinking green-dragon-wise is a determinate of the determinable thinking dragon-wise (and also, probably, of the determinable thinking green-wise), and thereby that when we think (or perceive) green-dragon-wise, we can draw the inferences that Jackson requires: we are thereby thinking (or perceiving) dragon-wise (and green-wise) (2008: 89). In fleshing out this position, Kriegel then asks the following question: if the apparently relational property perceiving Bigfoot is really the non-relational property perceiving Bigfoot-wise, then can we say anything else about this property? He answers this by suggesting that “there is a non-relational property that is a good candidate for constituting the property of representing Bigfoot-wise, namely, the representation’s phenomenal character” (2007: 317). The underlying idea is that, if there is something it is like to have an experience, then that alone suffices for the experience to be the kind of thing that could be accurate or inaccurate. Take the experience you are having now: it seems plausible to suppose that, given only what it is like for you, your experience is accurate if and only if there is a pink book (or suitably book-like object) in front of you (Siewert 1998). Kriegel also supplements this by appeal to the transparency intuition: the reason we only find the world when we attempt to introspect our experiences, he suggests, is because phenomenal character is inherently intentional (2007: 320–1; see also Horgan and Tienson 2002: 521; Chalmers 2004: 370).

The Determinable-Determinate Relation and the Many Property Problem One objection to this approach, as argued by Alex Grzankowski (2018), is that Kriegel’s invocation of the determinable-determinate relation may create a new problem for this view. Typically, alongside the rule that something instantiating a determinate property instantiates all the determinables under

Adverbialism and Qualia Theories  97

which it falls, there is a further rule: that a determinable property cannot be instantiated without one of its determinates also being instantiated. In the strawberry case, this tells us that it is not possible to simply be a berry without being some particular kind of berry. But this doesn’t seem to be the case for intentional states, as Kriegel himself recognizes: one can think of a parrot without thinking either of a blue-eyed parrot or of a brown-eyed parrot [because] thinking of a parrot is a matter of having a thought that represents parrot-wise, and a thought can represent parrot-wise without representing either blue-eyed-parrot-wise or brown-eyed-parrot-wise. (2007: 315) Yet if thinking blue-eyed-parrot-wise is a determinate of the determinable thinking parrot-wise, as it would need to be to licence the inference from my thinking blue-eyed-parrot-wise to my thinking parrot-wise, then given this rule, it would look to be the case that it should not be possible to merely think parrot-wise without thinking about a parrot with particular colored eyes.

Above, we discussed Alex Byrne’s argument that aimed to show that once we accept that there is a difference in phenomenology between two experiences, it follows that there must be a difference in the way things seem to the subject or, alternatively, the way the world is presented to be. If successful, this argument would allow us to insist that wherever there is a phenomenal difference, this will be accompanied by a representational difference. Now in our discussions above, we saw that this alone may not be enough for the representationalist – there may still be reasons to think that even if every experience has a unique content, we can nonetheless attend to properties of our experiences, which gives us reason to reject the core representationalist thesis. In this context, however, Byrne’s argument can provide a way of showing that every experience with phenomenology will attempt to tell us something – that is that, for any experience that possesses phenomenal character, it will also possess representational content in virtue of this phenomenal character. This view – phenomenal intentionality – is thus an inversion of the representationalist theory we met in the previous chapter. That view, recall, accepted the Mirroring Thesis – that there can be a difference in the phenomenology of an experience if and only if there is a difference in the experience’s representational content – and argued that this is explained by the fact that phenomenology is really grounded in representation. This alternative also accepts the Mirroring Thesis, but contends that it is explained by the fact that intentionality is really grounded in phenomenal character.

98  Philosophical Theories

The claim is that such conscious states have an intentional content which they carry purely in virtue of their phenomenal character. They may have intentional properties unrelated to their phenomenal character, but they also have some intentional properties that are instantiated in virtue of, indeed are constituted by, their phenomenal properties. When a mental state M has an intentional content that is constituted by its phenomenal character, we may say that M exhibits a phenomenal intentionality, or is phenomenally intentional. (Kriegel 2007: 320) Now, given that the phenomenal intentionalist gives explanatory priority to an experience’s phenomenology in this way, they allow us to explain what it is for an experience to have content by appealing to its phenomenology. But there is still a need to provide a theory of the phenomenal features of visual experiences. A qualia theory, or an adverbial theory, would therefore seem to be a natural partner of the phenomenal intentionalist claim. Yet given that both adverbialism and qualia theory have been introduced as rejecting the Representational Principle, this might look to be inconsistent with these properties being the source or ground of intentionality. To this extent, then, the phenomenal intentionalist’s understanding of qualia is importantly different to the traditional understanding. As Angela Mendelovici argues, “at least some phenomenal states are inherently intentional. They are not ‘raw feels,’ mere feelings or qualia. Instead, by their very nature, they automatically represent a content or ‘say something’”. As Loar (2003, §12) puts it, phenomenal character is not mere mental paint; it is, by its nature, “paint that points” (2018: 84). So whether or not we use the term “qualia”, the key idea is that experiences have intrinsic properties that determine their phenomenal character. For the phenomenal intentionalist, these properties will also determine the experience’s representational properties too. As theorists of this inclination are typically inclined to view qualia as determined narrowly, supervening only on the internal states of the subject, this means that any representational content that is fixed in this way will also be narrow – that is, shared by phenomenal twins who inhabit different environments. For this reason, many phenomenal intentionalists allow for experiences to have wide contents in addition to the narrow phenomenally determined contents, although the majority of them think that the intentionality of these additional contents is in some way derived from phenomenal intentionality (e.g. Loar 2003; Horgan et al. 2004). Others, such as Farkas (2008) and Mendelovici (2018), take a stronger line, arguing that there is no intentionality that is not phenomenal intentionality. When it comes to the question of how well phenomenal intentionalism can wear the two hats, the situation with respect to the phenomenological hat seems similar to that of the traditional qualia theory. When it comes to the epistemological hat, however, Kriegel argues that despite the theory’s refusal to understand intentionality as fundamentally relational, intentional states thus understood will connect us to the world “contingently rather than constitutively. … An accurate … experience as of trumpets connects us to trumpets, not in virtue of being an experience as of trumpets however, but in virtue of being accurate” (2011a: 165). On this approach,

Adverbialism and Qualia Theories  99

a state’s being intentional need not itself put us in contact with the world – after all, non-veridical states are intentional, yet fail to put us in contact with the world – but must make it possible for us to be in contact. And, Kriegel suggests, they do so by providing accuracy conditions that when met, ensure that we are in contact with the world.

Questions • • •

Is the adverbialist better placed than the sense datum theorist when it comes to wearing the epistemological hat? How compelling do you find the phenomenological arguments for the introduction of qualia? If the phenomenal intentionalist is correct and the intentionality of perceptual experiences is grounded in their phenomenal character, how plausible do you think it will be to extend this treatment to all intentional states?

Further Reading Uriah Kriegel’s defense of adverbialism as a way of fixing a problem for representationalism can be found in his papers “Intentional Inexistence and Phenomenal Intentionality” (2007), “The dispensability of (merely) intentional objects” (2008), and in Chapter 3 of his 2011 book The Sources of Intentionality. For a good outline of the original adverbialist approach, together with the original development of the many property problem, see Frank Jackson’s 1975 paper, “On the Adverbial Analysis of Visual Experience”. Color adverbialism is developed in Outside Color by Mazviita Chirimuuta (2015) and Primitive Colors by Joshua Gert (2017). Tim Crane discusses the connections between adverbialism and qualia theory in his 2000 paper, “The origins of qualia”. Recent defenses of the view include Amy Kind (2008) “How to Believe in Qualia”, Ned Block (2010) “Attention and Mental Paint”, and David Papineau (2014) “Sensory Experience and Representational Properties”. For phenomenal intentionality, see the Kriegel papers mentioned above. For fullstrength versions that attempt to ground all representation in phenomenology, see Katalin Farkas’s 2008 paper “Phenomenal Intentionality Without Compromise” and Angela Mendelovici’s 2018 book The Phenomenal Basis of Intentionality.

Taylor & Francis Taylor & Francis Group

http://taylorandfrancis.com

5

Naïve Realism

Overview Naïve realist theories of perception endorse the Phenomenal Principle, but hold that the objects we are aware of in experience are everyday material objects. For this reason, they also reject the Common Factor Principle and endorse disjunctivism: the view that indiscriminable experiences can nevertheless be experiences of different kinds. This places an onus on the naïve realist to provide theories of hallucination and illusion that explain how these states can be indiscriminable from veridical perceptions yet not “screen off” those perceptions from playing an explanatory role. Different attempts to meet this demand are considered, before exploring the claim that disjunctivism, and hence naïve realism, is inconsistent with visual science. In our discussion of sense datum theories in Chapter 2, we saw that theories that endorse mental objects are the result of two key commitments. The first of these was a commitment to the Phenomenal Principle: that in order to do justice to what it is like to have an experience as of an object possessing a certain quality, then there must be an object instantiating that property that we are aware of. As we noted in that chapter, in cases where we see such an object, we might think that the mindindependent external object itself is the object that we are immediately aware of. Yet the second commitment – to the Common Factor Principle that phenomenologically indiscriminable perceptions, hallucinations, and illusions have an underlying mental state in common – rules this out. Once we recognize that the possibility of indiscriminable illusions and hallucinations means that external objects cannot play this role for all visual experiences as of objects possessing certain qualities, this has the consequence that our fundamental analysis of perceptual experience cannot appeal to external objects to play this role, which, in turn, leads to the introduction of mental objects. As it is the acceptance of these two principles together that requires the introduction of mental objects, in Chapters 3 and 4, we explored a range of theories that

102  Philosophical Theories

attempted to avoid the appeal to mental objects by rejecting one of the two principles. For these theories, the principle to go was the Phenomenal Principle. This therefore placed the onus on these theories to provide an alternative explanation of the phenomenology of visual experiences. A different approach, however, is to go in the opposite direction: to continue to endorse the Phenomenal Principle, at least for the limited case of perception, and to reject the Common Factor Principle. This yields a philosophical theory of the good cases of perception known as Naïve Realism or Relationalism.

Naïve Realism and the Representational Principle Typically, naïve realists also reject the Representational Principle on the grounds that perceptions do not represent things to be a certain way; they simply present them as being as they are. There is therefore no question of experiences being incorrect, as the Representational Principle would require. As Austin puts it, “our senses are dumb—though Descartes and others speak of the ‘testimony of the senses’, our senses do not tell us anything, true or false” (1962: 11; see also Travis 2004).

Naïve Realism Phenomenal Principle Common Factor Principle Representational Principle

✔ ✘ ✘

By endorsing the Phenomenal Principle whilst rejecting the Common Factor Principle, the naïve realist can agree with the sense datum theorist that “the most fundamental characterization of perceptual experience is to be given in terms of a relation of conscious acquaintance with certain direct objects of perception”, yet nonetheless insist that “this insight is perfectly consistent with the thesis that the direct objects of perception are the persisting mind-independent physical objects that we all know and love” (Brewer 2011: 93). The word “fundamental” is important here – all the theories of perception we have been considering allow that, in perception, we are related to mind-independent physical objects. The difference is that in those cases, our being related to physical objects can be reduced to our having a perceptual experience of a certain kind – be that being acquainted with a certain sense datum, sensing in a certain way, having an experience with certain qualia, or being in a state with a certain representational content – plus some other factors. For the naïve realist, however, this is not the case: saying that, in perception, the subject is acquainted with a mind-independent physical object is to give the most fundamental characterization we can of the perceptual relationship (Logue 2012b: 211). We might wonder why the view is qualified by the adjective “naïve” – what is it about the view that makes it naïve in this way? Well, this name has a long history.

Naïve Realism  103

Hume, for example, describes something very like this view as one that people are almost compelled to adopt without thought or consideration. He tells us that: men are carried, by a natural instinct or prepossession, to repose faith in their senses …. [W]hen men follow this blind and powerful instinct of nature, they always suppose the very images, presented by the senses, to be the external objects. (1748/2011: 133) So for Hume, the view that we were acquainted with external objects in perception was held to be extremely natural, yet nonetheless “soon destroyed by the slightest philosophy”. The view could therefore be considered naïve, both in the sense of not requiring any kind of specialist knowledge for its formulation, but also in the sense of not being able to withstand any kind of philosophical scrutiny.

Naïve and Direct Realism On this understanding, naïve realism is an example of, but not the only example of, direct realism. The direct realist insists that when we perceive an external object, our perception of that object is not mediated in the kind of way that indirect realists, such as sense datum theorists, claim. On this understanding, adverbialists, representationalists, and (possibly) qualia theorists would all qualify as direct realist theories. What is distinctive about a naïve form of direct realism is the claim that the direct objects of perception are actually involved in the most fundamental analysis of the experience, typically by being constituents of the experience’s phenomenology. This has the consequence that the fundamental kind of experience a subject has when they perceive a certain object would not be available if the experience were of different objects or did not involve objects at all (see Genone 2016 for more discussion of this issue).

Today, of course, philosophers who hold broadly naïve realist views tend to have less naïve reasons for doing so. John Campbell (2002), for instance, argues that only this kind of view can explain our capacity to have thoughts about mindindependent objects, M.G.F. Martin (2002) argues that naïve realism is required to explain the phenomenology of sensory imagination, and William Fish (2009) argues that naïve realism can allow us to close the explanatory gap between the facts about phenomenal consciousness and the physical/functional facts. Others have more epistemic motivations: Mark Johnston (2006, and possibly John McDowell 1986/1998) argues that a broadly naïve realist view can provide a way to respond to skepticism, and Heather Logue (2012b) argues that only naïve realism can explain how the phenomenal character of veridical experience puts its subject in a position to know what the objects in the environment are like independently of experience.

104  Philosophical Theories

Despite this, however, phenomenological motivations for naïve realism do still loom large. For example, Benj Hellie (2007) develops an argument for naïve realism on the grounds that many philosophers have made naïve realist-sounding claims about how experiences seem after initial reflection on what it is like to have them. He then argues that these judgments qualify as judgments made by experts in the subject matter under ideal conditions for making such judgments and concludes that this provides at least a prima facie reason to think that these judgments will be accurate. More recently, Harold Langsam has argued that naïve realism really is the theory of the naïve and that “the only way to make sense of the phenomenal character of a perceptual experience is in terms of a naïve realist ontology” (2017: 117). He contends that the fact that there is something it is like for a subject to be in a state with phenomenal character puts the subject in a position both to know that they are in a state with phenomenal character and to know that they are in a state that instantiates particular phenomenal properties (and that this is the case even if the subject is naïve and doesn’t have the concepts required to articulate what they knew in these terms). When we then reflect on the phenomenal properties that make up the phenomenal character of an experience of a tomato, say: reflection suggests that certain features of the tomato are in some sense constituents of the phenomenal character of my experience …. Here I am, experiencing the tomato, and part of what it is like for me to do so is a matter of having the redness of the tomato before me and revealing its nature to me, so to speak. (2017: 110–1) The only way this could be the case, he suggests, is if the tomato itself were a constituent of the experience as the naïve realist contends. As Langsam’s argument indicates, the core of naïve realism is the view that features of our external environment can literally be constituents of an experience’s phenomenology or, as Martin puts it, that in the good cases of perception, external objects and their properties “shape the contours of the subject’s conscious experience” (Martin 2004: 64). We can capture this commitment by employing our terminology of presentational character. At a first pass, the suggestion is that the presentational character of a perception is actually constituted by the piece of mind independent reality that is being looked at. Such a view of presentational character enables the naive realist to claim that external reality is literally experientially present. To the extent that presentational character is constituted by external properties, the naïve realist’s take on presentational character is actually quite similar to the intentionalist’s. What about the phenomenal character of a perception? What does the naïve realist have to say about the nature of the property of the experience that types it by what it is like to undergo it? Here, naïve realists depart from representationalists and instead go along with the sense datum theorists. Naïve realists hold that the phenomenal character of a perception is its property of being a sensing of the elements of the presentational character, or its property of acquainting the subject with the elements of the presentational character. This claim does raise the question of what is involved in a subject’s sensing or being acquainted with external reality. This is a critical part of a naive realist position, but

Naïve Realism  105

without further explanation it can seem to be almost magical. For example Paul Coates complains that “it is totally unclear … what the metaphysical relation of intrinsic connectedness could be” (2007: 75). Yet for the reasons given above, many theorists maintain that this relation cannot be given a deeper analysis, claiming that acquaintance is “simple” (Campbell 2002: 114), “sui generis and unanalyzable” (Fumerton 2006: 63), or “irreducible” (Fish 2009: 14). Although this may seem concerning, it is worth noting that the naïve realist is not the only one to face this problem: sense datum theories also employ an acquaintance relation and some qualia theorists use the notion to describe the relation we stand in to qualia (e.g. Chalmers 1996: 197). So to the extent that naïve realism might seem to miss something out by not providing a deeper analysis of acquaintance, then the same could be said of a number of other theories of perception (Hobson 2013: 557). Moreover, even if a philosophical analysis of acquaintance is not forthcoming, this does not entail that we can’t say anything interesting about it. Fish (2009: 75–9), for example, hints at the idea that a functionalist story could be told that entails that a suitable subject – one undergoing the kinds of cognitive and behavioral functions that underpin our ability to have experiences – would have to be acquainted with aspects of its environment. Alva Noë goes even further and provides a broadly naturalistic explanation of what is involved in being acquainted with the world. This theory – “actionism” – begins from the start-point that “visual experience [is] an active form of engagement with the environment” (2001: 50). Noë rejects representationalism, arguing that we are not consciously aware of what the brain represents, and claims instead that we are rather consciously aware of what is made available by the various processes going on in the brain and body. As it is aspects of the creature’s environment that are made available by these processes, Noë describes this theory as being “as naïve realist as one can hope to be” (2008: 702). How are parts of the world made available? By our having the ability to access the world using what he calls sensorimotor skills, where to possess a sensorimotor skill of a certain kind is to have a tacit understanding of how sensory stimuli change as a result of active movement. To use an example of Noë’s, we have access to – we see – the (whole) cat even though part of it is occluded behind a picket fence because we implicitly know that were we to move our bodies in this particular way, some bits of the cat would come into view whilst others would become occluded by the fence and so on. On the basis of these sensorimotor skills, “objects and properties are available to us; they can show up in perceptual consciousness” (2008: 702). In developing this suggestion, Noë claims that “you visually experience parts of the tomato that, strictly speaking, you do not see, because you understand, implicitly, that your sensory relation to those parts is mediated by familiar patterns of sensorimotor dependence” (2005: 77). Yet the appeal in this passage to the idea that some things can be seen “strictly speaking” is in fact a ladder that can be kicked away. To explain: Noë’s claim that we can still visually experience the back of a tomato even if we don’t see it “strictly speaking” can make it look as though his theory has two stages – one that explains what seeing “strictly speaking” amounts to and a second that explains how seeing “in the full sense” is built on this basis. If this were the right reading of Noë’s project, then one might see the appeal to sensorimotor contingencies as relating to the second stage of the project, whilst the

106  Philosophical Theories

kind of traditional theories of perception we have been considering can be held up as responses to the first stage. On this way of looking at things, Noë’s theory would not actually be in competition with the more traditional theories; instead, it would be a supplement to them – a supplement that explains how, from the platform of such a theory, we can rightly be said to be aware of whole objects. Yet whilst this discussion makes its point by playing with the idea that we do see the front of the tomato “strictly speaking”, in fact this is not the case: “perceptual content is virtual all the way in” (2005: 193). This is the key point: nothing is ever given to consciousness in its totality, so nothing is ever seen “strictly speaking”; instead, everything that we see is seen in virtue of our possessing and employing sensorimotor skills. As Noë insists that it is worldly objects and properties (including aspects and appearances) that are seen, then if this reading is correct, he can usefully be seen as endorsing naïve realism, broadly speaking, together with a detailed, action-based theory of what it is to stand in the acquaintance relation to external reality.

Naïve Realism and Disjunctivism So the naïve realist accepts the Phenomenal Principle for the case of perception, claiming that the objects that bear the properties that we are aware in such cases are the familiar, mind-independent objects that inhabit our environment. If we accept the reasoning from Chapter 2, however, which argued that this approach is unavailable in cases of hallucination, then this commitment has the consequence that the naïve realist must reject the Common Factor Principle. This means that the naïve realist denies that phenomenologically indiscriminable perceptions, hallucinations, and illusions have an underlying mental state in common. To this end, the naïve realist will also need to endorse an approach known as disjunctivism, the core idea of which is the claim that the mental states involved in a “good case” experience of perception and an indistinguishable “bad case” experience of hallucination are different, even in those cases in which the two experiences are indiscriminable for their subject.

Perceptual Theories of Hallucination Having said this, there has been a strand in discussions of naïve realism/ relationalism that has considered the prospects of a broadly perceptual theory of hallucination, and thereby reestablishing the Common Factor principle. For example, Alston suggests that when Macbeth hallucinates a dagger, we might conceive of that as a misperception of “the portion of space apparently occupied by the dagger” (1999: 191). While Alston concluded that this approach would not be available for all cases of hallucination – suggesting that “total” hallucinations, in which every aspect of an experience is hallucinatory, will be outside its scope – recent attempts to extend this approach to accommodate such cases can be found in the work of Raleigh (2014) and Ali (2018).

Naïve Realism  107

Why “disjunctivism”? We can see the answer to this question when we consider the disjunctivist’s analysis of the neutral category of visual experiences. Putting illusions to one side for a moment (we will come back to them shortly), the disjunctivist proposes to analyze visual experiences as follows: A subject S has a visual experience as of a property F, if and only if: • EITHER S perceives an F • OR S has a hallucination of an F. To see how this is supposed to work, consider the following example due to William Child (1994: 145). Suppose that there are only two ways of producing a “likeness” of subject S – by either drawing S or photographing S. The resultant objects – the photograph of S and the drawing of S – both qualify as likenesses of S. Yet whilst there is something common to both photographs of S and drawings of S inasmuch as they are both likenesses of S, this is not because they have some common property in virtue of which they both qualify as likenesses of S. Instead, the right way of understanding the category “likeness of S” is as disjunctive: something is a likeness of S if and only if it is either a drawing of S or a photograph of S. Indeed, we can present an analysis of the category of “likenesses” that parallels the analysis of the category of visual experience as follows: An object O is a likeness of S, if and only if: • EITHER O is a drawing of S • OR O is a photograph of S. For the disjunctivist, the category of visual experiences (of F) is like the category of likenesses (of S): it is fundamentally disjunctive. This way of understanding the neutral category of visual experience enables the disjunctivist to offer an interesting reinterpretation of seems-statements. Recall that, in the opening chapter, we discussed a motivation for the Common Factor Principle that turned on the fact that if we did not know whether we were seeing or hallucinating a pink elephant, we might naturally say that we are having the experience of seeming to see a pink elephant. On the face of it, such a statement can appear to commit us to a good case/bad case common factor: something that could occur in both a case of perception and a case of hallucination. However, the disjunctivist will follow J.M. Hinton (1967, 1973) and argue that a statement such as “I seem to see a pink elephant” is just “a more compact way of saying” something like this: “Either I see a pink elephant, or I am having a hallucination of a pink elephant”, where having a hallucination of a pink elephant is understood simply as being in a state that is not a seeing of a pink elephant, but that I cannot distinguish from such a seeing. Once this translation of the seems-statement has been provided, there is no longer any pressure to think that the truth of “I seem to see a pink elephant” commits us to the presence of a good case/bad case common factor.

108  Philosophical Theories

Where the statement “Either I see a pink elephant, or I am having a hallucination of a pink elephant” is concerned, it is natural to see there being two different ways in which this disjunctive sentence can be made true – either by its being true that I actually do see a pink elephant or by its being true that I don’t see a pink elephant, but that it is for me as if I did. The core disjunctive claim is therefore that: We should understand statements about how things appear to a perceiver to be equivalent to a disjunction that either one is perceiving such and such or one is suffering a … hallucination; and that such statements are not to be viewed as introducing a report of a distinctive mental event or state common to these various disjoint situations. (Martin 2004: 37) So, disjunctive theories hold that perceptions and hallucinations of a particular kind do not have an underlying mental state in common. In the literature, we can find a number of different ways in which the core disjunctive idea – that indiscriminable perceptions and hallucinations are different mental states – has been developed. However, not all of these varieties of disjunctivism are sufficient to support naïve realism, as it has been laid out above. For example, if we focus on the naïve realist idea that perceptual experiences are fundamentally a relation between a perceiver and a mind-independent object, then if the existence of a relation entails the existence of its relata, the existence of a perceptual experience would entail the existence of its mind-independent object. We might attempt to capture this on a metaphysical level with the claim that mind-independent objects are constituents of successful perceptions of those objects. If we then coupled this with the individuative principle that two mental states qualify as distinct so long as they have different constituents (or perhaps different supervenience bases), then this view would be a version of disjunctivism. However, whilst such a view would accommodate the key naïve realist idea that that “no experience like this, no experience of fundamentally the same kind, could have occurred had no appropriate candidate for awareness existed” (Martin 2004: 39), it would still be compatible with other significant similarities between the experiences. In particular, in and of itself, a purely metaphysical disjunctivism of this kind would be compatible with the claim that these other significant similarities fix the experience’s phenomenology. So, although metaphysical disjunctivism could give us naïve realism at the level of the fundamental constituents of mental states, on its own it wouldn’t give us naïve realism at the level of external objects as constituents of the phenomenal character of perceptual experiences. Another possible form of disjunctivism would be based on the individuative principle that two mental states qualify as distinct so long as they have different intentional contents. For example, if we were inclined toward the view that perceptual experiences were representational and had contents that were singular for objects as well as properties, then we could hold that if two indiscriminable experiences had different contents, then they would qualify as fundamentally different kinds of experience. This would not only give us a disjunctivism at the level of content, but would also enable us to accommodate the phenomenon of particularity – that our

Naïve Realism  109

visual experiences appear to be of a unique, individual object – and to thereby capture the naïve realist insight that perceptual experiences have their external objects as constituents on the grounds that they are constituents of the experiences’ contents. As we saw in the previous chapter, a view with this general shape is defended by Susanna Schellenberg (2013, 2016, 2018). Recall that, on Schellenberg’s view, the perceptual capacities deployed in experience fix a content type for that experience, and this content type, in turn, fixes the experience’s phenomenal character. However, when perception is successful, these perceptual capacities single out a unique particular, in virtue of which this particular becomes a constituent of the content token. In a case of illusion or hallucination, however, this process fails to pick out a unique particular, and the resulting content is “gappy” – it has a gap in the place where the successful perception’s content has a particular. For the reasons given above, such a view could easily qualify as a version of disjunctivism, so long as we held that if two experiences had different contents, then they would qualify as fundamentally different kinds of experience. As a veridical perception and its corresponding hallucination have different contents – one having a particular where the other has a gap – then these two experiences would qualify as experiences of different fundamental kinds. However, Schellenberg herself rejects this approach, claiming that she sees her view as non-disjunctive, on the grounds that “subjectively indistinguishable experiences share a metaphysically substantial common element” (2016: 50). Moreover, this metaphysically substantial common element – the common content type that is fixed by the capacities deployed – fixes the experience’s phenomenology. To this extent, a Schellenberg-style view detaches questions about the singular aspects of perceptual content from questions about perceptual phenomenology. On such a view, it would be appropriate to say that indiscriminable perceptions and hallucinations share phenomenology despite differing in content. Such a view therefore treats phenomenology as a common factor between perception and hallucination, and would therefore appear to be unable to accommodate the naïve realist claim that external objects are constituents of perceptual phenomenology. To capture this central claim, then it appears that the naïve realist will need to adopt a disjunctivism at the level of phenomenology itself, combined with the individuative principle that two mental states qualify as mental states of different kinds so long as they have different phenomenal character. According to this variant of disjunctivism, perceptions and hallucinations do not share phenomenal character. This leaves it open to claim that in the good cases, the perceived objects are actually constituents of the experience’s phenomenology.

Epistemological Disjunctivism There could also be a purely epistemological disjunctivism, which held simply that perceptions and hallucinations differ in their status as perceptual evidence (Snowdon 2005). In other words, the epistemological disjunctivist would deny that one’s perceptual evidence is the same across indiscriminable

110  Philosophical Theories

cases of perception and hallucination. However, as Snowdon points out, if this were to amount to no more than the claim that “a single basic sort of (inner) experience [has] quite different epistemological significance in different cases” (2005: 140), then this view would be compatible with the claim that indiscriminable perceptions and hallucinations do have an underlying mental state in common after all. It would therefore not only be insufficient to support naïve realism, it would also fail to even qualify as a type of disjunctivism as we are using the term. Such a position can seem strange, however: perceptions and hallucinations are accepted to be indiscriminable. Given this, it would seem that what it is like to have both experiences would be the same. Doesn’t this just entail that they have the same phenomenal character? Clearly, a disjunctivist about phenomenology cannot accept this claim: indiscriminability cannot entail sameness of phenomenal character. However, this does put the onus on the naïve realist to explain and defend the contention that experiences with different phenomenal characters can be indiscriminable. Mike Martin argues that to simply assume that indiscriminability has to be accounted for by the experience’s sharing phenomenal character is to attribute to subjects unreasonably substantive epistemic powers. Consider a perception of an F and a perfectly indiscriminable hallucination of such. In virtue of what do both count as visual experiences as of an F? A modest answer to this question is that this is enough: something is a visual experience of an F just in case it is indiscriminable from a perception of an F. If indiscriminability is assumed to be a matter of sameness of phenomenal character, however, then: when I come to recognize the possibility of perfect hallucination just like my current perception, what I do is both recognize the presence of [the phenomenal character] … in virtue of which this event is … an experience [of an F], and also recognize that an event’s possessing these characteristics is independent of whether the event is a perception or not. (2004: 47; see Siegel 2004 for discussion) But what about the possibility of “a situation in which [phenomenal character is different or] absent but in which a subject would be unable to discriminate through reflection this situation from one in which [an F] was really being seen” (2004: 49). The modest conception of what is required for an event to qualify as visual experience would allow us to count such an event as an experience (as) of an F simply in virtue of this failure to discriminate. The alternative conception, however, could not count this as a visual experience. In order to rule out the possibility of such scenarios, the defender of the idea that indiscriminability entails sameness of phenomenal character will have to assume that a careful subject simply cannot fail to recognize the presence and nature of phenomenal character when it is present or the absence of phenomenal character when it is absent. This approach therefore has to “attribute to responsible subjects potential infallibility about the course of their experiences” (2004: 51).

Naïve Realism  111

William Fish (2013) also addresses this concern by looking at the tendency to move from the claim that two experiences are indiscriminable, to the claim that what it is like to have these two experiences is the same, and then to the claim that these experiences have the same phenomenology/phenomenal character. He suggests that there is a certain ambiguity to our talk about “what it is like” to have an experience: sometimes, when we say there is something it is like to have an experience, we are saying that the experience has a certain robust property that accounts for this – a phenomenal property or phenomenal character; at other times, however, he suggests we might be talking more generally about how things are from the subject’s point of view when they have the experience. He uses a locution borrowed from Susanna Siegel (2008: 213) – “felt reality” – to capture this sense of there being something it is like. He suggests that one way in which a state can contribute to how things are from the subject’s point of view (to there being something it is like in the second sense) is by possessing phenomenal character (there being something it is like in the first sense), but that this is not the only way. Against this background, he goes on to suggest that the apparently plausible slide from indiscriminability to sameness of phenomenal character actually turns on an equivocation on “what it is like”. The first stage – the move from indiscriminability to sameness of what it is like – makes use of the second, how things are from the subject’s point of view, sense of what it’s like, whereas the second stage – from sameness of what it is like to sameness of phenomenal character – makes use of the first, robust sense. To this extent, he suggests, this move is not as plausible as it initially seems to be.

Naïve Realism and the Two Hats The Phenomenological Hat As we have seen, one of the most significant motivations for endorsing naïve realism is the thought that it is the best way of doing justice to the phenomenology of perceptual experience. Yet naïve realism does face some phenomenological objections. For instance, consider the first pass claim that on a naïve realist view of perception, the presentational character of a perception is constituted by the scene that the subject is looking at. However, as Hume recognized, appealing only to the perceived objects and their properties is not enough. As he noted, “the table, which we see, seems to diminish, as we remove farther from it; but the real table, which exists independent of us, suffers no alteration” (1748/2011: 134). If the table alone supplies the phenomenal character of the experience, then as there is no change in the table, there should be no change in the phenomenal character. Yet there is. Contemporary naïve realists are alive to this concern, however; as John Campbell puts it: [T]he phenomenal character of your experience, as you look around the room, is constituted by the actual layout of the room itself: which particular objects are there, their intrinsic properties, such as colour and shape, and how they are arranged in relation to one another and to you. (2002: 116)

112  Philosophical Theories

As this quote indicates, to fully account for the contours of the subject’s conscious experience and accommodate the fact that our experiences of objects can change without any change in the objects themselves, most naïve realists therefore take the perceiver’s perspective on the objects (together with other features of the environment, such as the lighting conditions) as also being involved in shaping the subject’s conscious experience. Others have argued that this is not enough, however. For example, Quassim Cassam argues that blurred vision creates similar problems for naïve realists to those that it created for representationalists: two people have their eyes tested by visiting the same optician and looking at the same eye chart from exactly the same postion. To one of them, the fourth row in the chart is crystal clear. To the other, the letters look blurry. (2014: 110) Cassam argues that there is a clear difference between the characters of the two subjects’ experiences, but as both the chart and the perspective on the chart are stipulated to be identical, the experiential difference cannot be grounded in differences in either the environment or the subject’s perspective on that environment. Cases such as this suggest that one and the same scene can be seen in different ways by different subjects (or indeed, by the same subject at different times) and that by attempting to ground phenomenal character solely in the presented environment, naïve realism is left without the resources to account for this. Another inter-subjective example includes the surprising finding that different perceivers, all of whom qualify as “normal”, will differ over precisely which surfaces are perceived as being a certain color. For example, as we shall explore in more detail in Chapter 7, perceiver A might claim that a surface that reflects light of 500 nanometers (nm) is unique green – green with no blue and no yellow component – whereas a surface that reflects light of 490 nm is a slightly bluish green; perceiver B, however, in the same observation conditions, might see the second (490 nm reflecting) surface as unique green and the first (500 nm reflecting) surface as a yellowish green! An intra-subjective version of this puzzle can be found when we consider the role of attention. Macpherson (2006) and Block (2010) discuss a range of scenarios in which a subject’s distribution of attention to a scene yields a phenomenal difference. In some cases, this could be explained as attention to (and hence acquaintance with) different aspects of the scene, but other cases cannot be dealt with in this way. For example, Block discusses experiments by Carrasco et al. (2004) that suggest attending to a grid makes that grid look darker, or higher contrast. All of these cases describe scenarios in which one and the same surface appears to be seen in phenomenally different ways. Given that the naïve realist contends that the phenomenal characters of our perceptual experiences are grounded in the perceived environment, such phenomena raise challenges for the naïve realist to address. In tackling cases such as this, naïve realists appear to have three broad possible responses available: they can deny that there is a phenomenal difference in such cases, attempt to locate the phenomenal differences in the aspects of the world the

Naïve Realism  113

subjects are acquainted with, or explain the phenomenal differences as involving a subjective addition to experience (albeit in such a way that the core naïve realist aspects of the theory are not undermined). All three of these approaches can be seen in the work of naïve realists. For example, Fish suggests the first option as a way of responding to the case of unique hues (2009: 154n.3) and both Fish and Brewer attempt to accommodate blurred vision in the second way. They suggest that blurred vision is not a matter of blurriness being subjectively added to experience of an object, but of a failure to see an object’s sharp edges – blurriness involves “systematically degraded acquaintance”, as Brewer puts it (2011: 117; see also Fish 2009: 55–7). Others take the third approach. For example, in recent work, Campbell has added a subjective modification into his account of qualitative character: “the qualitative character of your experience is constituted by the point of view from which you are observing the scene, any relevant adverbial modification of the relation of experience, and the relevant qualitative aspects of the external scene” (2014: 28, my emphasis). Similarly, Craig French argues that blurriness is a subjective addition to experience, but contends that “blurriness can form part of the character of a visual experience only if it, so to speak, attaches to the presentation of entities in experience” (2014: 411). So even if blurriness is a subjective element of phenomenal character, he argues, this would not constitute an objection to the claim that perception is a matter of acquaintance with mind-independent reality.

The Boundaries of Visual Experience Some naïve realists argue that what it is like for the subject of experience is still not entirely determined by external objects and their properties. For example, Matthew Soteriou argues that it is part of what it is like to have a visual experience that it is limited – that at the peripheries of the visual field there is a boundary that cannot be seen beyond – but that this is experienced not as a boundary “of something one is sensing, like the frame of a painting” (2013: 118), but as a limitation in me, the perceiver. To the extent that awareness of this limitation shapes the contours of my conscious experience, these contours are not determined entirely by the environment I perceive.

A further objection to the naïve realist’s account of phenomenology concerns the fact that we can have experiences of objects that no longer exist, as in the case of extinct stars, or experiences of the way objects were rather than the way objects are, as in the case of, for example, our experience of things happening on the surface of the sun. As we saw in Chapter 2, the explanation of this turns on the fact that in order to perceive an object, light reflected or emitted by that object has to reach our visual systems and be processed by them, which – where such vast distances are concerned – takes a non-negligible amount of time. In these cases then, the phenomenal character of our experiences does not reflect what our environment is like now but what our environment was like some finite time ago. This claim, however,

114  Philosophical Theories

seems to be in tension with the naïve realist’s claim that the phenomenal character of one’s perception is a matter of one’s standing in a relation of acquaintance to external objects: when I am acquainted with the sun, the phenomenal character of my experience will not be explained by the properties of the sun at the time of the experience, but by its properties around eight minutes prior. One way for the naïve realist to reply to this objection would be to deny that such experiences can be cases of perception proper; but this would seem desperate, particularly when we bear in mind that in everyday cases of perception, the light reflected from the objects around us takes some small amount of time to reach our eyes. An alternative response has therefore been to challenge the assumption that the objects that we are aware of have to coexist with the act of awareness. As Pitcher argues, time lag considerations: [do] not entail that we do not directly see things and states of affairs in the “external world,” but only that we must see them as they were some time ago. We see real physical things, properties, and events, all right, but we see them late, that is all. According to a [naïve] realist, it is a mere prejudice of common sense—and one on which the time-lag argument trades—that the events, and the states of objects, that we see must be simultaneous with our (act of) seeing them. (1971: 48) Another way to reconcile time lag cases without denying that perceptual relations are relations between existents is for the naïve realist to endorse an eternalist view of time, which claims that “past objects still exist. That is to say, they are a part of reality, despite not being located at the present moment of time” (Moran 2018: 216). Whether these responses are adequate or not is open to further discussion.

The Epistemological Hat Supposed epistemological advantages have been cited as a significant motivation for disjunctive theories more generally. Consider a skeptical argument that runs as follows. When we hallucinate, the kind of experience we have clearly fails to put us in a position to know anything about the external world. In the case of a perception indiscriminable from this hallucination, the experience we have is an experience of exactly the same kind. As the experience failed to put us in a position to acquire knowledge when it was a hallucination, having the same kind of experience in the good case cannot place us in a better epistemic position. According to this argument then, even when we perceive, the underlying metaphysics of experiences has the consequence that we are not in a position to know anything about the external world. A theory that endorses disjunctivism about visual experiences offers to block this argument by denying the premise that the experience we have when we perceive is the same as the experience we have when we hallucinate. This would not, of course, prove that we do know anything about the external world, merely that

Naïve Realism  115

such knowledge is not impossible. Yet it would block the skeptic from using the impossibility of knowledge as a premise in an argument for the conclusion that we do not know anything. Of course, this is not to deny that, given the acknowledged indiscriminability of perception and hallucination, we cannot know, on any given occasion, whether we are hallucinating or perceiving (Wright 2002, 2008). It is not that this approach offers an argument to prove that we do have knowledge; rather, it offers a rebuttal to an argument that we cannot. To illustrate this, consider the familiar skeptical claim that all of our experiences might have been just as they are even if we were in the clutches of Descartes’ demon. If a naïve realist theory of experiences is correct, this is no longer possible. If any of my experiences are in fact perceptions, then, to be as they are, they have to be perceptions. It is therefore not possible that they could have been as they are and misleading. Suppose, then, that the skeptic were to reformulate the skeptical hypothesis as follows: all of your experiences might have been of the misleading kind. Now we can ask, so what? As long as they are not misleading, then many of our empirical beliefs will be justified. As McDowell puts it, this leaves the door open for us to hold that “our knowledge that [the skeptical] possibilities do not obtain is sustained by the fact that we know a great deal about our environment” (2008: 379).

Naïve Realism and Theories of Hallucination Positive Theories As the name suggests, positive theories of hallucination attempt to tell a positive story about the nature of the hallucinatory state. For example, one might claim that hallucination involves the awareness of nonphysical objects – sense data – that are indiscriminable from physical objects. Such a claim is hinted at by Austin’s observation, mentioned in Chapter 2, that “generically different” objects such as lemons and bars of soap can nevertheless look exactly the same (1962: 50). An alternative positive theory is presented by Mark Johnston (2004). Johnston contends that when we have a successful case of perception, we are aware of an instantiated sensible profile: “a complex, partly qualitative and partly relational property, which exhausts the way the particular scene before your eyes is” (2004: 134). Importantly, the sensible profile that we are aware of, says Johnston, is a type not a token; had we stood before an array of different particulars instantiating the same sensible profile, what we are aware of – the sensible profile – —would have been the same. Then, when you have a hallucination that is indiscriminable from this experience, “you are simply aware of the partly qualitative, partly relational profile. … When the visual system misfires, as in hallucination, it presents uninstantiated complexes of sensible qualities and relations” (2004: 135). On Johnston’s view, there are then clear similarities between good cases and bad cases—in particular, in both cases the subject is aware of the same sensible profile. Yet there are important differences too. “When we see”, says Johnston, “we are aware of instantiations of sensible profiles” (2004: 135 emphasis added). When we hallucinate,

116  Philosophical Theories

however, we are only aware of the structured qualitative parts of such sensible profiles. Any case of hallucination is thus a case of “direct” visual awareness of less than one would be “directly” aware of in the case of seeing (2004: 137 emphasis added). The objects of hallucination are instead “proper parts” of the objects of seeing (140). The difficulty faced by positive views such as these is that they flirt with what is known as the screening off problem (Martin 2004). The screening off problem begins from the plausible claim that all that is needed for a hallucination to take place is the right kind of activity in the subject’s brain. Take a simple, Austin-inspired sense datum theory of hallucination first. It seems reasonable to suppose that a certain pattern of neural activity must suffice for subjects to be aware of sense data in the bad case of hallucination. But if this is so, what about the neural activity that occurs in the good case of perception? If this also suffices for subjects to be aware of sense data and if – as is suggested – this awareness explains why the subjects take themselves to be perceiving a real-world object, then this awareness of sense data would seem to “screen off” the supposed object-involving phenomenal character of the perception itself from explaining why the subject’s experiences are as they are. If, however, it is claimed that the neural activity does not suffice for awareness of sense data in the good case of perception, then we might wonder why: if the mechanism or brain state is a sufficient causal condition for the production of an image, or otherwise characterised subjective sense-content, when the [objects] are not there, why is it not so sufficient when they are present? Does the brain state mysteriously know how it is being produced … or does the [object], when present, inhibit the production of an image by some sort of action at a distance? (Robinson 1994: 153–4) It is less clear how Johnston’s view fares here. At one point he asks: Why isn’t awareness of a sensible profile a common act of awareness as between seeing and hallucination? It may be held to be … But it does seem that once we adopt the act/object treatment of visual experience it is more natural to individuate an act of awareness occurring at a time in terms of an object that includes all that one is aware of in the relevant time. (2004: 171) As the perceiver is aware of more than the hallucinator (in that the perceiver is aware of the particulars that instantiate the sensible profile whilst the hallucinator is aware of the sensible profile alone), Johnston’s suggestion seems to be that when we account for the perceiver’s awareness of the particulars, we thereby account for the perceiver’s awareness of the sensible profile. There is then no need to introduce an additional awareness of an (uninstantiated) sensible profile. Yet it is not clear that this suffices to ward off the screening off problem. After all, if neural activity does suffice for awareness of an uninstantiated sensible profile in the bad cases, it should suffice in the good cases too. And this is the case, whether or

Naïve Realism  117

not we need to appeal to this to explain the fact that the subject is aware of a sensible profile at all. So Johnston’s view may also be threatened by the screening off worry. Alternatively, the positive theorist could draw on the idea that indiscriminability is an epistemic notion, and allow that a hallucination need not involve the awareness of identical objects – or, more generally, involve identical phenomenal character – so long as the subject remains unaware of any differences. In this light, Fiona Macpherson (2013) suggests an approach to hallucinations that treats them as states with perception-like phenomenal character – she offers “perceptual imaginings or perceptual rememberings” as candidates – yet in which “the subject misidentifies the nature of their experience by introspection” (24). Keith Allen (2015) defends a version of this approach, arguing that hallucinations are in fact a kind of sensory imagining – understood as mental events with phenomenal character that is similar to the phenomenal character of the corresponding perceptual experiences – by looking at discussions of hallucinations in the psychopathology literature. These discussions, he suggests, give us reason to think that many actual cases of hallucination appear to have imagination-like phenomenal character. Again, on Allen’s view, sensory imaginings become hallucinations in situations when “subjects are unable to tell through reflection whether they are perceiving or whether they are imagining” (2015: 293). On this kind of view, hallucinating subjects make what Dokic and Martin (2012: 536) call a “metacognitive error”: their subjects mistake a mental state of one kind – an imaginative state with imaginative phenomenal character – for a mental state of a different kind. As they note in their discussion of such errors, placing weight on the subject’s imperfect reflective capacities in this way has the consequence that “the nature of the first-order state which causes the [metacognitive] error is irrelevant. It need not be an intrinsically conscious state. It need not even be the same in each case of hallucination” (537). This is because, so long as the subject mistakes a certain mental state for a perception of a certain kind, it doesn’t matter what the intrinsic nature of that state actually is; all that we can say, given that the subject mistakes it for a perceptual state, is that it is not perceptual.

Multidisjunctivism Reflection on this claim suggests the possibility that “there is no common fundamental kind to all hallucinatory experiences” (Hellie 2013: 164). On such a view, the class of states that are mistaken for perceptions – the class of hallucinations – is itself a disjunctive kind. For this reason, Hellie calls this view “multidisjunctivism”.

Epistemic/Eliminativist Theories As Dokic and Martin note, once we start paying attention to the fact that concepts such as discriminability and indiscriminability are epistemic notions, the positive nature of the hallucinatory state seems less obviously central to an account of hallucination. What matters, we might think, is not the intrinsic metaphysical nature

118  Philosophical Theories

of the hallucinatory state, but the epistemic fact that it is mistaken for a perception. When combined with the concern that any positive account of hallucination will run into difficulties with the screening off problem, some naïve realists therefore prefer to offer a purely epistemic account of hallucination. In response to the question of what account can be given of the phenomenal character of hallucination, the epistemic theorist says that all we can say about indiscriminable hallucinations is that they are not perceptions but are indiscriminable from them. On this view, the phenomenal character of a hallucination – the property of the hallucination that types it by what it is like to undergo it – is its property of being indiscriminable from a perception of a certain kind. An important feature of this indiscriminability-based approach is that it enables the epistemic theorist to avoid the screening off problem. The key feature of the picture that enables this turns on the claim that indiscriminability properties have “inherited or dependent explanatory potential”, and that such cases therefore “offer us exceptions to the general model of common properties screening off special ones” (Martin 2004: 70). To explain the notion of a property having “inherited or dependent explanatory potential”, consider the property of being an unattended bag in an airport, which causes a security alert. Sometimes objects with this property are harmless, but sometimes they contain a bomb. Now ask: does the property common to harmless and non-harmless objects – that of being an unattended bag in an airport – explain why there is a security alert in such a way that the non-common property of being a bomb in an airport is screened off from being explanatory? Not at all. Instead, the only reason the common property of being an unattended bag in an airport has the explanatory role it does is because, sometimes, this property is correlated with the special property of being a bomb in an airport. In such a case, we can say that the explanatory potential of the common property of being an unattended bag in airport is “inherited from” or “dependent upon” the explanatory potential of the special property of being a bomb in an airport. The epistemic theorist exploits this by arguing that the key property of being indiscriminable from a perception of a certain kind – a property common to both perceptions and their indiscriminable hallucinations – has just this kind of inherited or dependent explanatory potential. Why did James shriek like that? He was in a situation indiscriminable from the veridical perception of a spider. Given James’s fear of spiders, when confronted with one he is liable so to react; and with no detectable difference between this situation and such a perception, it must seem to him as if a spider is there, so he reacts in the same way. (Martin 2004: 68) So in order to explain why James shrieks when he hallucinates a spider, we not only have to say that his experience has the property of being indiscriminable from a perception of a spider, but also that a perception of a spider constitutes, for James, a reason for shrieking. So the property of being a perception of a spider is, like

Naïve Realism  119

the property of being a bomb in an airport, a special property that is needed to explain why the common property has the explanatory potential it does. Whatever explanatory potential an indiscriminability property has in explaining why James behaves as he does is inherited from the explanatory potential of the associated case of real perception. Martin therefore suggests that the screening off worry can be avoided by characterizing the hallucinatory state purely epistemically. The naïve realist should say that “when it comes to a mental characterization of the hallucinatory experience, nothing more can be said than the relational and epistemological claim that it is indiscriminable from the perception” (2004: 72). Other naïve realists who endorse Martin’s broad approach to hallucination have nonetheless attempted to say a little more about why a hallucination is indiscriminable from a veridical perception. For example, Fish (2009) argues that two states will qualify as indiscriminable for their subject so long as they result in the same cognitive effects, where both behavior and (in conceptually sophisticated creatures) introspective beliefs qualify as species of cognitive effect. He argues that so long as a non-veridical mental state led its subject to have the same beliefs that they would have had in a perceptual case – including, crucially, beliefs that they are having a visual experience of a certain kind – then it will be, from the subject’s perspective, just as though they were having a veridical experience of that kind. On this view, it doesn’t matter what the intrinsic nature of the state is. Fish even contends that, in the limiting cases, hallucinations do not possess phenomenal character at all. So long as it yields the same cognitive effects, Fish argues, a mental state will qualify as indiscriminable from, and hence a hallucination of, a veridical perception of a certain kind. So whilst this view says a little more than Martin’s purely epistemic characterization of hallucination, it is nonetheless not a positive theory inasmuch as it is eliminativist about hallucinatory phenomenal character. Logue (2012a) attempts to add yet more explanation to the picture without thereby falling foul of the screening off worry. Whilst she joins Fish in endorsing eliminativism about hallucinatory phenomenal character, Logue doesn’t attempt to ground the indiscriminability of perception and hallucination solely in the common effects of the two states. Instead, Logue provides a further explanation of why the two states have similar effects. This turns on the idea that total hallucination generates a defective context for the employment of introspection. Because of this, she suggests, in cases of total hallucination, introspection will only be sensitive to the first-order states’ representational features, and not their phenomenal characters. This opens up room for a more robust characterization of hallucination, on which “total hallucinations fundamentally consist in the subject perceptually representing her environment as being a certain way, but they lack phenomenal character” (2012: 183). As introspection is only sensitive to representational features in these defective contexts, both perceptions and hallucinations with common representational properties will “seem” the same to introspection, despite one having and one lacking phenomenal character. This would therefore explain why hallucinating subjects believe that they are in states with phenomenal character when they are not.

120  Philosophical Theories

Yet, whilst Logue does acknowledge that the property of perceptually representing the environment as being a certain way is a common factor across indiscriminable cases of perception and hallucination, she argues that this can nonetheless avoid the screening off worry. To do this, she begins by arguing that the relevant perceptual representational state – the state that in hallucinatory cases is the hallucination – is merely a constituent of the indiscriminable veridical experience. Given that the veridical experience possesses phenomenal character, being in the representational state that is a constituent of this experience will lead the subject to believe that they are in a state with phenomenal character. Then, in a hallucinatory case, being in the same representational state will cause the subject to believe that they are in a state with phenomenal character, even though they are not. To this extent, Logue follows Martin in holding that the explanatory potential of the common representational property is inherited from the explanatory potential of the special case of veridical experience. Yet when pressed to say something more about why a hallucination has the epistemological and behavioral effects that it does, Logue can appeal to the presence of this representational state, which gives the theory a level of explanation that goes beyond that offered by Martin and Fish.

Objections to Epistemic Theories Given Martin’s characterization of the hallucinatory state as a state that is simply indiscriminable from a perception of a certain kind, a lot hangs on the way in which the key notion of indiscriminability is understood. Martin himself suggests that a hallucination of an F “is such that it is not possible to know through reflection that it is not one of the ... perceptions [of an F]” (2006: 364). We can therefore define indiscriminability as follows: x is indiscriminable from a perception of an F, if and only if x is such that it is not possible to know through reflection that it is not a perception of an F. Yet elements of this analysis of indiscriminability have been the source of objections. Both the restriction to the relevant knowledge being acquired “through reflection” and Martin’s interpretation of the modality present in “not possible to know” have been challenged. To take the “through reflection” clause first, one way of coming to know that your experience is not a perception of an F is by testimony. If I tell you that I slipped a drug in your coffee, you may on these grounds come to realize that your experience of pink elephants is non-perceptual. In this way, we can see that it is possible for you to know that your experience is not a perception of pink elephants. Yet even though you can know that your experience is not a perception of pink elephants in this way, we do not want this to preclude your experience’s being indiscriminable from a perception of pink elephants. This is why Martin includes the “through reflection” clause into the definition of indiscriminability, in order to rule out the possibility of knowledge gained from testimony being relevant to the question of whether or not two experiences are indiscriminable. Sturgeon, however, argues that this proposal cannot be made to work (2006). On the one hand, the “through reflection” restriction must be strong enough to rule out any of the routes by which a hallucinating subject might “figure out” that they are hallucinating. He suggests that it must therefore be taken to stipulate that

Naïve Realism  121

the “information involved in background beliefs cannot be generally available to reflection …. Otherwise the possibility of everyday knowledge of [hallucination] will slip through the net [and] count as knowledge obtainable by reflection” (2006: 209). On the other hand, when one hallucinates an F, one is thereby in a position to know a vast array of things. As a hallucination of an F is discriminable from perceptions of Gs, Hs, and Js, Martin’s definition will require that for each case, a subject hallucinating an F can know that, by reflection alone, their experience is not one of these cases of perception. But Sturgeon suggests that this: is a huge amount of knowledge to be got solely by reflection … and not by reflection on the visual character of [the hallucination], recall. … The only way that could be true, I submit, is if background beliefs were generally available to reflection on context. (2006: 210) Sturgeon therefore presents Martin with a dilemma. On the one hand, to rule out the possibility we might simply use our background beliefs to figure out that we are hallucinating; the “through reflection” clause must restrain us from making use of background beliefs. Yet, on the other hand, to make sense of all the reflective knowledge Martin’s theory allows that we are in a position to acquire when we hallucinate, the “through reflection” clause must allow us to make use of background beliefs. But this, suggests Sturgeon, is just to say that Martin cannot give an adequate account of the “through reflection” restriction. Another source of objections has stemmed from Martin’s interpretation of the “not possibly knowable” condition. This condition is in the picture to enable Martin to acknowledge that creatures that lack the sophistication to know things might nonetheless have hallucinations. This is a prima facie problem because if a creature cannot know things at all, then for any hallucination it might have, the creature will not know that it is not perceiving an F, or a G, or an H, and so on. So as it stands, Martin’s definition of indiscriminability threatens to count any creature hallucination as indiscriminable from each and every kind of creature perception. To avoid this, Martin states that whilst a creature: might fail to discriminate one experience from another, making no judgment about them as identical or distinct at all, that is not to say that we cannot judge, in ascribing to them such experience, that there is an event which would or would not be judgeably different from another experience. (2004: 54, emphasis added) In other words, Martin suggests that when we talk about what it is or isn’t “possible to know”, we are not talking about what the subject – with the subject’s idiosyncratic capacities – is or is not in a position to know. Rather, we are talking in an impersonal way. When we say that a hallucination is not possibly known to be distinct from a perception of a certain kind, we do not mean not possibly known by the subject, but rather not possibly known in some impersonal sense.

122  Philosophical Theories

Siegel argues that this approach renders us unable to pick out the hallucinatory “experience” – the state or event that is reflected upon – in an appropriate yet non-question-begging manner (2008: 212). Given Martin’s view, the hallucinatory experience cannot be identified by its having a certain robust property. This would conflict with Martin’s insistence that nothing more can be said of the hallucination than that it is indiscriminable from the perception. Yet we cannot pick out the relevant state in virtue of its indiscriminability property either, because we are trying to explain what it is for a state of the creature to have the indiscriminability property in the first place. This means that we cannot identify the experience we are talking about by appeal to its being the one that has the indiscriminability property.

Indiscriminability as Identity of Cognitive Effects In an attempt to avoid these objections, Fish (2009) departs from Martin on both of these issues: where Martin endorses an impersonal sense of indiscriminability, Fish endorses a personal sense; where Martin rules out testimony, Fish rules it in. He contends that this solves the dog problem on the grounds that as both behavior and (in conceptually sophisticated creatures) introspective beliefs qualify as species of cognitive effect, hallucination in a conceptually unsophisticated creature will therefore qualify as indiscriminable from a perception of a certain kind so long as it yields the kinds of behavior that a perception of that kind would have yielded. When it comes to known hallucinations, Fish asks what the effects of a perception in a subject would be who believed through testimony that they are hallucinating. He suggests that in such a case, both the hallucination and the perception would yield the same kinds of belief, including the belief that they are having a visual experience of a certain kind. He therefore contends that in such a case, the hallucination would have the same cognitive effects as a perception would have had, and thereby qualifies as indiscriminable from that perception. Siegel (2008) and Dokic and Martin (2012: 538) raise objections to these claims.

Naïve Realism and Illusion We have seen two naïve realist approaches to hallucination: positive theories, which claim that a positive account can be given of hallucination, and eliminative/epistemic approaches, which claim that the defining feature of a hallucination is that it is indiscriminable from a perception of a certain kind. A further question for the naïve realist is this. If perceptions and hallucinations are experiences of different kinds, what should we say about cases of illusion? These are similar to perceptions inasmuch as something is seen, but similar to hallucinations inasmuch as we get things wrong. The two obvious possibilities are to place illusion into one of the two

Naïve Realism  123

disjuncts that we already have: to treat illusions as either like hallucinations or like perceptions.

V vs IH Disjunctivism The terminology of V vs IH Disjunctivism is taken from Byrne and Logue (2008). It indicates that illusion (I) is being treated along with hallucination (H) and as distinct from veridical perception. McDowell seems to endorse this approach when he claims that “an appearance that such-and-such is the case can be either a mere appearance or the fact that such-and-such is the case making itself perceptually manifest to someone” (1982/1998: 386–7). As McDowell’s V disjunct contains cases in which a “fact” is made manifest, as there is no such thing as a non-obtaining fact, then any scenario in which it appears to the subject that such-and-such is the case when it is not could not be a case of a fact being made manifest. So illusions look to fall into the category of cases in which it merely appears as though a fact is made manifest along with hallucinations. However, there are concerns with an attempt to treat illusions as hallucinations. Robinson protests that: if all non-veridical perceptions were treated in the same way as hallucinations, then every case of something not looking exactly as it is would be a case in which one was aware of some kind of subjective content. Only perfectly veridical perceptions would be free of such subjective contents. (1994: 159) This leads A.D. Smith to ridicule the view: the “picture of our daily commerce with the world through perception that therefore emerges is one of a usually indirect awareness of physical objects occasionally interrupted by direct visions of them glimpsed in favoured positions” (2002: 28).

VI vs H Disjunctivism So, perhaps we would do better to bring illusion under the perceptual (V) rather than the hallucinatory disjunct. This gives us VI vs H Disjunctivism. If illusion is treated as a special case of perception, then the specific way in which illusion is treated will be dictated by the particular theory offered of perception. This approach can therefore seem unavailable to the naïve realist. If perception involves acquaintance with objects and their properties, then if illusions really are cases in which objects look to be a way that they are not, this approach to illusion would not appear to be available to a naïve realist. However, a number of naïve realists have nonetheless attempted to treat at least some illusions as cases of veridical perception. For example, Louise Antony develops a version of this approach for the illusion of seeing a straight stick in water that looks bent. She argues that in such cases, we are aware of the stick’s bentishness: “appearance properties are properties of objects: it is an objective feature of some

124  Philosophical Theories

sticks in water … that they are bentish” (2011: 35). Mark Kalderon gives a similar account of color illusion. He argues that if we see a blue bead under pink light that looks black, we still see the bead’s blue color: “Why should the fact that the bead looks black mean that we do not see the blue colour of the bead – after all, that is just what blue things look like in pink light” (2011: 769). At the beginning of this chapter, we noted that the naïve realist cannot hold that the phenomenal character of an experience is fixed solely by the nature of the objects of perception, but must also take into account features of the subject’s perspective on those objects as well as facts about the subject themself. By emphasizing the role of the environmental circumstances in which perceptions of this kind occur, this approach to illusion locates differing color experiences of the same object in the differing perspectives subjects take on that object. Similar approaches can also be found in Fish (2009) and Genone (2014). Another approach that is available to the naïve realist in treating illusions as cases of perception is to locate the differences in “illusory” experiences as differences in the way the subject experiences the objects (Beck 2019). Essentially, this approach extends the Campbell/French approach to blurred vision that we saw earlier to accommodate many of those cases that are described as illusions. A related approach, which is often combined with the kind of view developed in the previous paragraph (e.g. Fish 2009; Genone 2014), is to insist that illusion is not, strictly speaking, a feature of experiences themselves, but rather a feature of our cognitive response to our experiences. On this view, as far as their experiential nature goes, illusions are perfectly veridical. As Brewer puts it: The intuitive idea is that, in visual experience, a person is simply presented with the actual constituents of the physical world themselves. Any errors in her world view which result are the product of the subject’s responses to this experience, however automatic, natural, or understandable in retrospect these responses may be. Error, strictly speaking, given how the world actually is, is never an essential feature of experience itself. (2008: 169) In recent work (2011, 2013), Brewer has developed this approach in more detail. Consider the Müller-Lyer illusion once more. According to Brewer, the addition of the arrowheads makes it the case that the Müller-Lyer lines have visually relevant similarities to paradigm cases of uneven length lines at different distances from the perceiver, where visually relevant similarities “are similarities relative to the sensitivities of the various processes underlying vision” (2013: 425; for a similar empirically informed approach, see Phillips 2016a). Although the subject still perceives the Müller-Lyer lines as they are, if they register these visually relevant similarities, then the subject will thereby be inclined to judge that the lines are of different lengths. So where the other theories we have considered attempt to explain this as a perceptual phenomenon – with sense datum theorists claiming that we are aware of different length sense data, adverbialists that we sense different-length-ly, and intentionalists that our experience represents the presence of two different length lines – Brewer’s alternative holds, in accord with naïve realism, that the phenomenal

Naïve Realism  125

character of our experience is part-constituted by two same length lines, but that because of the arrowheads, we take ourselves to be confronted with two different length lines. Thus the illusory aspect of the experience – the fact that we go wrong about the way the world is – is not a feature of experience itself, but a feature of the way we (cognitively) take that experience.

Naïve Realism and the Causal Theory of Perception In contrast to the various common factor theories we have discussed so far, the naïve realist has no need to stipulate extra conditions under which a visual experience as of an F constitutes a successful perception of an F. For the naïve realist, successful perceptions of F are fundamentally different mental states from their unsuccessful, yet indiscriminable, counterparts. Whether or not a subject is successfully perceiving an F, then, is merely a matter of the fundamental kind of mental state the subject is in. As far as the question of whether the subject is perceiving or not is concerned, questions about the causal history of this state are simply irrelevant. Paul Coates has argued that this feature of the view leaves the naïve realist unable to adequately accommodate the possibility of veridical hallucinations. If the causal history of the state is irrelevant, Coates asks, then how are we “able to classify an experience as hallucinatory, when there is an object present that matches that experience” (2000: 328)? Yet the answer to this question (limited to the case of total hallucinations for simplicity) is that the experience is hallucinatory because it does not have world-involving phenomenal character, yet is indiscriminable for its subject from a state that does (Hobson 2013: 563). Essentially, for the naïve realist, the very concept of a veridical hallucination is unacceptably problematic, as the two constituent concepts – being veridical and being a hallucination – do not come apart as they do for theories that accept the Common Factor Principle. For the naïve realist, a hallucination is fundamentally not a successful perception, despite being indiscriminable from one: if a case is hallucinatory, then it cannot also be veridical; if a case is veridical, it cannot also be hallucinatory. Having said this, it seems to be possible that we might find ourselves in a situation where a subject was having a hallucination of an apple (understood as above) in a situation in which not only was an apple present, but all of the consequences of that hallucination – the subject’s utterances, their behavior, and so on – were appropriate. That is to say, on the basis of their hallucination, the subject was enabled to coherently talk about the apple (and their apparent experience of it) as well as reach for the apple, grasp the apple, bite the apple, and so on. However, not only would this still not constitute a veridical hallucination, it might cause us to have a good look at why we think the case is hallucinatory in the first place. If, for example, we deem the case to be a hallucination because the causal link between the apple and the subject is nonstandard, then so long as the subject’s capacities with respect to the apple were robust enough, we might think that rather than take the nonstandard causal chain to mean the subject must be hallucinating, we may in

126  Philosophical Theories

fact have discovered another causal mechanism that can underpin our ability to perceive apples. This does raise an interesting question, however, concerning precisely how the naïve realist understands the role of causation. There are a number of reasons to think that the claim that in order for a subject to see an object, the object must causally affect the subject, could be a conceptual truth about perception – that it would be “immediately acknowledgeable by any person, whatever their education, who can count as having the concept in question” (Snowdon 1980: 176). For example, P.F. Strawson argues that the: idea of the presence of the thing as accounting for, or being responsible for, our perceptual awareness of it is implicit in the pre-theoretical scheme from the very start. For we think of perception as a way, indeed the basic way, of informing ourselves about the world of independently existing things: we assume, that is to say, the general reliability of visual experiences; and that assumption is the same as the assumption of a general causal dependence of our visual experiences on the independently existing things we take them to be of. (1979: 51) In addition, the role of causation is arguably implicated in a plausible explanation of why perception has the defeaters it does: why perception fails in the conditions in which it does fail. For instance, we don’t see the object if our eyes are closed, or other opaque objects are interposed between us and the object, and so on (Child 1994). What these have in common is that in each case, the causal link between the object and the subject is broken. If we were to accept that it is a conceptual truth about perception that an object must causally affect a subject in order for the subject to see that object, we might wonder whether this could be used as a premise in an argument for the Common Factor Principle. For example, we might ask how the presence of causation could be a necessary condition for perception to take place unless it was also the case that the seen object O causes the subject S to have an experience that is, in its intrinsic nature, independent of that object. To satisfactorily address this challenge, the naïve realist therefore needs to explain how we can accommodate the thesis that it is necessarily true that O must causally affect S in order for perception to take place without its being the case that O produces in S a visual experience, conceived of as a good case/bad case common factor.

Is the Common Factor Principle a Conceptual Truth? In the quote from Strawson, he appears to argue that the claim that objects must causally affect subjects to be seen is not merely a potential premise in an argument for something like the Common Factor Principle – that there is “a general causal dependence of our visual experiences on the independently

Naïve Realism  127

existing things we take them to be of” – but is in fact essentially the same claim. However, Snowdon argues that the very coherence of disjunctivism means that the Common Factor Principle itself cannot be a conceptual truth. He argues that even if disjunctivism turns out to be false, it will only be “scientifically established facts about perceptual and hallucinatory processes” that disprove it (1990: 130). But these are results that the man on the street could not be expected to know merely in virtue of having the concept of perception. Even if it does turn out to be false then, disjunctivism is not a conceptual falsehood and therefore the claim that the intrinsic nature of the experience a subject has when perceiving an object is independent of anything outside the subject cannot be a conceptual truth, even if it is a part of the concept of perception that we can only see objects if they causally affect us. William Child offers a way of accommodating causation without committing to the Common Factor Principle’s ontology of experiences by making use of what is known as a distinction between personal and sub-personal levels of explanation. This distinction separates out the level of talking about human beings qua minded agents from the level of talking about human beings qua physical systems or, as Dennett puts it, “the explanatory level of people and their sensations and activities … and … the sub-personal level of brains and events in the nervous system” (1969: 93). For example, consider Lewis’s account of the causal processes involved in perception: Ordinarily, when the scene before the eyes causes matching visual experience, it happens as follows. Parts of the scene reflect or emit light in a certain pattern; this light travels to the eye by a more or less straight path, and is focused by the lens to form an image on the retina; the retinal cells are stimulated … and the stimulations comprise a signal which propagates up the optic nerve into the brain; and finally there is a pattern of stimulation in the brain cells which either is or else causes the subject’s visual experience. (1980: 83–4) Here we find that Lewis moves quite freely from talking about the causal processes at the sub-personal level, such as the stimulation of retinal cells, to an effect – a visual experience – at the personal level. According to Lewis, processes in the subject’s brain (sub-personal level) ultimately are or cause an experience (personal level). Others, however, suggest that these levels of explanation ought to be kept separate: The mental state of affairs, o’s looking F to S, is not a state or event at the end of a causal chain initiated by o; it is, rather, a (larger-sized) event or state of affairs which itself consists in the whole chain of physical events (not merely events within S) by which o causally affects S. The experience is the complete

128  Philosophical Theories

state of affairs, o causally affecting S. The ultimate effect in this causal state of affairs—the state or event which lies at the end of the causal chain which starts with o—is something physical in S; but that ultimate effect is neither identical with nor constitutive of the experience itself. (Child 1994: 161–2) Essentially, Child attempts to secure a role for causation at the physical (or sub-personal) level, whilst denying the common factor conception of mental states at the mental (or personal) level. This would enable him to both endorse the claim that causation is necessary for perception while rejecting the Common Factor Principle. However, the availability of this response does depend, in part, on whether or not we ought to keep a strong distinction between the personal and sub-personal levels, and as we shall see in the next section, not everyone agrees that we should. Moran (2019) develops this idea and suggests that the causal theory can be reconciled with a naïve realist framework to yield a new way of avoiding the screening off problem. He argues that if a veridical perceptual experience supervenes on the whole chain of physical events by which the perceived object causally affects the perceiver (which, as the case is veridical, we can assume is an example of a “standard” causal history), then we could also claim that cases of hallucination supervene on their entire deviant causal history. This, he suggests, will enable the naïve realist to resist the screening off problem on the grounds that even if the proximal stages of the causal chains are similar, the overall causal chains will differ. If hallucinations supervene on the entire deviant causal chain while veridical perceptions supervene on the entire standard causal chain, then in every case in which the common neural activity occurs, only one kind of experience will occur. There is therefore no scenario in which a non-veridical experience occurs at the same time as a veridical perception, and thereby threatens to “screen off” the veridical perception from playing an explanatory role. A similar view is outlined by Allen (2015) who, recall, endorses a positive theory of hallucination as a degenerate kind of sensory imagination. Allen accepts that as perception and mental imagery share neural substrates, it is possible that the same neural activity that occurs during a case of veridical perception could occur in a case of hallucination and that in such a case, the subject would have a hallucination that was indiscriminable from the original perception. However, as his view claims that hallucinations have imaginative phenomenal character in order to stay clear of the screening off problem, he therefore has to resist the claim that this neural activity would produce this hallucinatory/imaginative phenomenal character when it occurs during a perceptual episode. He therefore claims that the phenomenal character of hallucination is not determined by neural activity alone, but by the right kind of neural activity occurring in the absence of an appropriate object.

Is Naïve Realism Refuted by Empirical Science? As the discussions in the previous section might suggest, naïve realism has somewhat of a particular relationship with empirical science. The causal facts about perception have, for a long time, been taken to provide an argument in favor of the

Naïve Realism  129

common factor theory, and against naïve realism and disjunctivism. The thought behind this argument is that given the different stages in the causal chain between the object and perceiver, we might think that we could intervene and activate one of the intermediate stages of this causal chain in a nonstandard manner (say by direct stimulation of the retina, optic nerve, or visual cortex). In such a situation, we might then attempt to infer that so long as the proximal causes are the same, then the same effects will be produced. And if we include perceptual experiences themselves as an effect, then this would produce the same experience, despite having been caused in a nonstandard way. As we saw, however, the naïve realist can respond to this by denying that perceptual experiences are the endpoints of causal chains in this way, holding instead that perceptions have a much wider supervenience base than this. To see how this might work, consider an alternative case in which an intermediate stage of a causal chain is produced in a nonstandard manner. There is a “causal chain” for the production of legitimate banknotes, which begins with (I guess) an order from a competent authority and concludes with notes being printed out on a certain kind of machine. Suppose however that forgers manage to either duplicate or get illegitimate access to the machine that prints legitimate banknotes. In this way, we can say that the forgers “activate the intermediate stages of the causal chain in a nonstandard manner”. Yet, even though the latter stages of the causal chain are the same – those involving the functioning of the machine itself – the banknotes the forgers print on it will still be counterfeit. The “same kind of banknote” will not result, even though the immediate “cause” of the forgeries is the same as the immediate “cause” of genuine currency. If we can make sense of the possibility of two different mental states having equivalent proximal neural causes, then the disjunctivist can resist the causal objection and allow that perceptions and hallucinations are distinct mental states, even in situations where they have the same proximal causes. In recent work by Tyler Burge, however, a variant of this argument has been resurrected, which claims not that the same proximal causes must give rise to the same mental state, but that “the methodology of all serious empirical theory of vision guarantees that given types of visual state can be veridical in some circumstances and non-veridical in others” (2005: 23). To the extent that naïve realism requires disjunctivism, which rejects this claim, the view is thereby revealed to be inconsistent with visual science and should therefore be rejected on this basis. In outlining this argument, Burge makes two key claims. First that it is part of the core commitments of this scientific field that it concerns the “visual states of individuals, ordinarily so-called” (2005: 22); second that when it comes to such states, visual science is committed to the Proximality Principle – the principle that if we hold a subject’s antecedent psychological state constant, a particular proximal stimulation will produce a particular perceptual state. If Burge is right, then not only is visual science committed to the view that the same proximal causes will produce the same effects, but that these effects are the everyday perceptual states that philosophers of perception are interested in. If this is correct, then even though the alternative view of perceptual states endorsed by the naïve realist is not incoherent, it is inconsistent with the science: “It is fairly unusual, at least since the days

130  Philosophical Theories

of Descartes and Newton, for philosophical views to be as directly at odds with scientific knowledge as disjunctivism is. Hegel’s claim that there are seven planets comes to mind” (2005: 29). In response to Burge’s arguments, Campbell points out that “classical cognitive science has no constitutive interest in problems of consciousness” (2010: 209), which undermines Burge’s claim that the perceptual states studied by vision science are indeed the same things that philosophers of perception – for whom consciousness tends to be a primary feature – are interested in. Once we recognize this, he suggests, we are not compelled to see vision science and naïve realism as incompatible: “Vision scientists might characterize [perceptions and hallucinations] in the very same information-processing terms. But it does not follow from this that there is no distinction to be drawn between those states” (2010: 210). Likewise, McDowell insists that the philosophy of perception and the psychology of perception are engaged in different explanatory projects, and that there is a difference between attributing a perceptual state to a perceptual system (as psychology does) and attributing a perceptual state to a perceiver. He agrees with Campbell that once we take note of this distinction, we can see that a defender of naïve realism or disjunctivism is not flouting established scientific knowledge in their rejection of common state types in vision: [Burge’s discussion] seems remarkably insensitive to the possibility that it might matter who or what is in a state. Experiences are states of perceivers; the states that perceptual systems get into when they as it were solve the problem of moving from sensory input to representations of the environment (which is not something perceivers do even in a metaphorical sense) are states of perceptual systems. … I urge that a good theory of the workings of a perceptual system yields accounts of what enables a perceiver to get into perceptual states, not accounts of what it is for a perceiver to be in those states. Burge scoffs at this. But he does not understand its point. … I raise a conceptual question about how we should understand the language used in such theories. Burge responds, bizarrely, as if to a case of empirical ignorance. (2010: 250–1) In response to these claims, Burge insists that it is not open to the naïve realist to take this line on the grounds that the science not only specifies that a particular visual state type can be veridical in some circumstances and non-veridical in others, it also specifies that the visual states it is making theoretical claims about are the very states of perceivers that Campbell and McDowell contends are the distinctive subject matter of philosophy. So to the extent that the naïve realist rejects the claim that the states of perceptual systems just are states of perceivers, that alone places them in conflict with the science (2011: 44). McDowell responds that this claim – that the states of perceptual systems studied by vision science are one and the same as the personal level states of perceivers – is not the kind of assertion that can itself lay claim to being a scientific finding: it “is a disputable philosophical claim, not one that can demand the respect due to results of empirical science” (2013: 275).

Naïve Realism  131

The crux of the disagreement then is not on whether the science stipulates that there are information-processing states in the visual system that can be veridical or non-veridical – both parties seem to accept that it does – but on whether the science can also legitimately stipulate that these visual states are the very states that philosophers are talking about when they advocate naïve realism. Burge thinks that it can – and that naïve realism is thereby revealed to be inconsistent with vision science – whereas McDowell thinks that, to the extent that empirical scientists do indeed make such a claim, they would thereby be stepping outside of science and into the realms of philosophy. If McDowell is right, then even if naïve realism is inconsistent with some of the things that scientists may say, this does not mean that it is inconsistent with the science itself. Another recent attempt to show that naïve realism can be refuted by empirical science can be found in the work of Adam Pautz. This approach is very different from Burge’s, inasmuch as Pautz focuses not on general considerations, but on specific empirical results he claims refute naïve realism. Naïve realists claim that the phenomenal character of experiences is determined by the objects and properties in the world. Pautz contrasts this with a “brain-based view” on which “our experiences of sensible properties are most directly dependent on our subjective neural response to external items” (forthcoming). In any case of perception, then, there are three relevant elements – the phenomenal character, the neural processing, and the nature of the world – that may correlate or fail to correlate with one another. The basic structure of Pautz’s argument is that science has demonstrated that similarities and dissimilarities in the phenomenal characters of experiences correlate well with similarities and dissimilarities in the character of the underlying neural processing, and correlate badly with similarities and dissimilarities in the structure of the objects of experience. Pautz describes three empirical case studies in support of this thesis: one involving smell, one hearing, and one visual color perception. In the smell case, he describes the results of a study (Margot 2009) in which subjects were asked to smell the following three chemicals and describe how they smell.

O O

Citral

H

(+)-R-limonene

H

(–)-R-carvone

Figure 5.1 Interestingly, subjects who smell all three substances tend to describe both citral and (+)-R-limonene as smelling citrus-like (albeit distinctly so) yet describe (−)-R-carvone as smelling minty. Pautz argues that when it comes to the way the chemicals smell, we should therefore group citral and (+)-R-limonene together as similar, with both being different to (−)-R-carvone. However, he argues, when

132  Philosophical Theories

it comes to the chemicals themselves, our similarity and difference groupings would be different: that (as can be seen from the figure above) (+)-R-limonene and (−)-R-carvone would be grouped together as similar in structure, with citral being distinct. He then considers a further study (Howard et al. 2009) that explores the neural processing that occurs in the posterior piriform cortex (PPC) when subjects smell the same three chemicals. This shows that the “PPC neural representation of [(+)-R-limonene] does resemble your PPC neural representation of [citral] more than your PPC neural representation of [(−)-R-carvone]” (forthcoming). So when we look at the three similarity groupings, we see that both phenomenal similarity and neural similarity group citral and (+)-R-limonene together as similar, with (−)-R-carvone distinct, whereas a focus on external structure groups them differently. The cases Pautz presents for both auditory and color experience have the same pattern. The crux of Pautz’s argument is that, although naïve realism is not flatly inconsistent with these findings, they do make naïve realism implausible. To the extent that the motivation for naïve realism is the claim that we have unmediated access to the nature of the external world, we might think that access to similar chemicals ought to yield similar experiences, and access to dissimilar chemicals yield different experiences. Yet these experiments suggest that this is not the case. To maintain naïve realism in the face of this, Pautz argues, the naïve realist therefore has to bite the bullet and hold that very similar physical properties ground very different smells, and that very different physical properties ground very similar smells. That is, to explain the similarities and differences in our experiences, the naïve realist has to posit “a swarm of inexplicable, irregular grounding connections” (forthcoming). When this complexity is set alongside the fact that the differences in our experiences correlate well with differences in neural processing, Pautz argues that the best, most simple explanation of these findings would seem to be that phenomenal character is fixed by neural processing and not, as the naïve realist suggests, external objects. One possible avenue of response for the naïve realist would be to deny that the objects of smell are chemicals – to reject the claim that particular chemicals (or clouds of particular chemicals) are the external objects that we are acquainted with in olfactory experience. For example, Wilson and Stevenson reject the idea that the function of olfaction is to respond to individual chemicals, but to “odor objects”, understood as “a synthesis of multiple (potentially hundreds) volatile molecules” (2006: 7). This could enable the naïve realist to maintain that similarly/dissimilarly structured chemicals do not necessarily constitute similar/dissimilar olfactory objects. (Perhaps a similar response could also be given to the other examples Pautz explores, although this would require further discussion.) As we shall see when we explore the objects of smell in more detail in Chapter 8, however, this would go against dominant views of olfactory objects, both in philosophy and in science. To this extent then, even if naïve realism is not flatly refuted by empirical science, it does appear to be a philosophical position that exhibits interesting tensions with some of the things that empirical scientists think and say.

Naïve Realism  133

Questions • • •

How plausible is the naïve realist’s claim that experiences can be indiscriminable without sharing phenomenal character? Which naïve realist-friendly approach to hallucination do you think is preferable: a positive theory or an epistemic/eliminativist theory? How plausible is the naïve realist’s claim that the states of perceptual systems studied by vision science are not the same as the personal level states of perceivers?

Further Reading For a recent defense of the idea that naïve realism is truly a view of the naïve, see Harold Langsam (2017) “The intuitive case for naïve realism”. For general developments of a broadly naïve realist or relationalist view, see John Campbell (2002) Reference and Consciousness, William Fish (2009) Perception, Hallucination and Illusion, Bill Brewer (2011) Perception & Its Objects, and the works of M.G.F Martin, particularly his 1997 “The Reality of Appearances”, his 2004 “The Limits of Self-Awareness”, both of which are reprinted in Byrne and Logue (2009), and his 2006 “On Being Alienated”. Campbell and Quassim Cassam’s Berkeley’s Puzzle is also an interesting written debate on these issues. Alva Noë’s “as naïve realist as one can hope to be” actionism is developed in his 2005 Action in Perception. For positive theories of hallucination, see Mark Johnston (2004) “The Obscure Object of Hallucination”, Jérôme Dokic and Jean-Rémy Martin (2012) “Disjunctivism, hallucinations, and metacognition”, Keith Allen (2015) “Hallucination and Imagination”, and Fiona Macpherson’s introduction to the 2013 collection she edited with Dimitris Platchias, Hallucination: Philosophy and Psychology. For epistemic/eliminativist approaches to hallucination, see the works by Martin and Fish mentioned above, along with Heather Logue’s 2012 “What should the naïve realist say about total hallucinations”? For criticism, see Susanna Siegel’s 2008 paper, “The Epistemic Conception of Hallucination”. For naïve-realist friendly theories of illusion, see the works by Brewer and Fish mentioned above, along with Louise Antony (2011) “The Openness of Illusions” and James Genone (2014) “Appearance and Illusion”. For the criticism that disjunctivism, and hence naïve realism, is incompatible with vision science, see Tyler Burge’s 2005 “Disjunctivism and Perceptual Psychology” and 2011 “Disjunctivism Again”. For responses from naïve realists, see John McDowell’s 2010 “Tyler Burge on Disjunctivism” and 2013 “Tyler Burge on Disjunctivism (II)” as well as John Campbell’s 2010 “Demonstrative Reference, the Relational View of Experience, and the Proximality Principle”. For the criticism that empirical results disprove naïve realism, see Adam Pautz (forthcoming) “Naïve Realism and the Science of Sensory Consciousness”.

✔

✘

✘

Belief acquisition theory Weak intentionalism

✔

✘

✘

✘

✔

Qualia theory

Phenomenal intentionality Naïve realism

✔

✔

✘

✔

✘

✔

✔

✘

✘

✔

✘

✔

Pure representationalism Adverbialism

Higher-order perception theory Higher-order thought theory Representationalism

✔

✘

Percept theory

✔

✔

Sensory core theory

✔

✔

Common Factor Principle

Sense datum theory

Phenomenal Principle

✘

✔

✘

✘

✔

✔

✔

✔

✔

✔

✔

✔

✘

Representational Principle

Philosophical Theories of Visual Perception: Summary Table

Acceptance of PP with CFP shows that all visual experience involves acquaintance with mental objects. Experience of external objects is thus indirect. Awareness of sense data is accompanied by a “taking for granted” of propositions about the mind-independent world. Sense data themselves have representational properties. The details suggest that the theory does not, in fact, endorse PP. Rejection of PP removes the need for mental objects. Perception is no more than the acquiring of beliefs. Perception involves a distinctly perceptual relationship to content; the phenomenal aspects of perception are explained independently. Perception involves a distinctly perceptual relationship to content; first-order perceptual states are conscious when they are the object of a higher-order perception or experience. Perception involves a distinctly perceptual relationship to content; first-order perceptual states are conscious when they are the object of a higher-order thought or belief. Perception involves a distinctly perceptual relationship to content; first-order perceptual states are conscious when they play a particular functional role. Perception involves a distinctly perceptual relationship to content; first-order perceptual states are conscious simply in virtue of bearing this relationship to content. Perception involves adverbially modified events of sensing. When we see a red ball, a red ball causes us to sense redly. Perception involves experiences possessing qualia: intrinsic, consciously accessible properties that are responsible for the experience’s conscious aspects. Perception involves awareness of qualia, understood as inherently intentional phenomenal properties; perceptual states have content fixed by these phenomenal properties. Acceptance of PP with the rejection of the CFP allows that veridical experience involves acquaintance with mind-independent objects.

Notes

PART II The Philosophy of Perception and the Sciences of the Mind

Taylor & Francis Taylor & Francis Group

http://taylorandfrancis.com

6

The Philosophy of Perception and Vision Science

Overview This chapter focuses on the interplay between the philosophy of perception and the empirical sciences that investigate the mind, with a focus on scientific investigations into vision. The chapter begins by exploring the dominant information processing paradigm with cognitive science and looks at recent developments of and challenges to this paradigm. It then moves on to look at the division between perception and cognition and investigates the empirical debate over whether or not perception is cognitively penetrable, as well as the associated philosophical puzzle concerning which properties can feature in the contents of perceptual experiences. The chapter then concludes by considering the relationship between cognitive processing and phenomenal consciousness. In the previous chapter, we explored Tyler Burge’s arguments that disjunctivism, and hence naïve realism, is inconsistent with basic tenets of vision science. In the present chapter, we will look more closely at the interplay between philosophy and the empirical sciences of the mind (a set of disciplines that are often brought together under the heading Cognitive Science).

Cognitive Science and the Information Processing Paradigm In the opening chapter, we explored the way in which the development of mechanistic explanations in science influenced the Modern philosophers, which in turn played a role not only in setting the background understandings of the problems in the philosophy of mind and perception, but also in the nascent discipline of psychology. For example, in one of the earliest works in what we now know as psychology, Hermann von Helmholtz investigates the workings of our sense organs and concludes that “it is invariably simply the nervous stimulations that are perceived directly … but never the external objects themselves” (1867: 1059). This claim

138  Perception and the Sciences of the Mind

clearly has similarities to the conclusions of the Modern philosophers, but from a psychological perspective, the problem of perception was subsequently framed as one of how the brain could go on to work out the layout of the world from the clues contained within these nervous stimulations. A recent example of this picture can be found in the work of Jakob Hohwy, who illustrates the brain’s basic perceptual task by way of an analogy to a person trapped in a room with no access to anything outside of the room, but needing to figure out the source of a tapping noise (2013: 15). This deduction is challenging, however, for two reasons. On the one hand, many different external causes could give rise to the same patterns of sensory stimulation (consider large objects at a distance and smaller objects close to; trompe l’oeil pictures, holograms, or other representations of an object and the object itself). This means that “the information available to the visual system underdetermines the true state of affairs in the world” (Pylyshyn 2003: 94, my emphasis). However, we might also note that one and the same external cause could give rise to wildly different patterns of sensory stimulation (consider an object seen close up, at arm’s length, and in the far distance; from the front, back, side, top, and bottom; in bright light, artificial light, and twilight; and so on). This means that it is not straightforward, given a particular pattern of sensory stimulation, to identify one unique cause that explains that stimulation. The question then becomes: how does our visual system select the correct interpretation on the basis of this input? Given the stability and appropriateness of our visual experiences, it appears that our visual system does do this; so on this view of the nature of the problem of perception, it is the job of vision science to understand how the visual system solves this problem. In the latter half of the twentieth century, Ulrich Neisser – the founding father of cognitive psychology – added a critical metaphor with his suggestion that “the task of a psychologist trying to understand human cognition is analogous to that of a man trying to discover how a computer has been programmed” (1967: 6). The methodology subsequently adopted by cognitive psychology was thus to treat the mind as an information processing system that contains a range of internal representational states, and then to address the problem of perception by asking how the visual system, taking only the retinal image as input (which itself can be considered as a representation), succeeds in producing something like what Marr calls a “3-D model representation”: an object-centered description of the objects around the perceiver (1982: 37). The classical answer to this problem, which has its origins in the work of von Helmholtz, is that the visual system infers the layout of the world “on the basis of (usually implicit) knowledge of the correlations that connect” features of the world to features of the retinal image (Fodor and Pylyshyn 1981: 168). For example, when faced with the kind of retinal image that could be caused by a concave object lit from below or a convex object lit from above, the visual system will usually infer that it is a convex object lit from above on the basis of the implicit assumption that light typically comes from above. This combination of ideas  – incorporating a way of understanding the problem of perception (as illustrated by the underdetermination problem), a way to understand how the visual system functions in broad terms (the information processing paradigm), and the outlines of how this particular problem is solved by a visual system thus understood (involving a

Perception and Vision Science  139

bottom-up, stimulus-driven process, which makes use of implicit assumptions that are hard-wired into the visual system to produce a representation of the subject’s surroundings) – continues to inform a significant body of work in psychology as well as, as we have seen, influencing the intentionalist/representationalist paradigm in the philosophy of perception.

Predictive Processing Approaches An important recent development that endorses this overall approach builds on the work done by empirical scientist Karl Friston and his colleagues. Jakob Hohwy (2013; see also Clark 2013, 2017) developed a view that treats perception not as the registration of sensory stimulation, but rather as the result of the brain’s attempt to minimize its prediction errors. Imagine lying in bed and hearing a low growling, rumbling noise, but being too tired to actually get up and investigate, so just trying to figure out what might be causing the noise from the confines of your warm bed. This kind of case, Hohwy suggests, is analogous to the problem the brain faces in perception: we have to identify the cause of sensory input based only on those incoming sensory signals. Given this, we can use this analogy to explain key aspects of the Bayesian position Hohwy goes on to develop. The first is that of a hypothesis – a possible explanation of the cause or source of the noise. Of course, there would be numerous possible hypotheses we might develop – perhaps it is the refrigerator (although the refrigerator has never sounded quite like that before), perhaps it is an inquisitive lion just down the hall, perhaps it is something else entirely – so the problem then becomes: how to choose the right hypothesis. The next key feature is that of the likelihood of each hypothesis – for each possible hypothesis, what is the probability that this particular event would indeed cause precisely that noise that the person is aware of? Importantly, this is not a measure of how plausible the hypothesis is, just a measure of how well the hypothesized cause would explain the effect. To see the difference, consider the inquisitive lion hypothesis and suppose that the noises you can hear sound just like the noises an inquisitive lion was making on a recent nature documentary you watched. This hypothesis would therefore have a high likelihood, in that it would account for the particular noises heard extremely well; based purely on the nature of the noise and what you know about the sounds made by inquisitive lions, this hypothesis would accommodate the data very well. However, roaming lions not being very common in the part of the world in which you reside, this hypothesis would be implausible for other reasons. To capture this aspect of how antecedently plausible each hypothesis is, we need another notion: prior probability. Given what we know about the world, we can assign a prior probability to each hypothesis to sit alongside its likelihood. In this case, although the likelihood of the inquisitive lion hypothesis is higher than the likelihood of the refrigerator hypothesis – because the noise sounds just like an inquisitive lion, but you’ve never heard the refrigerator make a noise quite like this before – the prior probabilities are reversed: given where you live, it is far more likely that the refrigerator is making an unusual noise than it is that a lion has ventured into your house. These features can then be used to assign a posterior

140  Perception and the Sciences of the Mind

probability to each hypothesis using a simplified version of Bayes’ rule: essentially, we find the posterior probability of each hypothesized explanation of the noise by multiplying its likelihood by its prior probability. The rational inference to make is then that the evidence – the noise – is explained by the hypothesis with the highest posterior probability. (In this case, given the extremely low prior probability of the inquisitive lion hypothesis, we would probably expect the refrigerator hypothesis to come out on top.) Against this background, Hohwy states the core idea behind his theory: in this kind of way, the “brain infers the causes of its sensory input using Bayes’ rule – that is the way it perceives” (2013: 18). Hohwy is clear that when he uses the word “perceives”, he is talking in a phenomenological sense. To illustrate, he uses an example of binocular rivalry. In this experiment, a picture of a house is presented to one eye and a picture of a face to the other. What are the hypotheses that could account for this visual input? Hohwy suggests three: that the world contains a face, a house, or a face-house blend (2013: 21). Although the face-house blend hypothesis has the highest likelihood, in the sense that it best explains the sensory input, it has far and away the lowest prior probability, in that everything we know suggests that face-house blends are not found in the world. Given this, although the face-only and house-only hypotheses have lower likelihoods, in that they don’t explain this particular sensory input quite so well, their higher prior probabilities mean that they have higher posterior probabilities than the face-house blend hypothesis. So where a bottom-up theory might predict that subjects would combine the low-level sensory stimulation in the way that best explains the input – that is, to a face-house percept – the Bayesian inference theory thus predicts that subjects will fix on either the face or house hypothesis and thus see either a face or a house. And indeed, this is what subjects report: they report their phenomenal experience alternating between perceptions of a house and perceptions of a face rather than being a stable house-face amalgamation. And as Hohwy is careful to point out, this account is at the level of phenomenology, not merely conceptualization (2013: 25–6) – subjects’ phenomenal experiences switch, it is not that they have a stable phenomenal experience about which their judgments alternate. The content of the winning hypothesis thus explains the phenomenal aspects of perception. To this extent, Hohwy’s view looks to be broadly in the representationalist camp as far as the phenomenology of experience is concerned, just with a distinctive account of how the contents that underpin phenomenology are fixed. Yet we might worry that this way of fixing perceptual contents is as of yet underexplained. The appropriateness of our inferences – and hence the veridicality of our experiences – is kept in check by the assignment of prior probabilities to the various hypotheses: if the prior probabilities are not appropriate, then the system could spiral off into wild inferences that do not fit the world appropriately (and thus engender illusions and hallucinations). This theory therefore turns on the assumption that the prior probabilities are appropriate, but this generates a further problem: we are not simply programmed with suitable prior probabilities at the outset but must come to them using the information we receive in perception. So we might wonder how we come to an appropriate assignment of prior probabilities to hypotheses?

Perception and Vision Science  141

To answer this question, Hohwy introduces the key notion of prediction error. When the visual system decides upon a hypothesis about how the world is, it also makes a prediction about the sensory stimulation that should be received if the hypothesis is accurate and the world is as it is hypothesized to be. This prediction is then compared with the actual incoming sensory stimulation to yield a prediction error. This is a way of field testing the hypothesis. As Clark puts it, this picture “offers a form of self-supervised learning, in which the ‘correct’ response is repeatedly provided, in a kind of ongoing rolling fashion, by the environment itself” (2017: 18). If there is a prediction error, this is then used to revise the hypothesis so as to minimize the prediction error. This revised hypothesis is then tested in the same way and so on. In this way, the goal is that the brain arrives on a hypothesis that generates a minimal prediction error and cannot be improved: If this is all done in a reasonably optimal fashion, the resulting hypothesis must then be close to the real world’s state of affairs, at least as signaled in the sensory input …. This iterative edging towards accurate models makes sure that the perceptual inference is not going around in the threatening explanatory circle … Hypothesis revision guided by prediction error signals shapes prior [probabilities], and makes sure I do not simply see what I predict and predict what I see: the reason is that the prediction error is an objective corrective caused by the objects in the world, which shapes inference. (Hohwy 2013: 45) This feature of the view also enables it to explain the alternation in the binocular rivalry experiments (Hohwy 2020: 7): regardless of whether the prevailing hypothesis is house or face, the nature of the experimental set-up will ensure that there will always be prediction errors – that is, features of the incoming sensory stimulation that do not fit the hypothesis. It therefore makes sense, after a while, for the system to try the alternative hypothesis, which will switch the subject’s phenomenology; yet of course, as there will also be prediction errors for this hypothesis, this would lead to further switches in the future. Although there is a great deal more empirical and mathematical complexity behind this theory than we can cover here – and a far more significant role for active engagement than the treatment above might suggest – we can nonetheless see the ways in which this picture is an intriguing inversion of the traditional “bottom-up” view of perceptual phenomenology. Instead of thinking of sensory stimulation as providing the basic material for the construction of a phenomenal experience (be that by fixing what is represented by perceptual states, what aspects of the world are selected, or by generating qualia or sense data), this approach suggests that it is the operational hypothesis about the expected sensory stimulation – the hypothesis that performs best at minimizing error – that fixes what appears in phenomenal consciousness, and the incoming sensory stimulation itself is only used to identify errors in this prediction: “You open your eyes and see a bird flying past: this perceptual content arises by predicting the sensory input and, on the fly, being sensitive

142  Perception and the Sciences of the Mind

to how the prediction error landscape changes” (Hohwy 2013: 49). On this view, incoming sensory information isn’t used to inform us about what is in the world, but only about what in the world is surprising – what we are not expecting or have not predicted. As we have already noted, Hohwy presents this theory in broadly representationalist terms, so we might wonder: would accepting this view of the way the brain processes sensory information commit us to a representationalist theory of the kind discussed in Chapter 3? Zoe Drayson (2018) explores this question in detail, and notes that the proponents of this view have in fact offered a range of claims in this space. For instance, although as we have seen Hohwy sounds broadly representationalist, at other points he does suggest that perception is “at one remove from the world” (2013: 48) and, in earlier work, suggested that we literally perceive “the brain’s best hypothesis” (2007: 322), which makes it sound as though the hypothesis itself is an intermediate object of perception. These claims can make it sound as though the contents of the hypotheses are themselves objects of perception, which runs counter to most representationalist thinking. Clark, in contrast, flirts with (although ultimately falls short of) suggesting that this approach could underpin naïve realism by explaining how “beings like us are able to achieve genuine access to the causal structure of our environment” (2012: 767). On the basis of this, Drayson suggests that these pronouncements are therefore revealed as metaphysical speculations about, rather than implications of, the empirical findings. Her conclusion reflects one that we have already seen – that the science alone does not decide these philosophical questions: [E]ven where scientific theories are informing metaphysical claims [about the nature of perception], the latter cannot be simply ‘read off’ the former. … If predictive theories of perceptual psychology support a particular metaphysical approach to perception, this support must be established by argumentation that has yet to be provided. (2018: 3258)

J.J. Gibson and Ecological Psychology Even during the early stages of the development of cognitive psychology, there have been dissenting voices. For example, the father of “ecological” psychology, J.J. Gibson, rejects the underlying conception of the problem of perception – that the information available to the visual system underspecifies the layout of the environment (1966, 1972). Yet he doesn’t claim (implausibly) that the information contained in a single 2D pattern of light can specify a unique 3D interpretation. Instead, he suggests that this misunderstands what the visual system is trying to do: the role of the visual system is not to be stimulated by the light that impinges upon it and then draw inferences from that stimulation to the layout of the world, but to pick up on the information present in the light available to it (what Gibson calls the “ambient optic array”). To do this, the perceiving creature must “perform sample-taking in order to perceive the environment” (1972: 83), and it does so by moving its eyes,

Perception and Vision Science  143

head, and body to enable it to access enough information to specify a unique layout of the environment. Moreover, in addition to conceiving of the perceptual process differently, Gibson also developed an alternative action-and-exploration-based picture of the objects of perception, arguing that an important role for perception is to pick up on relevant “affordances” – the opportunities for action offered to animals by their environment – that are specified in the ambient optic array (1979: 127–41). In outlining the basic structure of the standard information processing approach, we saw that the traditional understanding of the problem of perception was one of discovering how, using only the (underdetermining) retinal image as input, the visual system could construct a unique internal representation of the perceiver’s surroundings. Gibson argues however that these considerations show that this is unnecessary. Instead, the ambient optic array contains all the information that the perceiver needs, so long as the perceiver is mobile and can sample this array, and also that this information includes information about the affordances – the opportunities for action – that the environment offers to the perceiving creature. As Anthony Chemero puts it, “if the information required to guide behavior is available in the environment, then organisms can guide behavior just by picking that information up” (2009: 111). There is thus no need for the organism to construct an internal representation of this information. In addition to undermining the idea that perceivers need to construct representations of their environment, there are also reasons to think that we don’t. To introduce this idea, consider the following passage from Daniel Dennett in which he outlines an experiment you can perform for yourself: Take a deck of playing cards and remove a card face down, so that you do not yet know which it is. Hold it out at the left or right periphery of your visual field and turn its face to you, being careful to keep looking straight ahead (pick a target spot and keep looking right at it). You will find that you cannot tell even if it is red or black or a face card. Notice, though, that you are distinctly aware of any flicker or motion of the card. You are seeing motion without being able to see the shape or color of the thing that is moving. Now start moving the card towards the center of your visual field, again being careful not to shift your gaze. … You will probably be surprised at how close to center you can move the card and still be unable to identify it. (1991: 53–4) The moral Dennett draws from this experiment is that the perceptual system doesn’t provide rich color and shape information about the peripheries of the visual field. This experiment hints at the possibility that, despite its seeming to be the case that we are aware of the scene in front of us in rich, full-color detail – and hence that the visual system must have constructed a rich, full-color representation of our surroundings – this might not be the case. A variety of recent experiments have built on this observation and discovered that subjects can often fail to see some aspect of their immediate environment. One of these phenomena is known as change blindness. In a well-known change blindness

144  Perception and the Sciences of the Mind

experiment, subjects view alternate presentations of a picture of a natural scene and a modified version of that picture. The kinds of modification include such things as significant objects appearing and disappearing, such as an airplane’s engine or a garage, or significant portions of an object changing color. While these images are cycling, the visual system is prevented from seeing the change happen by techniques such as adding a flicker – a brief period of no image – between the presentations or by ensuring that the changes take place during subject saccades (eye movements). Yet even if viewing conditions are optimal – for example, if subjects are told that they are studying the images for a memory test (to ensure they are concentrating on features in the image) and even told that the scenes depicted may change (so changes are expected) – the changes can go undetected for a significant length of time. Intriguingly, this phenomenon can also be found in real-world interactions. To demonstrate this, Simons and Levin (1998) set up two experiments in which a pedestrian was approached by an experimenter asking for directions. After these two individuals had interacted for 10–15 seconds, two people carrying a door passed between them, and whilst the conversation was thus interrupted, the experimenter who had initially asked for directions switched places with one of the experimenters carrying the door. Even though the two experimenters differed in height, voice, clothing, and so on, in 66% of trials (8 of 12), subjects continued the conversation and when asked if they had noticed anything unusual, did not report noticing the change. A closely related phenomenon is inattentional blindness (Mack and Rock 1988). In experiments studying this phenomenon, subjects are given an attention-demanding task to perform. In one example, participants are shown a recording of two intermingled teams of basketball players – one wearing black, one white – each team passing a ball to one another. Participants are then asked to count the number of passes made by members of the white team to one another. Then, whilst the subject’s attention is occupied by this task, an unexpected event occurs – someone dressed in a (black) gorilla suit walks across the court, beating their chest midway. As with the change blindness experiments, if asked beforehand, we would probably expect such an event to be seen immediately. Yet because attention is otherwise engaged, almost half of the subjects of this experiment failed to see the gorilla cross the basketball court (Simons and Chabris 1999).

Philosophical Prediction of Scientific Discoveries Given that this chapter is focusing on the interaction between philosophy and science, it is interesting to note that the phenomenon of change blindness was actually predicted by Dennett (1991: 468). So another way in which philosophy and empirical science can interact is through philosophical theory suggesting testable consequences. Noë argues that these results give us reason to think that perception does not involve a “process whereby a rich internal representation of experienced detail is built up” (2005: 50). In the change blindness cases, for example, we might think

Perception and Vision Science  145

that if the brain had constructed a representation of the entire scene, this could then be compared with the newly altered scene, thus enabling the change to be easily identified. The fact that the changes are in fact quite challenging to see suggests instead that perhaps this kind of rich, detailed representation of the perceiver’s surroundings is not constructed by the visual system after all. Of course, as before, these findings do not refute the idea that perception involves the construction of internal representations. For one thing, the interpretations of these results can be challenged – perhaps detailed representations are constructed but are not remembered (Pani 2000; but see Rensink 2000). For another, even if we accept that the findings show that the brain does not create rich internal representations, this is not to show that the brain does not create representations at all. However, it does at least open up the possibility that this is not how things have to be, and that the assumption that this is the way the brain works should not be taken for granted.

4E Cognition Views that depart from traditional cognitive science in this way are increasingly brought under the umbrella heading of “4E cognition”, where the four Es stand for “embodied, embedded, extended and enactive” (Newen et al. 2018: 6). The claim that a cognitive process is embodied signifies that it involves bodily processes outside of the subject’s brain in some way. Suppose, for example, that I can only successfully calculate the nine times table if I use my fingers; if this is the case, then it looks like my ability to calculate 9 × 6 (a cognitive process) is dependent upon processes going on in my hands and fingers. If the claim is that this cognitive process is causally dependent upon these bodily processes, then we can say that it is weakly embodied; if the claim is that this cognitive process is partially constituted by these bodily processes, then it is strongly embodied. Or suppose that I can only perform a certain type of calculation – long division, say – if my extra-bodily environment contains objects I can utilize, like pencil and paper. If my ability to divide 972 by 27 is causally dependent upon my use of pencil and paper, then we can say that this cognitive process is embedded in the environment. If we think that it is partially constituted by these aspects of my extra-bodily environment, then we can say that the cognitive process is extended in that the cognitive process itself reaches out into the environment. And if a cognitive process depends upon my ability to act in the environment, then this cognitive process can be said to be – again, weakly or strongly, depending on whether the claim involves causal dependence or part-constitution – enacted or enactive. On this approach, Gibson’s ecological theory – with its emphasis on the fact that “retinal inputs lead to ocular adjustments, and then to altered retinal inputs, and so on [illustrating that perception] is an exploratory, circular process, not a one-way delivery of messages to the brain” (1972: 80) would therefore qualify as at least embodied as it fundamentally relies on the perceiver’s ability to sample different parts of the ambient optic array by moving the eyes and head, and potentially even enactive to the extent that it relies upon the perceiver’s ability to actively explore its environment. As we saw in the previous chapter, these ideas have been developed by Alva Noë, in conjunction with Kevin O’Regan, into the enactive sensorimotor contingency

146  Perception and the Sciences of the Mind

theory of perception. In the previous chapter, we explored the extent to which this qualified as a version of naïve realism in virtue of its making aspects of the external environment available to perceivers. In the present context, however, it is relevant inasmuch as O’Regan and Noë (2001) argue that this action-based picture of visual experience, which treats perception as a temporally extended kind of interactive engagement with the world, can show how the construction of a rich, complete, internal representation of the environment is unnecessary. On this view, we are not consciously aware of what the brain represents, but rather are consciously aware of what is made available by the various processes going on in the brain and body: our sense that we perceive the rich detail of the entire scene before us is not a matter of our having constructed a detailed internal representation of the entire scene, but is rather a matter of our having the capacity to visually access rich detail concerning the aspects of that scene to which we are attending alongside our ability to visually access any particular aspect of the scene through active sampling. O’Regan and Noë’s focus on perception as a form of active engagement with the environment therefore challenges a core assumption of the information processing paradigm – that the task for the perceptual system is to construct an accurate representation of the environment in a context in which the input to that system underdetermines the layout of the world. If we accept that perception is a temporally extended activity, then the information the perceptual system has access to does not underdetermine the layout of the environment after all. If we accept that this activity involves accessing detail about the scene, then it suggest that the perceptual system need not construct an internal representation of the environment to ensure our behavior is appropriate. In the words of Rodney Brooks, the “key idea here is to be using the world as its own model” (1991: 148).

Representationalism and Ecological Theories Although as we saw in the previous chapter, Noë appears to align this view with a naïve realist approach, others have attempted to accommodate a broadly Gibsonian picture within a representationalist paradigm. For example, Nico Orlandi (2013) develops a version of an Embedded View that rejects the claim that the visual system constructs a representation of the environment using something like a process of inference yet allows that the visual state itself is representational (13–14). To motivate this position, Orlandi argues that in order to avoid claiming that the implicit assumptions employed by the visual system are simply innate (and hence ultimately unexplained), we must hold that they are grounded in the way the environment is – we implicitly assume that light comes from above because, typically, light does come from above. Yet if this is so, the “visual system does not need to know about them in any sense of knowing in order to produce the percepts that it produces. It can simply be wired by the environmental fact in question to produce” a representation of a convex object lit from above (41).

Perception and Vision Science  147

Perception and Cognition In a recent paper, Chaz Firestone and Brian Scholl suggest that the “challenge of understanding the natural world is to capture generalizations – to ‘carve nature at its joints’” (2016: 1). The suggestion here is that the aim of science is to identify the fundamental divisions within nature. For those sciences that study the mind, then the hope would be that cognitive science and its constituent disciplines will identify fundamental divisions within our understanding of minds and mental processes. In this light, Firestone and Scholl go on to ask: What are the joints of the mind? Easily, the most natural and robust distinction between types of mental processes is that between perception and cognition. This distinction is woven so deeply into cognitive science as to structure introductory courses and textbooks, differentiate scholarly journals, and organize academic departments. (2016: 1) Not only is the distinction between perception (or sometimes sensation) and cognition central to cognitive science, it also has significant intuitive support. Recall our discussions of nonconceptual content in Chapter 3; there we saw that there is a strong intuition that what we can see somehow overflows or outstrips our ability to think about what we see. If “perception” names the purely stimulus-driven aspects of the overall perceptual process and “cognition” names those parts of the overall process in which cognitive capacities can be brought into play, then this intuition says that perception delivers more information than can be processed in cognition. For there to be a meaningful distinction between perception and cognition in this sense, it will need to be the case that “although what is commonly referred to as ‘visual perception’ is potentially determined by the entire cognitive system, there is an important part of this process … that is impervious to cognitive influences” (Pylyshyn 1999: 342). After all, if all parts of the perceptual process are influenced by cognition, then on what basis could it be claimed that there is a natural “joint” between the two types of process? Given this, the claim that there is an important distinction between perception and cognition appears to be tied to the claim that at least some aspect of perceptual processing – Pylyshyn calls it early vision – that is “informationally encapsulated” (Fodor 2000: 58) or impervious to cognitive influences. Interestingly, Pylyshyn has a relatively robust conception of what counts as being pervious to cognitive influences. He explains what it means for a system to be influenced by cognition – to be “cognitively penetrable” – as follows: If a system is cognitively penetrable then the function it computes is sensitive, in a semantically coherent way, to the organism’s goals and beliefs, that is, it can be altered in a way that bears some logical relation to what the person knows. (1999: 343) On this understanding of the perceptual process, it breaks down into two elements: early vision – involving processes that are encapsulated from or impenetrable

148  Perception and the Sciences of the Mind

by broader cognitive processes – and other processes that may, and in fact are (1999: 344), penetrated by cognition. Moreover, there is a relatively clear statement about what kinds of interaction would qualify as cases of cognitive penetration – cases in which the processes in early vision “can be altered in a way that bears some logical relation to … beliefs, expectations, values, and so on” (1999: 343). Within psychology itself, the idea that there is an impenetrable, encapsulated element of the perceptual process has been challenged. In an influential 1947 paper, Jerome Bruner and Cecile Goodman reported the finding that when asked to make the size of a variable patch of light match that of a coin, children systematically set the size of the light-patch larger than the coin, but that this result was not replicated when the children were asked to match the light to cardboard disks, despite those disks being identical in size to the coins. To explain this finding, Bruner and Goodman hypothesize that the value of the coin makes it seem larger to the children than it really is. Moreover, when they separated out the results of the tests performed by poor children from those performed by rich children, they found that the “poor group overestimates the size of coins considerably more than does the rich” (1947: 39), and they explain this finding with the suggestion that the poor children’s greater need for money – or the greater relative value of the coins – leads to an accentuation of this effect. Bruner and Goodman conclude on this basis that cognitive phenomena – such as the value the children place on the coin – shapes the way they perceive the coin even at the most basic level and thus that perception and cognition are not encapsulated, but are “continuous”. This finding – that there are “top-down” or cognitive influences on perception – has also been reported in numerous other studies. Fiona Macpherson (2012) discusses particularly interesting experiments by Delk and Fillenbaum (1965), in which participants were shown a range of shapes that had been cut out from the same orange paper against a background whose color could be adjusted: from yellow to red through orange. When subjects were asked to match the background to the color of the shape presented, they found that if the shape of the cutout was of a characteristically red object (such as a love heart, lips, or an apple), the subjects adjusted the background to be more red than if the cutout shape was of a more generic – not typically red – object such as a geometric shape, a bell, or a mushroom. This provides a case, Macpherson suggests, in which a subject’s beliefs – in particular, their beliefs about the typical color of objects – influence the phenomenal character of their color experiences. In more recent years, studies purportedly showing cognitive influences on perception have proliferated. In their 2016 paper, Firestone and Scholl refer to a website they maintain that aims to provide an exhaustive list of all papers that purport to show a top-down influence of cognition on perception published since 1995 (http://perception. yale.edu/Brian/refGuides/TopDown.html). As of its most recent update in September 2018, this resource listed 182 studies, reporting such varied findings as that: desiring an object makes it look closer (Balcetis & Dunning 2010), that reflecting on unethical actions makes the world look darker (Banerjee et al. 2012), that wearing a heavy backpack makes hills looks steeper (Bhalla & Proffitt

Perception and Vision Science  149

1999), that words having to do with morality are easier to see (Gantman & Van Bavel 2014), and that racial categorization alters the perceived lightness of faces (Levin & Banaji 2006). (Firestone and Scholl 2016: 2) Yet, considered as proof that the perceptual process is influenced by cognition, these studies face some important problems. First is the fact that other similar observations appear to point in the opposite direction. For example, in our discussions of belief acquisition theories in Chapter 3, we considered an objection based on the fact that our experiences of optical illusions – there we looked at the Müller-Lyer illusion – do not change when we learn that the illusions are misleading (Pylyshyn 1999: 344). If cognition was influencing perception, then we might expect knowledge that the lines are the same length to make them look the same length, but this is not what transpires. Moreover, as Firestone and Scholl point out, the “stickability” of this kind of illusion suggest that in such cases, features that look like cognitive penetration may actually be “wired into the visual module itself” (2016: 4). If this were the case, then some of the cases that seem to show cognitive penetration of visual experience could in fact be explained by the hard-wired assumptions that underpin von Helmholtz-style unconscious inferences. In addition, the kinds of claims listed above could be given different interpretations, some of which may be consistent with the view that perceptual processes are encapsulated. To see this, consider Susanna Siegel’s discussions of Keith Payne’s (2001) experiment, which found that priming subjects with pictures of black men (as opposed to white men) made them more likely to misclassify a tool, such as a pair of pliers, as a gun. Siegel points out that these results can be interpreted in numerous ways (2015: 419–20). One interpretation is of course the full “cognitive penetration” interpretation – that the pliers look to the subject like a gun, and that this is a result of the priming. However, it may also be the case that there are interpretations that are consistent with the pliers looking to the subject like pliers. If, for example, we have a view of perception which separates out phenomenal and conceptual aspects, then perhaps the subject has a perceptual experience with plier-like phenomenology, but the interpretative component of the perceptual process goes awry and leads to the experience having a conceptual content involving the concept . This interpretation would allow that cognition influences perception, but nonetheless holds that there is a component of the perceptual process – perhaps corresponding to Pylyshyn’s early vision – that is encapsulated from these influences. Or perhaps the priming effects do not impact upon the perceptual process itself at all, but lead the subject to form an erroneous belief on the basis of perception, or make an introspective error about what they see or to misremember what they saw. Another possibility, suggested by Pylyshyn, is that many cases of cognition influencing early visual processing are legitimate, but do not operate in a way that undermines the hypothesis of informational encapsulation. He argues that cases involving an aspect switch – such as the vase-face illusion, the Neckar cube, or Jastrow’s duck-rabbit – can be explained by cognition directing “the allocation of attention to certain locations or certain properties prior to the operation of early

150  Perception and the Sciences of the Mind

vision” (1999: 344). This would allow cognitive features to affect the resulting phenomenal character of an experience without directly impacting upon the formation of that phenomenal character. To see how, consider the difference between a car driver and a railway points operator deciding whether the relevant vehicle is to go left or right. If the car driver wants to turn left, they have to turn the steering wheel that way; the driver’s actions are naturally seen as part of the car’s turning left. If the points operator wants the train to turn left, however, they just need to set the points. This can happen a long time before the train gets there; then, when the train does arrive, it simply follows the predetermined direction. In this case, the points operator’s actions are not part of the train’s turning left; rather, they are something that is, strictly speaking, external to that process, yet which has causally influenced it. This is analogous to the claim that cognitive processing impacts upon phenomenal character by determining the allocation of attention, which then causally affects which low-level properties are subsequently perceived. However, defenders of the view that vision can be cognitively penetrated have argued that whilst this might be true for overt attention – the kind of attention that is manifested by directing one’s eyes (or head) in a certain direction – it is not the case for covert attention, in which attention shifts happen without any outward behavioral manifestation. Wayne Wu argues that the mere intention to act on a specific object – with intention being a cognitive state – directly affects the way in which information is computed within the visual system, leading to covert attention shifts, which then, in turn, facilitate shifts of overt attention: think of the visual system as keeping an eye out for information about the animal’s current goals so as to shift its processing to better serve those goals. When it learns of the intended target, it selects that target for further taskrelevant processing. (2017: 23) In considering such cases, however, Firestone and Scholl argue that this does not qualify as cognitive penetration in Pylyshyn’s sense. Even if attention does indeed influence early visual processing, it does not do so in such a way that the outputs of that processing bear any relation to the contents of the intention. As they put it: attention may enhance what you see regardless of the reasons that led you to deploy attention in that way … such attentional … effects may be occasioned by a relevant intention or belief, but they are not sensitive to the content of that intention or belief. (2016: 14)

Cognitive Penetration and Reentrant Processing What these discussions suggest is not only that our intuitions can point each way, but that there are also ways of interpreting the experimental findings in line with each approach. How might we resolve this? Perhaps the answers to these questions

Perception and Vision Science  151

might be found in empirical study of the neural structures underpinning perceptual processing and the way incoming visual information is processed in and propagated through neural areas. When it comes to neural structure, however, the growing consensus is that there is a great deal of what is known as “re-entrant” processing, in which neural activity from higher brain regions can modulate the activity of lower regions (e.g. Rolls 2008; Gilbert and Li 2013). Defenders of the encapsulation thesis have disputed that this is evidence of cognitive penetration, however, arguing that the same “areas that process in a bottom-up way sensory specific information are also involved in higher-level activities (voluntarily attention-driven search, imagery, concept formation), except that in the latter case they are reentered in a top-down manner” (Raftopoulos 2001: S194). So the presence of reentrant processing only shows that top-down information can influence processing in perceptual areas, not that it always does. And when we focus on the way visual information is processed, Raftopoulos contends that we can see that there is a component of the perceptual process that is not cognitively penetrable (2009: 131–2). He follows Lamme in describing the progression of information processing in the brain as involving three key stages. The first is a feedforward sweep (FFS) in which an incoming signal is transmitted from the lower levels of the brain to the higher levels, with no information travelling in the opposite direction. This process, he suggests, can begin to extract “high-level information, which is sufficient to lead to some initial categorization of the visual scene” (2009: 132). The next phase is localized recurrent processing (LRP), in which information does flow in both lateral and top-down directions, but only within early vision. As this is “restricted within early vision, [it] does not involve memory or other cognitive functions” (2009: 132), and hence does not constitute cognitive penetration. Only after the LRP phase do we find the onset of global recurrent processing (GRP), which brings higher cognitive centers into play. Of course, even if this is correct as a timeline of visual processing, there remains the question of how these different stages of perceptual processing map on to more philosophical domains. For example, which stages of perceptual processing correlate with conscious awareness of objects? Raftopoulos himself aligns the different stages of processing with the familiar categories of sensation and perception: In neuroscientific terms, sensation consists in those processes that belong to Lamme’s feedforward sweep. … The processes that transform sensation to a representation that can be processed by cognition constitute perception. The output of these processes is a cognitively impenetrable content that is retrieved from a visual scene in a bottom-up way. A subset of this output, consisting of that which can be brought to a kind of awareness called phenomenal awareness, is the “phenomenal content”. (2009: 133) These two processes – sensation and perception, in Raftopoulos’s sense – make up the impenetrable part of the visual system Pylyshyn called early vision (2009: 134). However, although these nonconceptual outputs of early vision are held to

152  Perception and the Sciences of the Mind

be phenomenal, Raftopoulos claims that they are “not easily reportable” (2009: 132). The kind of awareness that is easily reportable – the kind that “accompanies our normal experience” (2009: 132) – only arises with the onset of GRP and the widespread influence of cognition. Yet for those who think that conscious visual perception is penetrated by cognition, this might seem open to an objection: why would it matter if there is indeed a stage of the neural process involved in perception that is encapsulated from cognitive influence if it is nonetheless conceded that the phenomenally conscious aspects of our perceptual experiences are cognitively penetrated?

Why Do We Care about Cognitive Penetration? As we have seen, Pylyshyn defines cognitive penetration as requiring visual processing to be impacted in quite a particular way – in a way that bears a logical relation to cognitive states, such as beliefs, desires, and values. As we saw when discussing Wu’s claims that intention can enhance perceptual processing, it is therefore possible that cognition might impact upon visual processing without this counting as cognitive penetration. Yet Pylyshyn’s is not the only definition of cognitive penetration around. For example, Susanna Siegel (2011: 205–6) and Fiona Macpherson (2012: 28) offer broadly similar accounts, according to which if two subjects (or the same subject at different times) have the same “inputs” to perceptual processing – what is perceived, the viewing conditions, and the state of the sensory organ (Siegel also adds the direction of the subject’s attention, overt or covert) – yet nonetheless have different perceptual experiences on account of differences in their cognitive systems, then these experiences would qualify as cognitively penetrated. Likewise, Dustin Stokes offers a definition according to which a perceptual experience counts as cognitively penetrated so long as there is an internal, mental causal link between a cognitive state and the perceptual experience (2015: 78). In none of these cases is there a constraint placed on the rationality of the effect that cognitive processes must have on perception/perceptual processing. So where Pylyshyn requires the impact of cognition to bear a logical relationship to beliefs, desires, or values, these theorists do not. As long as cognitive differences lead to perceptual differences, so long as they do not do so by altering anything in the input conditions, then we are dealing with a case of cognitive penetration. On these definitions, cases of intention enhancing what is seen would qualify as cases of cognitive penetration. In a more recent paper, Susanna Siegel therefore concludes that “there is no single phenomenon of cognitive penetration” and hence that “it is useful to postpone the existence and clarification questions until one or other theoretical purpose for the notion of cognitive penetration has been specified” (2015: 406). If we identify why we care about cognitive penetration – what problematic consequences there would (or wouldn’t) be if cognitive penetration were actual – then we can use this to define “the phenomenon of interest [as] a cognitive-perceptual relation that implies one or more of these … consequences” (Stokes 2018: 305). Pylyshyn’s original interest in the question of whether or not perception is cognitively penetrable appears to have been driven by the desire to demarcate the proper scope of cognitive science (see Stokes

Perception and Vision Science  153

2015: 77). If cognitive science is the study of computational mental processes and if a computational process involves computation over internal representations, then computational processes will require access to a system of internal representations – that is, they will be cognitively penetrable. So in rejecting the suggestion that early vision is cognitively penetrable, Pylyshyn denies that early vision falls under the scope of cognitive science, instead being a part of the underlying functional architecture. Within philosophy, one of the most significant reasons for interest in the question of whether perception is cognitively penetrable turns on epistemic considerations. One aspect of this concern is reflected in discussions in the philosophy of science on the role observation can play in decisions between scientific theories. In the 1960s, Hanson, Kuhn, and Feyerabend suggested that the scientific theory a scientist endorsed would literally affect the way that scientist saw the world – that empirical observations are “theory laden”. If this is true, however – if the theories one believes affect one’s perceptual experiences – then we might worry that the legitimacy of the scientific method of hypothesis and test has been undermined. If the theory a scientist endorses influence what they observe, then how could those observations be appealed to in support of that theory? And if two theorists coming from different theoretical standpoints describe the same events differently, how can they even communicate with one another to settle their disagreements? In this light, Raftopoulos suggests that arguments for cognitive penetrability are arguments for the theory-ladenness of perception, which, in turn, “clear… the way for relativistic theories of meaning and scientific theories” (2001: S187). To resist this conclusion, he therefore argues that we need to be able to maintain that “the content of perceptual states is free of cognitive modulation” for it to be the case that perception can provide: a theory neutral basis, in the sense that when persons with differing conceptual backgrounds view the same visual scene they perceive the world the same way, while remaining within their respective conceptual frameworks … This renders possible communication across paradigms. (2009: 313–4) As Jack Lyons points out, however, if it is admitted that perceptual beliefs are cognitively penetrated, which “even the staunchest defenders of encapsulation concede”, then it is unclear how this helps. If communication involves giving voice to the things that one believes, then the conceptual frameworks are going to have done their damage before potential cross-paradigm communication gets started (Lyons 2011: 291). A variant of this concern also appears within epistemology. The position Susanna Siegel calls “dogmatism” within epistemology claims that “absent defeaters, having a perceptual experience with content p suffices to give you justification for believing p” (2011: 208). Yet if my belief that p cognitively penetrates my experiences, thereby making it the case that I see that p, then this experience cannot stand as a justification of that very belief on pain of circularity. If it is the case that we both believe what we see and see what we believe, then we would seem to be

154  Perception and the Sciences of the Mind

trapped in a pernicious epistemic circle. Siegel illustrates the concerning possibility with the following example: suppose Jill believes that Jack is angry at her, and this makes her experience his face as expressing anger. Now suppose she takes her cognitively penetrated experience at face value, as additional support for her belief that Jack is angry at her (just look at his face!). … [So] while experience will seem to let you check your beliefs against the world, … really you’ll just be checking your beliefs against your beliefs. The tribunal will be corrupted. (2011: 202) As Siegel notes (2011: 201), not all cognitive penetration will be pernicious – some may actually be epistemically beneficial. For example, we might think that the kind of expertise that characterizes a radiologist, forensic print analyst, or ornithologist would be manifested in part by their hard-earned knowledge enabling them to perceptually access information – in x-rays, fingerprints, or aviaries – that others are simply unable to see. Crucially, this process is truth-conducive – in enabling the expert to perceptually access “hard-to-see” information, they enable the expert to acquire knowledge that would not otherwise be accessible. For this reason, Dustin Stokes argues that this kind of expertise therefore constitutes an epistemic good. He overviews a range of experimental evidence that, he argues, shows that this kind of expertise enables those possessing it to “perform more accurately, more rapidly, with less cognitive effort, and in ways that present advantages for working memory” (2020: 18). Of course, opponents might argue that the kind of perceptual expertise appealed to here is either not strictly perceptual or is not a result of cognitive penetration – although Stokes explores a range of evidence to counter both of these claims (2020: 6–17) – but this doesn’t affect the evidence against the claim that cognitive penetration, should it occur, would have to be epistemically problematic. Moreover, Lyons point remains: if cognitive penetration is problematic, then it is unclear that identifying an element of visual processing that is encapsulated helps. “The view about the informational encapsulation of early vision is controversial, but my point is that even if it’s right, the epistemological problem remains” (Lyons 2011: 291). In this light, Lyons goes on to argue that cognitive penetration is not, in and of itself, problematic. Even in the kinds of case Siegel describes, he suggests, things may not be as problematic as she contends. Lyons argues that if the epistemic impacts of cognitive penetration are probabilistic – that is, they make it more likely that people will have experiences that correspond with the things they already believe – then this doesn’t entail that perception cannot be a source of information about the world at all; it’s just another way in which perception is a fallible source of information. So long as “objective facts are still significant determinants of what is observed”, then even if cognitive influences do play a role, it will just be another factor – along with such things as “poor observation conditions, camouflage, distraction, sleepiness, etc.” – that stops our perceptual access to the world from being infallible (2011: 293). If Jack is angry with Jill, and Jill’s belief that this is the case makes her more likely to visually pick up on Jack’s expressions of anger, then this could also be an epistemic virtue: Jill’s enhanced sensitivity enables her to

Perception and Vision Science  155

see things that she may have missed otherwise. It only becomes a problem, Lyons suggests, if the cognitive influence of her belief that Jack is angry with her is so pervasive that this belief ceases to be determined by her environment (2011: 301). For this reason, Lyon suggests that cognitive penetration itself is not epistemically problematic – unreliability is epistemically problematic – but that cognitive penetration is, along with many other factors, a potential source of unreliability.

The Admissible Contents of Experience A closely related question to that of whether perception is cognitively penetrable is the question of which properties can appear in phenomenal character. Those who lean toward the encapsulation thesis and the claim that there is a purely bottomup-driven component of the visual process are likely to lean toward a “conservative” view (Bayne 2009) of the properties that can appear in phenomenal character, claiming that they are completely exhausted by low-level properties. For example, Jesse Prinz argues that “the only things that make a phenomenal difference” are appearances, which he characterizes as involving only “colors, shapes [and] textures” (2006: 451–2). Those who think that cognition can influence perception, however, are more likely to take a liberal stance and think that phenomenal character can also involve “higher-level” properties, such as: • •

•

Kind properties, such as the property of “being a house” or “being a tree” (Siegel 2006: 483; see also Bayne 2009); Causal properties, such as the property of a boulder’s flattening a hut (Strawson 1992: 118) or, more generally, one thing’s causing another (Butterfill 2009; Siegel 2009). Generic properties, such as the property of being “five to ten rectangles arranged in a circle” (Block 2008: 307) or the property of being nonspecific text (Grush 2007: 504; Fish 2009: 64).

Admissible Contents in Non-Intentionalist Theories Although discussions about these issues often take place under the heading of the “admissible contents” of experience, this is not just an internecine dispute within the intentionalist camp. However, as each theory of perception has a different view of phenomenal character, each theory will frame this issue slightly differently. Where a representationalist will hold that phenomenal character “involves” such properties in virtue of their being represented, a theorist who endorsed the existence of sense data might say that phenomenal character “involves” low-level properties in virtue of their being instantiated by sense-data, and a relationalist would hold that such properties are “involved” in phenomenal character in virtue of the subject’s being acquainted with those properties. For present purposes, though, we will just ask the question in a theory-neutral way by asking which properties feature in phenomenal character.

156  Perception and the Sciences of the Mind

Answering the question about what properties feature in phenomenal character is therefore similar to theorizing about the Bruner-Goodman findings. Most people would agree that we can see trees, for example, so we might then ask: do we see trees by having properties such as the property of being a tree feature in phenomenal character, or are such properties merely part of the interpretative overlay that perception places on phenomenal character? To answer this question, we might start by introspecting our experiences, but as Susanna Siegel points out, this can only take us so far. When we see a bowl of fruit, introspection can tell us that the phenomenal character of this experience does not involve the property of being a blowfly or the property of being a bass guitar. But does it involve the property of being a banana? Introspection is ambivalent. In light of this, Siegel introduces us to a way of answering this question using what she calls the “method of phenomenal contrast” (2007). This operates as follows: Suppose you have never seen a pine tree before, and are hired to cut down all the pine trees in a grove containing trees of many different sorts. Someone points out to you which trees are pine trees. Some weeks pass, and your disposition to distinguish the pine trees from the others improves. Eventually, you can spot the pine trees immediately. They become visually salient to you.... Gaining this recognitional disposition is reflected in a phenomenological difference between the visual experiences you had before and after the recognitional disposition was fully developed. (2006: 491) In essence, the method of phenomenal contrast proceeds by finding two experiences that involve the same “low-level” properties, and that are experienced by subjects who possess the same concepts, save for one. In this case, the before and after experiences of the grove involve the same low-level properties as they are experiences of the same group of trees; the subject of these experiences is also the same person, who we can assume possesses the same concepts, save for the fact that the subject has now acquired a recognitional concept for pine trees. Against this background, we can then ask whether or not there is a phenomenal difference between the two experiences. Siegel’s argument is that introspection can at least yield the deliverance that these two experiences differ in phenomenal character, even if it cannot tell us which properties appear in phenomenal character. But then, this can be used to argue that phenomenal character involves high-level properties: given that the two experiences involve the same low-level properties (as they are experiences of the same scene), and given that the only difference between the subjects is that the later subject possesses the concept of being a pine tree, then the best explanation for this difference in phenomenal character is that the phenomenal character of the later experience differs inasmuch as it also involves high-level properties, such as the property of being a pine tree, alongside the low-level properties that were there in the first experience.

Perception and Vision Science  157

Siegel on the Visual Experience of Causation Due to considerations of space, I won’t discuss the arguments surrounding whether or not causal properties and/or generic properties can be involved in phenomenal characters in any detail. It is worth noting however that thus far, these arguments have also had an important role for empirical results. For instance, Susanna Siegel (2009) explores Michotte’s (1963) findings that subjects will often describe their experiences in causal terms, and then employs the method of phenomenal contrast to argue that these reports actually reflect the contents of the subjects’ visual experiences. As Siegel recognizes, this is at heart an appeal to intuition: it just seems right to say that there is a phenomenological difference between the two experiences before and after gaining the relevant capacity (she also gives an auditory example involving the experience of hearing words spoken in a language before and after learning that language). The problem with this approach, however, is that people just don’t seem to agree. For example, Jack Lyons insists in a parallel case (involving recognizing faces as opposed to trees) that “it is clear that faces don’t look – experientially – any different on becoming familiar ones” (2005: 244). So where Siegel thinks it is intuitive that there is a phenomenal difference in such cases, Lyons disagrees. This suggests that philosophical intuitions are only going to go so far in enabling us to answer questions like whether the kinds of differences appealed to as evidence of cognitive penetration are really phenomenal or whether cognition can penetrate phenomenal character. It is for reasons such as these that Pylyshyn describes phenomenology as “an egregiously unreliable witness” (1999: 362).

How Do We Explain Clashes of Intuitions? What might explain these differences in intuitions? One possibility is that it depends on the antecedent commitments different theorists bring to the table when considering such cases. For the reasons we have seen, someone antecedently inclined to think that perception involves sense data would be likely to think that only properties that could be possessed by sense data could appear in an experience’s phenomenology. This could then influence what they “discover” when they introspect their experience. Alternatively, theorists’ responses could be influenced by the models they use to understand experience. For instance, after claiming that only colors, shapes, and textures feature in phenomenal character, Jesse Prinz suggests that to claim otherwise would be “like saying that phenomenology includes features that could not be captured by a camera” (2006: 451–2). Perhaps the model a theorist uses for understanding experience or phenomenal character might itself influence what is “found” when engaging in introspection.

158  Perception and the Sciences of the Mind

Perceptual Consciousness and Reportability Earlier, when we discussed Raftopoulos’s claim that the nonconceptual outputs of early vision are phenomenal but not easily reportable, we asked why this would matter if other phenomenally conscious aspects of our perceptual experiences were cognitively penetrated. Indeed, the problem runs even deeper than this: if the outputs of early vision are not easily reportable, what is the evidence for the claim that they are phenomenal at all? Perhaps, we might think, our evidence for this claim turns on our ability to engage in fine-grained motor interactions toward seen objects. If we can pick up a coin or post a letter through a slot, then wouldn’t this show that we are phenomenally aware of the coin and slot? This question has been complicated by the discovery of two different pathways in the brain that are both dedicated to the processing of visual information. According to this “two visual systems” hypothesis, all visual information begins on the same path: it is passed from the retinas, down the optic nerve to the lateral geniculate nucleus (LGN) and from there to the primary visual cortex V1. From there, however, this information travels down two distinct anatomical pathways, each of which is dedicated to performing different processing tasks. On Ungerleider and Mishkin’s classic presentation (1982), one of these pathways – the what pathway – travels along the base of the brain (the ventral stream) and the other – the where pathway – travels up and over the top of the brain (the dorsal stream). Although the precise functions of these two pathways have continued to be discussed, their existence is now widely accepted. According to a more recent and highly influential treatment by Milner and Goodale (1995), the purpose of the ventral stream is to underpin object identification/recognition, whereas the purpose of the dorsal stream is to enable real-time visual guidance of action (they therefore suggest the pathways subserve what and how rather than what and where). Milner and Goodale contend (although this can be disputed) that conscious visual awareness is connected with only one of the two streams – the ventral (object identification and recognition) stream. This therefore predicts that subjects could have the capacity to act in appropriate ways to visually presented stimuli even if, due to localized damage, they failed to consciously see those stimuli. A famous example of this is Milner and Goodale’s patient D.F. who developed apperceptive agnosia – the loss of the ability to discriminate visual forms, such as the width of objects or the orientation of a slit – following damage to her ventral stream. In one experiment, D.F. was presented with a vertically mounted disk with a slot cut into it, which was set at different angles for different trials, and asked to report the orientation of the slot. In response to this task, Milner and Goodale found that her reports of the slot orientation were unaffected by its actual orientation – they were effectively random guesses. On their hypothesis, this showed that the ventral stream damage meant that information about the slot was just not making it through to the ventral stream processes responsible for object identification and recognition. However, when asked to post a card through the slot, D.F. had no difficulty – her actions were accurate, smooth, and appropriate.

Perception and Vision Science  159

This suggests that information about the slot’s orientation was making it through to the dorsal stream responsible for action. Similar results were found when she was asked to open her thumb and forefinger to illustrate the width of an object. For this task, she performed at chance, but when reaching to pick up the object, D.F. would smoothly adjust her grip aperture to an appropriate width. Interestingly, Haffenden and Goodale (1998) reported similar results in experiments involving normal subjects. Their experiment makes use of the Ebbinghaus illusion (below), in which a disk of a particular size surrounded by small circles is judged to be larger than a disk of the same size surrounded by big circles. Haffenden and Goodale discovered that, despite the presence of the illusion of size difference, if subjects were asked to pick up the central disks, they automatically moved their finger and thumb to the same distance apart in both cases. This suggests that whilst the ventral pathway is subject to the illusion, the dorsal pathway is not.

Figure 6.1 Although these findings are fascinating in their own right, they are of particular significance for the philosophy of perception because they challenge something that can seem so obvious to us. For example, take your finger and place it slowly on this X. Naturally, it seems to us that any changes in the direction or angle or speed of our finger is made in response to our conscious awareness of the X, our finger, and the relationship between them. Yet if Milner and Goodale are right, conscious perceptual awareness does not play this role. Whilst it might play the role of singling out the X as something to put your finger on, once action begins it is guided by nonconscious processes. What this suggests is that we cannot necessarily infer claims about how things look, phenomenally speaking, from behavioral evidence derived from our interactions with the objects of perception. As the Haffenden and Goodale experiment suggests, it would be inappropriate to conclude that subjects have a phenomenally veridical experience of the size of the central disk from data about their grip scaling when interacting with those disks.

Types of Agnosia The loss of a mental ability is known as a form of agnosia. Patient D.F. developed visual apperceptive agnosia, but there are many different kinds of agnosia. For example, in color agnosia (also known as achromatopsia) subjects

160  Perception and the Sciences of the Mind

lose the ability to see color, typically due to damage to brain area V4, which supports color vision. Similarly, damage to area MT can result in motion agnosia (akinetopsia): the inability to see motion. Other interesting agnosias include prosopagnosia – the loss of the ability to recognize familiar faces; simultagnosia – the loss of the ability to simultaneously recognize more than one object; associative agnosia – the loss of the ability to recognize objects; and unilateral neglect, in which damage to the right hemisphere of the brain leads subjects to neglect or overlook any visual information coming from the left visual field. In psychopathology, studies on patients with deficits of this kind can provide important information concerning how different areas of the brain support different facets of our perceptual abilities. In philosophy, they also raise a range of interesting questions about the nature of conscious perceptual awareness. For example, in akinetopsia/motion agnosia, sufferers report seeing people disappear from one place and reappear in another – they cannot see them moving from one place to another – and when filling a teacup, sufferers see the liquid level jump from level to level rather than rise gently. Yet it is only moving objects that are perceived in this way – the experience of the parts of the scene that do not move don’t become jerky, even when an object in the scene is moving. Cases such as this raise interesting questions for how this should be captured in a philosophical theory of perception.

When it comes to thinking about how we might engage in empirical studies of the question of whether perception can be conscious in the absence of cognition, these results appear to make one source of evidence – behavioral evidence – problematic, in as much as they suggest that appropriate behavior could occur in the absence of conscious awareness. What other evidence might we appeal to? The most obvious source of evidence we have concerning both the presence of and nature of our conscious experiences is our first-person reports about consciousness. Yet when we are interested in finding out more about the relationship between perception and cognition, and in particular, whether there could be conscious perception in the absence of cognition, we run headlong into a methodological puzzle. As Ned Block points out, the most “natural methodology” for investigations of this kind “is to find the neural [activity that is taking place] in clear cases and apply it to the problem cases where for some reason there is no cognitive accessibility of the experience” (2008: 292). We might think therefore that all we need do is to identify the neural activity occurring in clear cases of conscious experience, and then investigate whether activity of that kind is also occurring in the absence of any activity in the subject’s cognitive centers. Yet the clearest cases of consciousness are those in which we know that the subject is having a conscious

Perception and Vision Science  161

experience because they tell us that they are – that is, the conscious experience is reportable. And of course, when we investigate those cases, we would expect to find activity in the neural areas underlying cognitive access and reportability. Given that this neural activity involves cognitive areas, this rules out finding exactly the same kind of activity in the absence of cognitive activity. How, then, could we determine whether or not consciousness can occur in the absence of this kind of activity? Block illustrates this problem with an example involving a further pathological syndrome known as visuo-spatial extinction, which is brought on by brain injuries of certain kinds. If a subject suffering from visuo-spatial extinction sees an object on one or other side of their visual field, they can identify it and report its presence. Yet if there are objects on both sides of the visual field, then the patient functions like a unilateral neglect patient: the patient can only identify and report one object, claiming to be unable to see the other (Davies 2004). The first scenario clearly shows that such a patient has the capacity to be conscious of objects on both sides of the visual field. But in the second case, is the subject conscious of the object on the unseen side yet unable to report it, or does the brain injury stop the subject from being conscious of it at all? Does science contain the answer to this question? Block reports studies by Nancy Kanwisher and colleagues that show that in normal cases, there is activity in the area of the brain known as the “fusiform face area” when subjects report conscious experiences of faces. What is more, experiments using magnetic resonance imaging (fMRI) techniques on the brain of a subject with visuo-spatial extinction have shown that even in situations in which the subject claims not to see a face on the left of the visual field, the area of the brain responsible for recognition of faces on the left (the right half of the fusiform face area) is active, just as it is when the subject successfully sees the face. Yet as Block points out, this result could only show that the subject is conscious of the face but cannot report it if we have already assumed that activity in the fusiform face area alone supports the capacity to have conscious experience of faces and that activity in the areas underlying cognitive access and report are not required. But this is the very question we are trying to answer. We want to know whether a subject can be conscious in the absence of any capacity for cognitive access or report, yet we have no (non-question-begging) way of determining whether a state is or is not conscious other than through the reports of the subject. There are clear parallels between the resolution to this question and the resolution to the question of whether or not subjects “see” the changed/unattended object in the change blindness and inattentional blindness paradigms discussed above (Block 2008: 295–7). Some philosophers – Block (2007: 486) cites Putnam (1981) and Dennett (1988, 1991) – have argued from this that the problem lies in the assumption that there must be a fact of the matter about whether subjects are conscious in these cases. In essence, they contend that the appearance of two distinct but unverifiable explanatory hypotheses shows that we are in fact operating with a problematic conception

162  Perception and the Sciences of the Mind

of consciousness. Block, however, argues that we can resolve the problem by adopting the “method of inference to the best explanation, that is, the approach of looking for the framework that makes the most sense of all the data” (2008: 293). The bulk of his papers on this issue then contain arguments, based on this method, for the view that there can be consciousness without cognitive access/reportability. What we must bear in mind, though, is that the method of inference to the best explanation functions as follows: a claim is supported if it is the best explanation of the evidence. Yet of course, in a way that echoes the issues about disjunctivism and vision science, this is entirely determined by what “evidence” we think needs explaining. As the sets of evidence that need explaining differ, so might our conclusion as to what constitutes the best explanation. In particular, if the relevant evidence set incorporates only the neuroscientific findings, the “best explanation” may indeed be, as Block suggests it is, that there can be conscious states that are unreportable/ inaccessible. If, however, the evidence set is expanded to include such considerations as the epistemological role of perception, the relationship between consciousness and the material world, and so on, then a different “best explanation” might be warranted. Yet surely, we might think, there really is a fact of the matter here (or perhaps that maybe Putnam and Dennett were right after all). This suggests once again that the interaction between philosophy and empirical science will continue to be critical as we progress in our understanding of the relationship between perception as a brainbased phenomenon and perception as a personal-level conscious activity.

Questions •

•

•

Do phenomena such as change blindness and inattentional blindness give us reason to think that the perceptual system does not construct a representation of the environment? Do you think there is a component of perceptual processing that is encapsulated from the influences of cognition? If so, what impact do you think this has on the question of the admissible contents of perception? Given the methodological challenges involved, how might we go about addressing the question of whether there could be phenomenal experience in the absence of cognitive access/reportability?

Further Reading Many relevant historical readings, including extracts from Muller, von Helmholtz, Gibson, and Marr, can be found in Volume III of Fish (2016) Perception: Critical Concepts in Philosophy. To find out more about predictive processing approaches, explore Jacob Hohwy’s 2013 book, The Predictive Mind, and Andy Clark’s BBS target article from the same year, “Whatever next? Predictive brains, situated agents, and the future of cognitive science”. For an exploration of the impact of this debate on the topics of the

Perception and Vision Science  163

first part of this book, explore Zoe Drayson’s 2018 paper, “Direct Perception and the Predictive Mind”. For an explanation of how a Gibson-inspired approach to perception reconceptualizes both the problem of perception and the need for internal representations, see Noë collaborator J. Kevin O’Regan’s 1992 paper “Solving the ‘Real’ Mysteries of Visual Perception: The World as an Outside Memory”. For both science and skepticism about cognitive penetration, Zenon Pylyshyn’s 1999 “Is vision continuous with cognition?” is still relevant, but for more recent approaches, consult Chapter 4 of Athanasios Raftopoulos’s 2009 book Cognition and Perception and Chad Firestone and Brian Scholl’s 2016 BBS target article “Cognition does not affect perception”. For more on cognitive penetration, consult Fiona Macpherson’s papers “Cognitive Penetration of Color Experience” (2012) and “Cognitive Penetration and Predictive Coding” (2015). For discussions of the epistemic impacts of cognitive penetration, explore two papers from 2011 – Susanna Siegel’s “Cognitive Penetrability and Perceptual Justification” and Jack Lyons’s “Circularity, Reliability, and the Cognitive Penetrability of Perception”. Susanna Siegel’s work has also been prominent in discussions of the question of what properties can be perceived; see in particular her 2010 book The Contents of Visual Experience and her 2014 paper “Affordances and the Content of Perception”. Ned Block’s papers, “Consciousness, accessibility and the mesh between psychology and neuroscience” (2007) and “Consciousness and Cognitive Access” (2008) are a good introduction to the issues relating to consciousness and reportability.

Taylor & Francis Taylor & Francis Group

http://taylorandfrancis.com

7

Color, Color Vision, and Color Science

Overview This chapter explores an issue that occurs at the intersection of the philosophy of perception, empirical science, and metaphysics: the philosophy of color. We begin by exploring the empirical discoveries about the physiology of our visual systems that have often been held to refute color realism and the philosophical theories that depend on it, before exploring a range of philosophical responses to this in the form of different theories of color. This chapter concludes by investigating the intriguing phenomenon of synesthesia, in which experiences in one sense modality give rise to unusual experiences in other modalities. Another critical point at which philosophy and empirical science interact concerns the deeply intertwined questions of what colors are and how we perceive them. As we saw in the opening chapter, the mechanistic physics that developed during the Modern period rejected the idea there were any such things as colors, at least as we pre-theoretically understand them. At best, we might identify the colors of physical objects – considered as properties of those objects – with their powers to produce experiences of color in perceivers. As we saw, these commitments then constrain the space of possibilities for these theorists in their philosophies of mind and perception. Given that we have experiences of color, despite there being nothing quite like the colors that we experience in the world, then these colors have to be accommodated within the mind as sensations or properties of sense data. This view of color was then taken up by many of the early psychologists. In the previous chapter, we saw von Helmholtz articulating aspects of this view, and it is also present in the writings of his teacher, Johannes Müller, who similarly claims: That which through the medium of our senses is actually perceived by the sensorium, is indeed merely a property or change of condition of our nerves; but the imagination and reason are ready to interpret the modifications in the state of the nerves produced by external influences as properties of the external bodies themselves. (1837: 1059)

166  Perception and the Sciences of the Mind

As psychology began to create its own results, many theorists took them to reinforce this view, and many present-day psychologists endorse a picture that is essentially the same in its key aspects. Consider, for example, the following quote from Stephen E Palmer, Professor of Psychology and the Director of the Institute for Cognitive Studies at the University of California, Berkeley: People universally believe that objects look colored because they are colored, just as we experience them. … As surprising as it may seem, these beliefs are fundamentally mistaken. Neither objects nor lights are actually “colored” in anything like the way we experience them. Rather, color is a psychological property of our visual experiences when we look at objects and lights, not a physical property of those objects or lights. The colors we see are based on physical properties of objects and lights that cause us to see them as colored, to be sure, but these physical properties are different in important ways from the colors we perceive. (1999: 95) This is an extremely important claim from the perspective of the philosophy of perception. As we saw in earlier chapters, a number of philosophical theories of perception are committed to realism about color – committed to the view that colors are real, mind-independent properties – and this is particularly true when we focus on those theories such as representationalism and naïve realism that are currently popular. If color realism is false, however, then of course any theory of perception that is committed to its truth will be thereby also shown to be false. So what are the empirical findings that have been taken to refute color realism?

Color Science One important discovery for the empirical argument against color realism was the discovery of the existence of metameric pairs (or simply metamers). It is well known that, generally speaking, objects that reflect different profiles of visible light look to be different in color. This suggests that the color realist might identify a particular color with a particular surface reflectance property: might say that to have a certain color just is to have the property of reflecting a certain profile of visible light (alternatively, the property of reflecting a particular color signal). However, it has also been discovered that in some cases, two surfaces that look the same in color (in particular viewing conditions) can nevertheless have markedly different reflective profiles. Such pairs of surfaces are known as metameric pairs or metamers (if they look the same in all viewing conditions, they are known as isomeric pairs or isomers). The following diagram shows the different reflectance profiles of four different objects that are metamers.

Color, Color Vision, and Color Science  167 Metamers 0.8

0.7

0.6

Reflectance (%)

0.5

0.4

0.3

0.2

0.1

0 400

450

500

550

600

650

Wavelength (nm)

Figure 7.1 The existence of metamers therefore has the consequence that the property of being green, say, cannot be identified with the property of reflecting such-and-such a color signal, as there exist green surfaces that lack this property. One possible response for the color realist would of course be to insist that one or other of these apparently green surfaces is not really green, and another to identify color with a disjunctive surface reflectance property (thus claim that the property of being green is the property of having surface reflectance property A or surface reflectance property B, and so on). The most problematic feature of metamers, though, occurs when we ask why surfaces that differ in reflectance profile look the same in color. When we look at the physiology of color vision, we discover that metameric pairs do have a property in common after all. The only problem for the realist is that it looks like that property is the property of affecting us – human perceivers – in the same way. To explain: the human retina comprises around 120 million rods, which only support achromatic (black, white, and gray) vision, and around 7 million cones, which are required to support chromatic or color vision. Both rods and cones are transducers – they convert energy from one form (light) into energy in another (electrical). So when light falls upon a cone (we shall focus on cones as the basis of color vision) it responds by firing – by sending an electrical signal to the neurons it is connected to. However, the electrical signal it sends is completely independent of the wavelength

168  Perception and the Sciences of the Mind

of light that stimulates the cone. No matter what wavelength the incoming light, if it stimulates the cone to fire, the signal it sends doesn’t change. So if all we have to go on are the signals being sent from the cone, we have no way of determining what the wavelength of light falling upon the retina is. To overcome this, evolution has given cones two critical features. First, it provides cones with a sensitivity curve. That is, cones are more likely to fire when stimulated by light of a particular wavelength (call that wavelength its peak sensitivity), and this likelihood of firing falls away as the wavelength of the incoming light departs from this peak sensitivity. The sensitivity of cones can therefore be represented with a bell curve. Second, in humans at least, evolution has equipped us with three different kinds of cone, each of which contains an opsin – a lightsensitive protein – with a different peak sensitivity. Roughly speaking, one type of cone is most sensitive to light of short wavelength (within the spectrum of visible light), another to medium wavelength light, and the other to long wavelength light. Typical sensitivity curves of human cones are shown in the graph below.

Cone sensitivities 1.0 0.9

Relative sensitivity

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

400

450

500

550

600

650

700

Wavelength (nm)

Figure 7.2 So when light of a particular wavelength falls upon the retina, each type of cone will respond in a different way and, from the different firing patterns of the cones, a great deal of information about the wavelength of incoming light can be recovered. Although the cones are often described in terms of the “color” of the light that corresponds to their peak sensitivity – so the short wavelength cones are sometimes called blue cones, the mid-wavelength cones called green cones, and the long wavelength cones called red cones – some scientists suggest this practice can lead to confusion: True, a particular physical wavelength near 580nm appears yellow, 540nm lime green, 600nm orange, and 660nm red. But a yellow indistinguishable from 580nm is seen also with a mixture of only 540nm and 600nm, or only

Color, Color Vision, and Color Science  169

540nm and 660nm. The identical color percept from all these physically identical lights is mediated by identical neural responses evoked by the lights. The determination of color by neural responses to light – not apprehension of color within the light rays themselves – is a foundational and functional property of color vision. (Shevell and Kingdom 2008: 144–5) So when we consider the incident wavelengths of light, we find that different combinations of incident light waves can yield identical color experiences. Echoing the discussions of Adam Pautz from Chapter 5, not only do the similarities in our phenomenal experiences of color correlate poorly with similarities in the wavelengths that affect our retinas, they correlate well with similarities in the character of the underlying neural processing. The same applies to different surfaces: we find that there can be surfaces that reflect color signals that, whilst different, nevertheless cause the same overall patterns of activation in the cones. As these surfaces affect the photoreceptors in the same way, they cannot be distinguished – these are our metamers. This suggests that the existence of metamers is, in some sense, subject relative – the fact that there are metamers, and the particular metameric pairs that there are seem to be a consequence not of the way the mind-independent world is, but rather of the particular way our visual systems work. This has been argued to be evidence that colors are not mind-independent after all. A further, related consideration is that certain features of our conscious experience of color seem to be likewise closely tied to the way our visual systems work. For instance, consider the distinction between unique and binary hues. The unique hues are red, yellow, blue, and green – hues that are not experienced as being “made up” of other hues. Binary hues are, by contrast, experienced as being composite. So, for example, orange is experienced as a composite of yellow and red, purple as a composite of blue and red, and so on. (It is important to be clear that, despite superficial similarities, the division of hues into unique and binary is not the same as the division of colors into primary, secondary, and tertiary.)

The Components of Color Strictly speaking, colors, as we normally conceive of them, have three dimensions: hue, saturation, and brightness. The best way to see what we are talking about when we talk about hue is to see it as what is left when we subtract saturation – the depth, density, or intensity of a color – and brightness from a color. Moreover, there are colors such as brown that do not correspond to hues. All the hues are represented in the color spectrum (the rainbow).

As Ewald Hering (1964) noted, however, the fact that there are only three types of photoreceptor, which (the caveats above being noted) are maximally responsive to red, green, and blue light, but four unique hues – red, green, blue, and yellow – raises

170  Perception and the Sciences of the Mind

some puzzling questions about how these things are related. In particular, where von Helmholtz had thought that the phenomenology of hues corresponded directly to the output of a single cone, the existence of four unique hues suggests that this can’t be the case. Moreover, Hering also noted two other intriguing findings that suggest the four unique hues pair off in an interesting way. First, while there can be a reddish-yellow and a greenish-yellow, there can’t be a bluish-yellow, and while there can be a bluish-green and a yellowish-green, there can’t be a reddish-green. Second, the afterimages created by overexposure to a stimulus of one color will have the color of its complement; so red stimuli elicit green afterimages (and vice versa), whereas blue stimuli yield yellow afterimages (and vice versa). On the basis of these findings, Hering concluded that the four unique hues break into two sets of opposing pairs: a blue-yellow pair and a red-green pair. In the 1950s, Leo Hurvich and Dorothea Jameson then developed this hypothesis into an opponent processing system, which aimed to explain how the visual system could take input from three cones and process them to produce two opponent-pair outputs. Somewhat oversimplified, this model suggests that the output from the photoreceptors in the visual system feeds into a system that performs a number of comparisons of the outputs (firing rates) of the cones (1957). These are represented in the following diagram.

L ( L-M )

+

–

(L+M)

M (L-M)+S ?

S

–

(L+M)-s

Figure 7.3 In this diagram, the boxes L, M, and S represent the long, medium, and short wavelength cones, respectively. For our purposes, the two important computations that the visual system performs are first, that it finds the difference between the output of the medium wavelength cones and the long wavelength cones, and second, that it finds the difference between the output of the short wavelength cones and that of the long and medium wavelength cones combined. Focusing on the first computation first (shown at the top left of the opponent processing diagram), if the output from the medium wavelength cones is greater than that of the long wavelength cones – if L – M < 0 – this signals green; if the output from the long wavelength cones is greater than that of the medium wavelength cones – if L − M > 0 – this signals red; if the output of both types of cone is the same – if L = M – it signals neither. As for the second computation (bottom right), if the combined long and medium cone output is greater than that of the short wavelength cones – if (L + M) − S > 0 – this signals yellow; if the other way round – if (L + M) − S < 0 – this signals blue. And again, if the output of the long

Color, Color Vision, and Color Science  171

and medium wavelength cones combined is roughly equal to the output from the short wavelength cones – if L + M = S – neither yellow nor blue is signaled. The way this opponent processing system responds to incoming light of different wavelengths is represented in the diagram below.

1.00

Relative visual responses

0.75 0.50 0.25 0.00 -0.25 Blue Yellow Red Green white

-0.50 -0.75 -1.00 400

500

600

700

Wavelength (nm)

Figure 7.4 With this graph in hand, we can see that if we were to show the subject a surface that reflects light of around 600 nm (nanometers), the first (red-green) opponent processing system will signal red whilst the second (yellow-blue) system will signal yellow. And indeed, the subject will experience a combination of yellow and red – orange. The binary nature of orange is therefore accounted for, on this picture, by the fact that both elements of the opponent processing system signals a particular color; the color the subject experiences is taken a composite of the colors that are signaled. The same is true of the other binary hues. Now suppose we show the subject a surface that reflects a lot of short wavelength light – on our diagram, light of approximately 475 nm. In this case, as we can see from the graph, the first (red-green) opponent processing system will be in balance and will not signal either red or green. The second (yellow-blue) system, however, will signal blue and hence the subject will experience (unique) blue. Here, the unique nature of the hue experienced is accounted for by the fact that only one of the two elements of the opponent processing system is signaling. Moreover, because the oppositional nature of the system means it is restricted to signaling either blue or yellow and either red or green – with signaling both being an impossibility, given the functional structure of the system – Hering’s observation that we don’t get bluish-yellows or reddish-greens can also be explained. And finally, with the assumption that afterimages are a product of retinal fatigue, this theory would predict that fatigue in the system’s signaling of red would, for example, yield a response in the opposite – green – direction once the stimulus is removed. Thus, the existence of complementary afterimages can also be captured by the opponent processing approach.

172  Perception and the Sciences of the Mind

Once again, the argument is that these particular features of our experience of color seem to be well explained by appeal to facts about visual processing, which can suggest in turn that colors are not out there in the world, but are artifacts of the contingent way that we process visual information. It should be noted however that despite the theoretical promise of the opponent processing explanation of the unique hues, it has proved difficult to find empirical support for it: “specialists in color vision agree that when the [opponent processing] theory is confronted with available physiological data, it does not adequately predict which hues are observed to be perceptually unique” (Philipona and O’Regan 2006: 331; see Broakes 2011: 603–10 for detailed discussion). Another challenge to the picture comes from the surprising finding that, given a suitably controlled experimental environment, subjects could be induced to see color blends such as a reddish-green that, according to Hering and the opponent processing theory, should be impossible. Building on the work of Crane and Piantanida (1983) and Billock et al. (2001), Livitz et al. developed an experimental protocol in which two techniques for inducing “illusory” colors were deployed to induce both a red hue and a green hue in the same area of the display, and found that about two-thirds of the subjects described (unprompted) the color at the point of overlap as “either reddish-green or reddish-yellowish-green” (2011: 9). Yet although these results raise important challenges for the opponent processing model of color experiences, the phenomena that gave rise to it – unique hues, color-complement afterimages, and the incongruousness, if not the sheer impossibility, of certain color combinations – remain to be explained. Most scientists working in this area would think it likely that their eventual explanation would make reference to idiosyncratic features of the human visual system. One final empirical finding that has been claimed to cause problems for color realism derives from what we know about the genetics of color vision. For a long time, we have been familiar with the phenomenon of color blindness, in which genetic disorders result in subjects lacking one of the three photoreceptors (or having one of the three photoreceptors respond to wavelengths that are much closer to one of the other two than is standard). In such cases, this impacts upon the phenomenology of the color experiences – in the most common cases, leading to subjects having difficulties telling red and green apart. Hardin also reports a recent study on mice – which normally only have S and M photoreceptors – where the genes for (human) L photoreceptors were inserted into a mouse genome. This enabled the altered mice to make color discriminations that normal mice were unable to (Hardin 2014: 385). Moreover, Hardin also suggests that minor differences in genetics can lead to differences in phenomenal color, even among subjects that we would class as “normal” perceivers. For example, when asked to adjust a mixture of red and green primaries to match a particular shade of yellow, subjects typically fall into one of two groups, which Hardin suggests “are correlated with genetically based polymorphisms of L (longwave) and M (middlewave) cone photopigments” (2014: 382). Another similar set of studies asks participants to adjust a light to or identify a chip with one of the unique hues. Hardin notes that every one of these studies with “a reasonably large number of observers has found a wide distribution of unique hue loci among normal perceivers” (2014: 383). So not only do different perceivers disagree as to which particular chip (or which particular setting on the dial) corresponds to a particular

Color, Color Vision, and Color Science  173

unique hue, but the range of hues that are identified as unique is quite significant, covering over 50% of the total array of hues. And once again there are reasons to think this might have a genetic explanation (Hardin 2014: 390, fn 13).

Philosophical Theories of Color All of these findings together make quite a substantial case for the claim, endorsed as we have seen by a number of scientists working in the field (and see Byrne and Hilbert 2003: 3–4 for further examples), that colors are not in the world, but are constructed or created by neural activity. If this claim is accepted at face value – if it constitutes, in ways that echo the discussions in Chapter 5, something that scientists have discovered – then it will place significant constraints on the scope of acceptable philosophical theories of perception. In particular, it would seem to rule out any of the theories, such as naïve realism and representationalism, that seem to rely on color realism. Yet, in ways that also echo our discussions of naïve realism, it may be possible to reinterpret the science in ways that accommodate the data, yet allow for colors to be inhabitants of the mind-independent world. As David Hilbert said back in the 1980s: [T]he question of the objectivity of color is in the end a conceptual one. To settle the question, we need to discover which way of conceptualising color allows us to account for both pre-theoretical intuitions regarding color and the wide range of known color phenomena. (Hilbert 1987: 16)

Eliminativism Suppose, then, that we decide to accept the science at face value and deny that there are any such things as colors in the external world. As we have seen, this eliminativist claim is then naturally coupled with the subjectivist claim that when we talk about colors, we are in fact talking about subjective properties of some sort, such as qualia or properties of sense data (e.g. Hardin 1988), and maybe also with the projectivist claim that we mistakenly project these internal, subjective colors onto the external objects out there in the world. As noted, such a view would be compatible with sense datum theory as well as traditional adverbialism and qualia theories. It would also be compatible with phenomenal intentionality, so long as the phenomenal intentionalist is prepared to accept that, should the phenomenal characters of our color experiences ground contents that represent aspects of the world to be colored – as it would seem they would have to, given the argument that any difference in the way things seem to the subject will correspond to a difference in the way the world is presented to be – then all such color-involving contents will be false. Eliminativism about colors faces some important challenges, however. The first is that, given a central argument in favor of eliminativism is the finding that similarities in color phenomenology map far better onto similarities in color processing than they do on to any similarities out there in the world, we might wonder what the epistemic role of color experience could possibly be. After all, this line of reasoning

174  Perception and the Sciences of the Mind

implies that color experiences cannot give us any information about similarities in the external world; if they did, then this mapping between color phenomenology and external similarities would undermine the argument just outlined (Brogaard 2012: 314). But why would evolution go to the trouble of equipping us with a faculty that couldn’t provide us with useful information about the environment? This might make us think that there must be something that color vision is telling us about the world, even if we haven’t yet worked out what that might be. Yet if this is the case, then these properties – whatever they are – would be good candidates for identifying with the colors. In addition, the phenomenon of color constancy – that we do see objects and surfaces to have a relatively constant color through a range of different observation conditions – also suggests that color vision is at least trying to track some persisting property of external objects (Brogaard 2012: 313–4). Brogaard also argues that eliminativism has no space for the distinction between veridical perceptions of color and color illusions. If there are no colors, as eliminativism claims, then all color experiences will be illusory. Yet it seems we should want to make room for there to be a difference in veridicality between seeing, say, a red object in everyday lighting conditions and seeing a white object illuminated in red light. Eliminativism appears unable to accommodate this (2012: 316–7). If the eliminativist claim that there are no colors in the external world feels a bridge too far, there are two other more realist theories of color that are available while still maintaining the spirit of the idea that color is primarily grounded in neural processing. The first is to take the Lockean approach we met in the first chapter and identify colors with the powers or dispositions that external objects have to cause subjective properties of the right sort; the second is to identify them with the physical properties, whatever they may be, that ground the powers that external objects have to cause subjective properties of the right sort. Both of these theories of color would enable us to reinstate the intuition that external objects have colors, and in the latter case, would also allow us to maintain that their having the colors they do is a perceiver-independent fact about them. Moreover, it would also enable us to accommodate what Johnston (1992: 222) calls explanation: the claim that a surface (or volume or radiant source) having a certain color sometimes causally explains our visual experience as of its having that color. If colors just are the properties that ground the powers that external objects have to cause color experiences – or if we allow that dispositions can be causes ( Johnston 1992: 147; McKitrick 2005) – then the colors, thus understood, will sometimes causally explain our visual experiences of those colors. Yoke this to a causal theory of perception – something that at least sense datum theorists and traditional adverbialists are likely to do – then you also get the claim that colors, thus understood, are the properties that we see when we have color experiences.

Color Relationalism In a recent book, Jonathan Cohen has developed and defended a broadly dispositional view of color that claims that objects have many – infinitely many – colors. Cohen calls his view relationalism on the grounds that it understands colors in terms of relations between perceivers and color bearers, and argues for this view by focusing on findings about both inter- and intra-personal variations in color perception.

Color, Color Vision, and Color Science  175

Inter-personal variations include findings such as the one discussed above that there is significant divergence across perceivers as to which color sample, from a range of possible candidates, is unique in its hue. Intra-personal variations include the phenomenon of simultaneous contrast, in which one and the same sample of color can be made to appear very different when placed against different backgrounds. A monochrome example of this is reproduced below: the center gray squares are the same shade, but the square on the left looks darker than the one on the right due to the different shades of the backgrounds.

Figure 7.5 Cohen argues that when faced with the ubiquity of perceptual variation, there is no principled reason to think that there is one, single right answer to questions about whether we see the sample to be the color it really is against the first or second background, or which perceiver is right about the sample that is really unique green. But if there is no unique right answer, then either there are no right answers at all (eliminativism or irrealism) or there are many (ecumenicism or pluralism), and Cohen argues that we should prefer a less revisionary alternative to eliminativism. He therefore endorses the pluralist claim that there are many right answers to questions about which color an object really is – that objects have many colors – and gives a dispositional account of these colors. On Cohen’s view, “a ripe tomato is not merely red, but red to one perceiver in one circumstance, blue to a second perceiver in a second circumstance, … and so on” (2009: 133). With this view, each perceiver would be right about the color they see the sample to be – sample a is unique green to perceiver a in circumstance a, but bluish-green to perceiver b in circumstance a; the center square is light gray to perceiver a in circumstance c (against a dark gray background) and dark gray to perceiver a in circumstance d (against a light gray background). One immediate concern with this kind of pluralism is that it may seem to render inter-personal agreement and disagreement about the colors impossible. Consider the tomato perceivers above: if the first claims that it is red and the second appears to disagree, saying that it is blue, pluralism would say that in fact both are right and that there is no disagreement here after all (Tye 2012: 300). Recognizing that this could also be seen as an unacceptably revisionist claim, Cohen supplements this view with the claim that, whilst we see fine-grained colors as described above, our everyday thought and talk about color represents more coarse-grained, yet still relational, properties. When the perceiver says that the tomato is red (in perceptual context K), this “expresses the property red for the perceivers relevant in context K under the perceptual circumstance relevant in context K” (2009: 100, emphasis in original). In typical cases, this just amounts to saying that “is red” expresses the property red for normal perceivers in normal circumstances (2015: 157). In such circumstances, then,

176  Perception and the Sciences of the Mind

it looks like our tomato perceivers will indeed be disagreeing about which of these coarse-grained properties the tomato actually has. From the perspective of the philosophy of perception, another important objection to dispositional theories, qua theories of what colors are, is that they cannot accommodate what Mark Johnston calls a “core belief” about color: Revelation. Revelation claims that the “intrinsic nature of [a color such as] canary yellow is fully revealed by a standard visual experience as of a canary yellow thing” (1992: 223; Campbell 1993 calls a similar-looking claim transparency). The concern is that, even though this approach enables us to identify properties in the world that cause color experiences and can stand as the referents of color terms, they are not really colors. As Colin McGinn argues: dispositions are not visible properties of things in the way that colors are, so the two cannot be identical … Dispositions lie strictly outside of what is immediately perceptually presented, but colors figure in the very pith of perceptual presentation. (1996: 540) Moreover, if we identify the property of being green with the disposition to look green (to normal perceivers in normal observation conditions), then color properties appear to be relations between perceivers and objects/observation conditions: But surely this misrepresents the phenomenology of color perception: when we see an object as red we see it as having a simple, monadic, local property of the object’s surface. … No relation to perceivers enters into how the color appears. (1996: 541–2) If colors really are dispositions, McGinn suggests, then this will have the consequence that “ordinary color perception is intrinsically and massively misleading” (1996: 537).

Color Relativism Berit Brogaard defends a very similar view to Cohen’s that she calls color relativism. She moves away from Cohen’s view that colors are relations to perceivers as well as to circumstances on the grounds that if such a theory of color were to be yoked to a version of representationalism, then it would have the consequence that the contents of different perceivers’ color experiences would include themselves as constituents, meaning that no two perceivers could possibly have color experiences with the same contents. And with the key representationalist claim that content fixes phenomenology, this would have the consequence that no two perceivers could have phenomenally indistinguishable color experiences (2012: 317–8). The relativism Brogaard prefers therefore takes colors to be perceiver dependent, just not perceiver relative.

Color, Color Vision, and Color Science  177

One response to this on behalf of the dispositionalist is that, even if we might pre-theoretically think that colors are intrinsic, monadic properties, the findings about simultaneous contrast above already suggest that this view must be mistaken. As one and the same color sample can look quite different against different backgrounds (never mind in different lighting conditions, etc.), we have reason to think that the color the sample looks to be does in fact involve relations – to its immediate surrounds, as well as to the circumstances in which it is viewed – even though (as the surprise with which we first discover simultaneous contrast suggests) colors do not intuitively seem to be relational in this way. To this extent, then, the claim of Revelation – that all aspects of the intrinsic nature of a color are revealed in a standard visual experience of that color – appears to make a stronger claim that can be defended. It is worth noting, however, that dispositional views of the nature of color may seem to constrain the potential theories of perception that could work alongside it. If we identify the property of being green with the disposition to look green (to normal perceivers, in normal observation conditions), then it can seem that we would need to understand what it is for something to look green prior to understanding what it is for something to be green (Boghossian and Velleman 1989: 89). Yet this suggests that, even though dispositionalism can lay claim to being a version of color realism, it will not obviously be available to theories of perception such as naïve realism and standard versions of relationalism that attempt to explain what it is for something to look green in terms of being related to something that is, or representing that something is, green (although see Cohen 2009: §6.5 for an attempt to avoid this objection).

Color Adverbialism Another important view of color in this broad space has recently been developed and defended by Mazviita Chirimuuta. She begins developing this view by overviewing recent empirical findings that suggest that color processing is not a discrete process, utilizing dedicated neural resources, but is actually much more integrated with other processing tasks, such as the perception of shape. She also suggests that the function of color vision is not to track intrinsic properties of objects, but to “flag up differences in surface properties” (2015: 74). When viewed in this way, she suggests, there is no reason to find simultaneous contrast perturbing: it is only if we think color vision is trying to identify “real” colors that this seems problematic; if we take color vision to be informing us about differences in surface properties, then “it makes sense that the appearance of an object should change with context, so as to enhance its contrast with surroundings” (2015: 74).

The Role of Assumptions in Shaping Science Chirimuuta also uses the history of experimental studies of color vision to sound a warning about the possibility of inadvertently creating a selfsustaining philosophical-empirical ecosystem: Prior belief in the model will affect experimental design, especially the choice of stimuli, and this in turn will affect the chances of finding

178  Perception and the Sciences of the Mind

evidence for the model. Psychological studies such as Livingstone and Hubel’s garnered evidence for the irrelevance of color for many visual functions … using isoluminant stimuli that elicit responses only from spectrally sensitive neurons. In choosing such stimuli, the researchers are already making the assumption that the color system can be fully characterized in isolation from the rest of vision. … If the alternative assumption is made – that color can only be understood in conjunction with other visual modalities – then different experimental methodologies will be used, ones that have the potential to reveal interactions between color and achromatic processing. (72–3)

To this extent, Chirimuuta appears to agree with Cohen that colors are essentially relational, but she ends up taking this in an importantly different direction. Against the background of the pragmatic claim that the purpose of perception is to guide an animal’s action by allowing it to access useful (i.e. interest-relative/not necessarily uniquely true) information about its environment (see also Thompson 1995), she suggests that the purpose of color perception is not to: acquaint… us with the colors of things; rather, it helps us to see things. Once one accepts the idea that color vision is integral to the perception of objects — their material boundaries, their 3D structure, and their very there-ness among other objects in space — it becomes intuitive to say that color vision gives us access to the presence of those things. (2015: 173) Colors, on this view, “are not properties of things (minds or extra-dermal objects) but of specific kinds of events, namely perceptual interactions” (2015: 140), where perceptual interactions are understood as relations between a subject and its environment. Yet colors themselves are not a relatum of these perceptual interactions, but a property of the event that is the perceptual interaction. This echoes the adverbialist claim that instead of understanding the assertion that a subject senses a red sense datum as committing us to a red object to which the subject bears the sensing relation, but as an event (of sensing) which is modified in a particular way. On the basis of these similarities, Chirimuuta calls her view color adverbialism; but where a more traditional adverbialist would have said that when we see a red apple, we sense-redly or are appeared-to-redly, Chirimuuta would say that when we see the apple, the apple appears-redly to us. So unlike traditional adverbialism, this view does commit us to an object (the apple) to which the subject is related, but it does not commit us to that object having a particular color to which we are related; instead, “when a subject looks at the sky on a cloudless afternoon, the expanse above her appears to her (shows up for her) in a blue way [or alternatively], she is perceiving it in a blue manner” (2015: 142).

Color, Color Vision, and Color Science  179

When considering the objection that an adverbial view of this kind cannot do justice to the idea that color experiences attribute properties to objects, Chirimuuta goes on to present this theory as a version of Fregean representationalism, according to which colors – ways of experiencing – are modes of presentation of properties. To this extent, the view is similar to those developed by Brad Thompson (2009) and Joshua Gert (2017), save for the fact that both Thompson and Gert take colors as they appear in perception to be modes of presentation of colors as they are in the world. So, for example, Thompson claims that “colours themselves are what colour experiences represent … understood as genuine mind-independent physical properties that external objects actually have” (2009: 108). Likewise, Gert follows Cohen in distinguishing between what he calls an object’s “rough” color – the kind of color expressed by everyday color words like “red” and “yellow” – and the precise shade of color that a rough-colored object will appear on any given occasion, and then endorses realism for rough colors and adverbialism about precise colors (2017: 88). Yet unlike Cohen, for whom a fine-grained color is a distinct relational property of the object, for Gert a fine-grained color is a mode of presentation of – a way of perceiving – a real “rough” color property in the world. The difference between Gert and Thompson’s theories and Chirimuuta’s is that while accepting that color experiences are modes of presentation of objective properties, Chirimuuta rejects the idea that the objective properties color experiences present are colors, suggesting instead that color experiences present “a complex of spatial, spectral, and material properties” (2015: 169).

Primitivism While both Gert and Thompson are realists about the colors that inhabit the world, they each endorse a different kind of realism. Gert endorses (realist) primitivism: a non-reductive realism, which claims that colors are primitive, sui generis properties that cannot be identified with any physically specifiable properties. (I have “realist” in brackets above, given the possibility of endorsing primitivism about color properties alongside the claim that no such properties exist, thus yielding an eliminativist primitivism. In what follows, I will focus solely on realist versions of the view, so will omit this clarification in what follows). Primitivism is often connected to Johnston’s thesis of Revelation, which we met above. Recall that Revelation claimed that the intrinsic nature of a color is “fully revealed” in a standard visual experience of that color: one thing that is definitely not revealed in a standard visual experience of, say, canary yellow is the nature of the spatial, spectral, and material properties of the canary yellow object. If we think both that colors are real properties of material objects and that Revelation states an important truth about the colors, then this appears to rule out a reductive version of realism. If a reductive realism were true, and colors were identical to “such things as sets of spectral reflectances [or] disjunctions of microphysical surface properties” (Gert 2017: 9), then Revelation would seem to require this fact to be part of the color’s “intrinsic nature” that is “fully revealed” in a standard visual experience. As this has not proven to be the case, it seems that realism + Revelation = primitivism. As McLaughlin points out, however, this does

180  Perception and the Sciences of the Mind

have the slightly odd consequence that, despite all we have learned and hope yet to learn about color vision and color science, none of this will enable us to learn anything whatsoever about the colors themselves (2003: 98).

Primitivism and the Chromatic Nature of Color When we consider what can be revealed about a color in standard visual experiences of that color, it seems that all that could be revealed would be broadly chromatic facts about the color. So a primitivism based on Revelation would also seem to imply that as far as the intrinsic natures of colors are concerned, colors are “chromatic through and through” (Byrne and Hilbert 2007: 77). Of course, this is not to say that colored objects lack any properties that can be specified in more austerely physical terms, just that these are not part of the intrinsic nature of what it is to be a certain color. For example, suppose that after empirical investigation, we discover that a particular canary’s feathers have certain spatial, spectral, and material properties – call them CYR – that account for how they reflect light (remembering given the possibility of metamers that other canary yellow objects might have different surface properties). The primitivist can then claim that: without there being any commitment to any thesis of property identity … colours [such as canary yellow] are supervenient upon physical properties [such as CYR], if only in the minimal sense that two possible worlds which share all their physical characteristics cannot be differently coloured. (Campbell 1993: 258)

Some recent primitivists have moved away from endorsing Revelation as presented by Johnston. Gert himself prefers a more modest version, which also requires that the subject both has normal vision and possesses color concepts (2017: 23), but Keith Allen, in developing his “naïve realist” view of color, rejects Revelation altogether. Revelation, Allen suggests, is more naturally paired with some form of subjectivism about color/eliminativism about (mind-independent) colors (2016: 134): the idea that everything there is to know about color can be known just on the basis of color experience is very naturally allied with the view that there is a subjective realm, in which colors appear, about which we are infallible. Moreover, he argues that Revelation is too strong. If colors are mind-independent, then this would appear to be an intrinsic or essential fact about them – colors as mind-independent properties of objects would just not be the sort of thing that could have been a property of something mind-dependent. Yet “the justification for the claim that colours are mind-independent properties plausibly goes beyond what is revealed in colour

Color, Color Vision, and Color Science  181

experience” (2016: 137). For example, reflection on the phenomenon of color constancy can give us reasons to think that colors are mind-independent, yet color constancy is not revealed in a standard visual experience, but at best in a succession of experiences across a range of different conditions. To explain the popularity of Revelation in recent discussions, Allen suggests that it is because it is easily confused with a distinct and plausible claim – the broadly naïve realist idea “that visual experience acquaints us with the colours” (2016: 153) – and that this claim, more than Revelation, has a claim to represent “a core common sense belief about perception” (2016: 154). Thus understood, the key feature of primitivism is more its claim that colors are irreducible, mindindependent, sui generis properties, and less that their intrinsic natures can be fully revealed in standard visual experiences. Moreover, despite Allen’s using acquaintance to capture an intuitive aspect of what we think colors are, primitivism is consistent with other theories of perception that involve color realism too. For example, a representationalist could hold that these are the properties that are represented by color experiences, and that it is those experiences representing these very properties that accounts for their phenomenal aspects. Primitivism seems to be naturally understood as claiming that there is one objective color that an object is; yet, as we have seen, there is not only intra-personal variation in the color objects look to have, as shown by the phenomenon of simultaneous contrast and many other more mundane examples, but also inter-personal variation as we saw when we discussed variation in the location of the unique hues. Moreover, there are also reasons to think that there will be significant inter-species differences. For instance, across different animals, “the number of photoreceptor types ranges from one in some nocturnal animals and deep-sea fish, to 16 in the mantis shrimp” (Chirimuuta 2015: 4). When it comes to animals with more than three photoreceptors – such as pigeons and the aforementioned mantis shrimp – then these creatures will be able to make spectral discriminations that humans cannot. This raises some interesting questions for realist theories of color. For example, Byrne and Hilbert (2007) ask us to consider the goldfish, which has a similar visual system to humans, save for the fact that it has a cone that is sensitive to wavelengths near the ultraviolet spectrum. Given this, if a human and a goldfish were to look at two objects that had the same reflectance patterns in the (human) visible light spectrum, but differed in how they reflected ultraviolet light, we might expect these objects to look the same for the human, but different for the goldfish. As we saw, it is to accommodate facts about perceptual variation that Gert distinguishes between an object’s rough color, understood as a sui generis primitive property, and the precise color that it appears to be on any given occasion, which is given an adverbial explanation. Another possible solution, at least for the intra-personal cases, is to identify one type of perceiver and one set of circumstances in which colors look to be as they truly are. Although this kind of approach has often been argued to be arbitrary (e.g. Chalmers 2006: 400–1), Allen contends that there are good reasons to choose natural daylight as the normal viewing conditions (2016: 52–7) and trichromats as the normal perceivers (2016: 58–9), and then to treat any remaining differences among normal perceivers in normal conditions as

182  Perception and the Sciences of the Mind

“disagreements over borderline cases that arise as a result of the inherent vagueness of our colour concepts” (2016: 61). Another option, championed by Mark Eli Kalderon and adopted by Allen for inter-species variations (2016: 66), is to follow Cohen and adopt a version of pluralism. Kalderon endorses a broadly naïve realist approach (although his theory of color is compatible with other theories of perception), according to which the phenomenal characters of experiences are inherited from those aspects of the environment that are picked out or selected by that perceiver’s visual systems. If this is the case, then one way of dealing with different perceivers (both inter-personal and inter-species) having different color experiences when faced with the same object is to claim that differences in their visual systems means that each picks out slightly different aspects of the material environment. If we then accept that: a determinate colour could fall under different determinables thus allowing it to bear different similarity relations to different properties and so participate in different colour property spaces, then colours would have multiple qualitative natures. A single determinate colour would have a qualitative nature perceptually available to a certain kind of perceiver and a different qualitative nature perceptually available to a different kind of perceiver. (2011b: 247–8) Essentially, then, in cases of perceptual variation, the claim is that perceivers with different visual systems will thus be open to different mind-independent aspects of the object’s color. Although this has similarities to Gert’s view, the critical difference is that for Kalderon, the precise colors are actually mind-independent aspects of the object’s color that appear in phenomenal character rather than adverbial ways in which that color is perceived.

Reductive Realism The other central form of color realism – reductive realism – holds that colors can be reduced to mind-independent properties specified in non-chromatic terms – typically physical properties of some sort – and which require the contributions of color science to identify. Which properties? Byrne and Hilbert explore a range of constraints on these properties – they should be properties (i) of surfaces that (ii) are involved in the causal processes underlying color perception as well as being (iii) relatively illumination independent, to account for color constancy, and (iv) something that the visual system could recover on the basis of photoreceptor responses – and conclude that surface spectral reflectances (SSRs) appear to be a good candidate (2003: 9). However, given the existence of metamers, it is likely that for any given color, there would not be a unique SSR that we could expect to be possessed by all surfaces with that color. To accommodate this, Byrne and Hilbert argue that colors should be identified with “reflectance types (or sets of reflectances) rather than with the specific reflectances themselves” (2003: 10). Michael Tye suggests that the relevant types can be

Color, Color Vision, and Color Science  183

identified using the insights of the opponent processing theory. For example, recall that on the opponent processing theory, if the output of the long wavelength cones exceeds that of the medium wavelength cones – if L – M > 0 – then this will signal red, and if the amount of short wavelength light is approximately the same as medium and long combined (so L + M = S), then the yellow-blue system will be in balance and will not signal either yellow or blue. This can be used as a signpost to the SSR type we are looking for: a surface will be (unique) red so long as it has an SSR of a type that in normal viewing conditions will reflect more long wavelength light than medium wavelength light, and as much short wavelength light as medium and long wavelength light combined (Tye 2000: 159–61). There are two significant objections to this kind of reductive color realism. The first is that if SSR types are to be identified with colors, then the relationships between the colors should be replicated at the level of SSR types. Another of the “core beliefs” about the colors elicited by Johnston is unity: “thanks to its nature and the nature of the other determinate shades, canary yellow … has its own unique place in the network of similarity, difference and exclusion relations exhibited by the whole family of shades” (1992: 222). If colors just are SSR types, then each SSR type should have its own unique place in this kind of phenomenal structure network. Yet if we only have the SSR types to go on, it is not possible to reconstruct this network of relationships (Thompson 1995: 120–1; see also Hardin 2003: 198; Brogaard 2012: 308; Chirimuuta 2015: 127). Byrne and Hilbert argue that with the help of the opponent processing approach for identifying SSR types, this objection can be met: for example, we can ground the claim that purple is more similar to blue than to green on the fact that the purple and blue SSR types are similar in that both reflect a high proportion of short wavelength light (2003: 15). Others, however, have rejected the demand. For example, Mohan Matthen asks: why can one not hold that experience of red and experience of green track properties that exist independently of us, but that the experienced opposition between them has no real counterpart? Why can we not say that color experience misrepresents the true resemblance relations of colors? (1999: 68) This discussion leads into an interesting question about the compatibility of reductive realism and naïve realism. Reductive realism pairs nicely with representational theories, as the colors thus understood can be the properties that are represented in color experience, which, in turn, fixes the experience’s phenomenology. Yet for the naïve realist, the phenomenology of color experience is fixed by acquaintance with the colors themselves. As we saw, this pairs naïve realism with primitivism relatively well, but it is not entirely clear how this can be finessed as acquaintance with sets of SSRs. To see the problem, consider McLaughlin’s claim that colors, understood reductively, cause perceivers to have phenomenal experiences of color, and that as the phenomenal characters of these experiences stand in similarity and difference relationships, then we can claim that the colors participate in these relationships derivatively (2003: 115). This response to the objection suggests that colors themselves,

184  Perception and the Sciences of the Mind

understood as SSR types or a similar physical property, are identified with the colors in virtue of being the properties that ground the powers that external objects have to cause color experiences. If this is the case, then these properties do not themselves appear to be phenomenal, which means that the phenomenal characters of color experiences could not be explained by the subject being acquainted with the colors thus understood. For reductive realism to pair with naïve realism, then it would seem to need to be the case that SSRs, SSR types, or any other candidate physical properties be not only visible properties, but inherently phenomenal properties. It is unclear whether this would be a plausible claim, but it certainly doesn’t appear to be any part of the claim of most reductive realists (Chirimuuta 2015: 46). As we might have come to expect, with explaining color constancy being an important touchstone for reductive realist theories of color, a second challenge concerns what to say about cases of color variation. As we saw when discussing primitivism, there are different dimensions to this problem: intra-subjective variation (simultaneous contrast), inter-subjective variation (unique hue location), and inter-species variation (alternate visual systems). The two primary responses to cases of variation have been to either to treat instances of variation as involving an illusion (Tye 2000; Batty 2010), in which one of the perceptions (although we may not know which one) is mistaken, or to endorse pluralism (Matthen 2005; Kalderon 2007) and thereby allow that both are correct. (Another possible response, inspired by Brewer (2006: 169), would be to deny that there is any variation at the phenomenal level, instead locating the disagreements in the subject’s conceptual or cognitive responses to the experience.) This is not a once-and-only choice, however – the reductive realist may offer different responses to different cases. For example, Tye (2000: 156) suggests that cases of simultaneous contrast can be treated as illusions on the grounds that the object’s surrounds block the subject from correctly perceiving its color. However, if this approach was adopted for cases of inter-species variation, then it would have the consequence that some species – maybe even humans, given that other species have superior visual systems to ours – never perceive colors veridically. For this reason, pluralism might seem preferable here, with the claim that there is a range of SSR types – a range of colors – that human beings cannot pick up on, given the limitations of their visual systems, but that other creatures can.

Synesthesia and Color Synesthesia A further interesting phenomenon for the philosopher of perception is color synesthesia, in which regular perceptual experiences can trigger anomalous – and often disconnected – color experiences; for example, where listening to music or seeing particular (monochrome) letters and numerals can lead some people to see colors, even when nothing in the environment is colored. Color synesthesia is a special case of a more general phenomenon, which was originally characterized as occurring “when stimulation of one sensory modality automatically triggers a perception in a second modality, in the absence of any direct stimulation to this second modality” (Harrison and Baron-Cohen 1997: 3). If we call the initial unexceptional experience the inducer and the experience that this yields the concurrent (Grossenbacher and Lovelace 2001: 36), then the most widespread form of synesthesia occurs when the

Color, Color Vision, and Color Science  185

inducer is the experience of hearing certain words or sounds and the concurrent is a visual experience of color (Harrison and Baron-Cohen 1997: 3). As the concurrent involves the experience of color, we can call this a case of color synesthesia. What makes cases like this interesting is that the concurrent – in this case, the color experience – is produced in an extremely nonstandard way.

Synesthesia in Popular Culture Sometimes, popular characterizations of synesthesia can be misleading. For instance, Richard Cytowic’s influential 1993 book on the phenomenon was called The Man Who Tasted Shapes. Such a description suggests that in synesthesia, objects or properties can be experienced via a nonstandard sensory modality (it suggests that, in this subject, shapes can be experienced by the modality of taste). Sadly, however, this is not an accurate description – it is not that shapes are tasted, but rather that experiencing certain tastes causes anomalous experiences of shape.

As the hearing-color synesthetic experience just described involves distinct sense modalities, we can call it a case of cross-modal synesthesia. Other cases of cross-modal synesthesia include Cytowic’s man who tasted shapes – who had taste (inducer) to shape (concurrent) synesthesia – as well as a range of other cases, such as word-taste, sound-taste, touch-color, sound-geometric vision, and more (Spiller et al. 2015: 73). In addition, however, intra-modal forms of color synesthesia have also been discovered. Most prominent is what is known is grapheme-color synesthesia, in which a visually experienced black grapheme – a numeral or letter written in black ink – yields an additional color experience. This finding suggests that the original definition of synesthesia above, which defined synesthesia as having inducers and concurrents in different modalities, needs revision. Another shortcoming of this original definition is that it limits the range of possible inducers to stimulation of a sensory modality. Yet recent studies have suggested that synesthetic effects can also occur at a cognitive level. For example, Ramachandran and Hubbard (2001) describe a range of experiments in which grapheme-color synesthetes will have different concurrents to the same stimulus depending upon how the stimulus is classified. For example, they describe a stimulus in which a large numeral 5 is made up of small 3s, and report that when synesthetic subjects switch back and forth between seeing it as a 5 or as an array of 3s, their concurrents switch along with the changing classifications. Similarly, they showed these subjects the stimulus “THE CAT”, where the H in THE and the A in CAT were actually the same shape – with the two vertical lines converging, but not touching, at the top – and discovered that the same physical form elicited different concurrents depending on whether the context was saying the shape was an H or an A. Figure 7.6

186  Perception and the Sciences of the Mind

These findings led Ramachandran and Hubbard to suggest that synesthetes may fall into higher and lower groups, where lower synesthetes respond purely to what in the previous chapter we called “low-level” visual properties of the stimulus, such as shape, but higher synesthetes can respond to cognitive properties of the stimulus, such as meaning. Another important distinction among synesthetes is the distinction between projectors and associators. Dixon et al. (2004) found that 11% of the grapheme-color synesthetes in their sample “described their colors as being ‘out there on the page,’ as though a transparency bearing a colored number was placed on top of the written digit” (2004: 335–6). As these colors are “projected” out into the external world, they described these synesthetes as projectors. The majority of synesthetes, however – who reported their concurrents as being “in my mind’s eye” or “in my head” – they called associators. They described the experiences of associators by drawing an analogy to the case of a non-synesthete viewing an achromatic picture of an object that is highly associated with a particular color – something like a banana or a stop sign – noting that when viewing the black-and-white picture, we both “know” the color of the object and can easily form an image of it in our mind’s eye, but do not in any way project it out onto the picture. In a recent study, Ward et al. draw further distinctions within these categories: between surface projectors, who experience the colors “in the same location as the attended grapheme”, and near-space projectors, who experience them as if “on a floating screen” (2007: 927); and between associators who see “a coloured copy of the grapheme in their mind’s eye” (61%), who see “blocks of colour that do not take the shape of the grapheme” (6%), and who merely express a strong sense of knowing the color without claiming to actually see it (32%) (2007: 928 fn.3). Although synesthetic experiences have been reported for well over a hundred years, it is only relatively recently that scientists and philosophers have started taking the condition seriously. Ramachandran and Hubbard lament that, “despite a century of research, the phenomenon is still sometimes dismissed as bogus”, with skeptics suggesting that synesthetes simply have a hyperactive imagination, or are attention-seeking, or are even reporting the effects of psychoactive drugs. Other skeptical explanations of synesthetic reports include the suggestion that they are either reports of memories that were associated with the things experienced, such as associating different colors with letters having been exposed to the well-known multicolored letter refrigerator magnets as children, or perhaps even due to an overactive sense of metaphor – an extension of something we all do when we say that the wind is biting, the cheese is sharp, or the shirt is loud (2001: 4). However, empirical studies have uncovered a number of reasons to think that this is a real, perceptual phenomenon. Pop-out experiments on grapheme-color synesthetes conducted by Ramachandran and Hubbard seem to show that there is a significant perceptual difference between synesthetes and non-synesthetes. The phenomenon of “pop-out” is straightforward. If an image contains a triangle of red “2”s in an array of randomly placed green “5”s, the triangle formed by the red “2”s will “pop out”; it will be easy for you to see (as indicated by the monochrome version of the figure on the right, below). However, if you are a normal subject, then if an image contains a triangle

Color, Color Vision, and Color Science  187

formed by black “2”s in an array of randomly placed black “5”s, the triangle formed by the “2”s will not pop out (left). 5 5 5 5 5 5 55 5 5 5 5 5 5 5 5 5 2 5 5 5 5 5 5 2 5 25 5 5 5 5 5 5 5 2 5 2 2 5 5 5 5 5 5 5

5

5 5 5 5 5 5 55 5 5 5 5 5 5 5 5 5 2 5 5 5 5 5 5 2 5 25 5 5 5 5 5 5 5 2 5 2 2 5 5 5 5 5 5 5

5

Figure 7.7 Yet where synesthetes who experience one color when they see a “5” and another color when they see a “2” are concerned, pop out actually occurs for the monochrome stimulus. These results suggest that such synesthetes really do “see” the colors when they see the numbers rather than merely associate the colors with the numbers (although recent studies, such as Mattingley et al. (2006), have suggested that things could be a little less straightforward than this suggests, inasmuch as this phenomenon may require attention to the inducers). Another experiment that appears to support the reality of synesthetic experience is a variation of the Stroop task. In the original Stroop task, subjects were required to name the color of the ink in which certain words were printed. What was found was that if the printed word was the name of a color other than the color of the ink – say if the letters BLUE were printed in red ink – then subjects would take longer to name the color of the ink than they did if the ink color and color name matched, or if they were merely asked to name the color of a patch. What this suggests is that processing of the color word interferes with the subject’s ability to recognize and name the ink color. Mills et al. (1999) designed a Stroop-type task for a grapheme-color synesthete, in which she was asked to name the ink color that graphemes were printed in. Mills and colleagues varied the colors of the inks, so that they either corresponded to or diverged from the subject’s stable synesthetic concurrent for that digit. Upon testing, they found that, as in the original Stroop test, where there was a mismatch between the ink color and the concurrent color, the subject took longer to name the ink color, suggesting that the synesthetic color experience was interfering with the processes underlying color recognition. Intriguingly, other studies suggest that this effect may not be limited to synesthetes. For example, Elias et al. (2003) found that the robust but non-synesthetic pairings of an experienced cross-stitcher, who associated numbers with colors of thread, also elicited Stroop-like interference. Meier and Rothen (2009) also found that interference could be created so long as participants were given a week’s training on particular letter-color pairings. These studies can be interpreted in more than one way, however. Perhaps we might think they cast doubt on the idea that synesthesia is a truly perceptual phenomenon, but perhaps they point toward an unexpected result: that training can induce genuine synesthesia-like experiences in non-synesthetic subjects.

188  Perception and the Sciences of the Mind

In addition, Blake et al. (2005) explored whether synesthetic colors could be used to induce the McCollough effect. The McCollough effect is an afterimage-based effect in which subjects adapt to both (say) a vertical red grating and a horizontal green grating by alternating between them during the adaptation period. What McCollough discovered was that subsequently presented achromatic gratings would induce different afterimage effects depending on whether the orientation of the achromatic grating was horizontal or vertical. Blake and colleagues developed synesthetic colored gratings by creating lines out of the achromatic graphemes that induced the appropriate colored concurrents and found that these were able to produce afterimage effects in their synesthetic subjects (2005: 53–4). Neuroscience has also found evidence that points to synesthesia being a real phenomenon. For example, Nunn et al. (2002) found that sound-color synesthetes show increased activities in neural areas associated with color perception when they hear triggering inducers; yet this increased activity is not found in non-synesthetes, even if they have been trained to form sound-color associations and even if they are asked to imagine the relevant color when they hear the paired sound. As for the suggestion that these could just be down to some unusual associations born out of memories – the multicolored letter refrigerator magnet hypothesis – studies suggest that synesthetic trigger/concurrent pairings are far more robust than would be expected if subjects were just remembering a set of color-letter associations. For example, Baron-Cohen et al. (1987) instructed both a word-color synesthete and a non-synesthetic control to record the colors that they associated with 103 different words. After an interval of two weeks, they retested the control (without warning) and found that she was only able to recall 17% of the color associations she had earlier provided; when they retested the synesthete, however – also with no warning, but this time after a period of over two months – they discovered that every single pairing matched the pairing recorded ten weeks earlier. A more recent, larger study (Eagleman et al. 2007), featuring 15 self-reported synesthetes and 15 controls, showed similar results, finding that each of the 15 synesthetes had more consistent color pairings than any of the controls. These findings strongly suggest that synesthetic word-color pairings are relying on more than just memory. For all of these reasons, we appear to have a case in which empirical research has demonstrated the existence of a robust phenomenon that is at the very least connected to the operation of our perceptual systems. It is therefore something that philosophers of perception ought to consider carefully. So how does synesthesia impact upon the philosophy of perception? On the face of it, synesthesia can seem to involve an unusual cause of a perceptual experience – consider, in this light, Harrison and Baron-Cohen’s definition above, which explicitly states that the unusual experience that is triggered in synesthesia is “a perception” – and hence that a theory of perception ought to cover synesthetic concurrents too. This is supported by Auvray and Deroy’s conclusion that the “documented reports and various observations converge to stress that a synaesthetic experience can be as vivid and forceful as other experiences induced by other kinds of stimuli, and that it can also spontaneously lead to beliefs” (2015: 644).

Color, Color Vision, and Color Science  189

Some philosophers have argued that if this is the case, then it looks to favor theories of perception that allow the phenomenal properties of an experience to vary independently of the experience’s connection to the world, as opposed to those that try to explain the phenomenal properties relationally or representationally. For qualia theorists, sense datum theorists, and adverbialists, the synesthetic concurrents would appear to be able to be accommodated simply as atypical subjective responses to the world. For those theorists who attempt to ground phenomenal properties in worldly properties – either by virtue of being acquainted with or representing those properties – things aren’t quite so straightforward. The problem for naïve realists is obvious – if synesthetic experiences are perceptual, then when the surface projector sees the monochrome letter to be red or the near-space projector sees a floating red patch, in neither case does it seem the phenomenal color in the experience can be explained in terms of acquaintance with anything red. For this reason, the naïve realist is far more likely to try and resist the suggestion that synesthetic experiences are truly perceptual, perhaps suggesting instead that they are misleading visual experiences – perhaps hallucinations or illusions. The same approach would seem to be the most natural response for representationalists too. If the synesthetic experience represents the monochrome grapheme to be red when it is not or there to be a red patch floating in space when there is not, then it could be argued that this is not inconsistent with representationalism per se, merely inconsistent with a representationalist treatment of such experiences as veridical. Some theorists have challenged the suggestion that these experiences can legitimately be treated as non-veridical. Wager (1999: 269), Rosenberg (2004: 101), and Auvray and Deroy (2015: 261) all emphasize that the unusual phenomenal responses synesthetes have to certain environment features are reliable – that is, that inducer – concurrent pairings tend to be stable over time, and can be used by their subjects as reliable indicators of the external states of affairs (Rosenberg 2004: 101). Given a fairly standard psychosemantics for determining what something represents then, this might make us think that those experiences do represent the inducer. Although this point can be made for both intra- and intermodal synesthesia, it is easier to see in the cross-modal case. For instance, Wager describes subject Cynthia (an imaginary case based on real descriptions) who is a music-color synesthete. When Cynthia hears Middle C, she experiences a red bar, six inches wide by one inch high (1999: 269). Although we might be tempted to say that this experience represents that there is a six inch by 1 inch red bar in her environment, Wager suggests that as this experience reliably covaries with Middle C, we should allow that it represents Middle C. (If you find it hard to shift the feeling that the experience nonetheless represents the presence of a red bar in the environment, Wager suggests you also imagine that Cynthia is retinally blind, so has no normal color experience). If this is correct, then both her auditory experience and her synesthetic experience represent the same external state of affairs, despite differing in phenomenal character, which suggests, in turn, that representational content does not in fact fix phenomenal character. If this is correct, then is there any way that the representationalist could hold on to the core idea that representational content fixes phenomenal character whilst

190  Perception and the Sciences of the Mind

maintaining that synesthetic experiences are cases of unusual veridical perception? One possible solution would be to contend that even though both experiences represent Middle C, there is nonetheless a representational difference between them, say by noting that the auditory experience represents the sound as having volume and timbre. Another option suggested by Torin Alter is to endorse the distinction between the state of affairs that is represented as obtaining (Middle C) and the specific phenomenal manner in which this information is presented (2006: 4), although it is worth noting that this does give up on the pure representationalist attempt to explain phenomenal character in terms of representational content alone. A third option would be to accept that this objection hits the mark, but only to the combination of representationalism and a particular theory of content. Perhaps, for example, a teleosemantic theory of content might enable the representationalist to claim that Cynthia’s red bar experience does not represent the pitch of musical notes, as it could not be said to have evolved to give information about musical pitch. The claim that synesthesia is not an evolved trait could potentially be defended by appeal to the fact that synesthesia of this kind is both rare in the population and different across individuals. Other theorists have challenged the claim that concurrent experiences are indeed perceptual. For example, Fiona Macpherson suggests the best (perceptual) explanation of the unusual experiences of projector grapheme-color synesthetes would be that they perceive two colors to be in the same place at the same time, but wonders whether it makes sense to say there could really be such an experience (2007: 76). Similarly, Casey O’Callaghan points out that there “is no reliable connection between the colors of things and the colors a synesthete experiences as a result of hearing sounds. Things usually just lack the determinate qualities synesthetic experiences attribute” (2017: 51). If these theorists are correct, then there would be scope to attempt to accommodate synesthetic experiences within a theory’s account of illusion or hallucination. This would look relatively straightforward from the perspective of representationalist theories – for instance, the experiences could involve false contents that represent either that objects have certain colors (for surface projectors) or that there are patches of color at certain points in the environment (for near-space projectors) – but it is less clear how a naïve realist might approach this. A particularly interesting approach from this perspective is Ophelia Deroy’s view of synesthesia as involving phenomenal enrichment (Deroy 2014, Auvray and Deroy 2015). According to this view, the concurrent experience is not an additional perceptual experience, but an imagistic enhancement of the perceptual experience of the inducer. Most of the time, researchers and philosophers seem to be misled by descriptions that make the synaesthetic ‘extra’ analogous to classical experiences—as if for instance, they were additional experiences of colour patches superimposed onto graphemes or floating after-images. But, on examining the reports more closely, the evidence suggests that the synaesthetic colour is often just an addition of some colour dimensions (such as hue) to a grapheme. In other words, the synaesthetic ‘extra’ is actually relatively poor—and seems to contain fewer

Color, Color Vision, and Color Science  191

dimensions than a typical experience of a coloured patch—which has hue, brightness, saturation but also extension, texture, orientation, etc. (Auvray and Deroy 2015: 646) If this is correct, and the additional phenomenology attaching to the concurrent is significantly less than the phenomenology of a typical perception of color, then this suggests that the synesthetic component of the experience is not fundamentally perceptual. Instead, suggests Deroy, it is a fairly typical perceptual state that has been enhanced with mental imagery. To explain this, Deroy introduces us to the idea of a reverse Perky (2014: 402–4). In the original Perky experiments, subjects were asked to imagine a particular object – classically, a banana – while a faint image of just such an object was projected onto the wall in front of them. Although subjects took themselves to only be imagining a banana, and didn’t take themselves to be perceiving at all, their perception of the projection did nonetheless appear to somehow enter the content of their imagination, as was shown by such things as the imagined banana reliably having the same orientation as the faint projection. As Deroy interprets this, these cases show that a subject’s imaginative state can be enhanced by a perceptual state of which the subject is unaware being unknowingly incorporated into it. A reverse Perky, then, would be a situation in which an imaginative state (of which the subject is unaware) is similarly incorporated into, and enhances, the subject’s perceptual state: “The incorporated state is felt as imagery in the Perky cases, and as perceptual for synesthesia. The mixed origin of the content gets un-noticed, and subjects experience a single (incorporated) content in a unified perceptual way” (2014: 403). Such a view of synesthesia would seem to align well with the view of hallucination as degenerate sensory imagination that we met in Chapter 5. One further interesting feature of synesthesia is that it seems to show that there can be interesting interactions between experiences that we would typically associate with distinct sense modalities. Indeed, Casey O’Callaghan suggests that crossmodal synesthesia is in fact “just one of a variety of effects in which one sense modality causally impacts and reshapes experience associated with another” (2017: 45; see also Cohen 2017). This suggests, in turn, that separating our multifaceted sensory engagement with the environment into distinct sensory streams may not be an accurate way of conceiving of the role of sense modalities (Keeley 2013). We shall discuss these issues further in the remaining two chapters.

Questions •

• •

Do you think that Johnston’s thesis of Primitivism is a core pretheoretic belief about color that must be respected by any plausible philosophical theory of color? Do the findings from color physiology refute color realism? What can we learn about color experience from considering synesthesia?

192  Perception and the Sciences of the Mind

Further Reading For an excellent overview of the history of the interplay between philosophy and color science, which goes back to Aristotle, see Kathleen Akins and Martin Hahn’s 2015 paper “Colour Perception”. Mark Johnston’s 1992 paper “How to Speak of the Colors” is getting on, but remains a good introduction to the philosophical issues regarding color as well as introducing the influential thesis of Revelation. Mazviita Chirimuuta’s 2015 book Outside Color: Perceptual Science and the Puzzle of Color in Philosophy is both an excellent, up-to-date overview of the relevant science and a defense of a view she calls color adverbialism. Broadly dispositional theories of color can also be found in Jonathon Cohen’s 2009 book The Red and the Real and Berit Brogaard’s 2012 paper “Colour Eliminativism or Colour Relativism?” Keith Allen (2016) defends a version of primitivism in the course of developing, as the book’s name suggests, A Naïve Realist Theory of Colour. Mark Eli Kalderon’s 2011 paper “Color Illusion” contains a discussion of color illusions and a defense of color pluralism as a response. Alex Byrne and David Hilbert defend the reductive realist identification of colors with types of surface spectral reflectances in their 2003 “Color Realism and Color Science”. A good philosophical introduction to the phenomenon of synesthesia can be found in Fiona Macpherson’s 2007 paper “Synaesthesia, Functionalism and Phenomenology”. Ophelia Deroy’s account of synesthesia as involving imagistic enhancement of phenomenal character can be found in her 2014 “Synesthesia: An Experience of the Third Kind?” and her 2015 paper with Malika Auvray, “How Do Synaesthetes Experience the World?”

8

Perception and the Nonvisual Sense Modalities

Overview In virtue of what do the senses differ from one another? And how do we distinguish between seeing something, hearing something, feeling something, smelling something, and tasting something? This chapter begins by considering a range of different answers to this question and exploring the problems they each face. We then go on to investigate empirically informed philosophical theorizing about the nonvisual modalities of hearing, touch, and the chemical senses, and consider whether the theories of visual perception explored in Part I can be extended to cover these nonvisual perceptual modalities.

Before we can begin to think carefully about concepts such as inter- or crossmodality, we need to spend some time thinking about the base concept of a sense modality. Traditionally, and potentially pre-theoretically, human beings possess five distinct sense modalities: vision, hearing, touch, smell, and taste (this dates back to Aristotle’s De Anima (1984: 424b)). However, recent scientific studies of the physiology of sensory systems have suggested that we have more senses than this – possibly 17 (Keeley 2002: 10) and maybe even 21 (Durie 2005). To begin our discussions, then, we shall explore the question of how many senses we have by focusing on the more fundamental issue of what distinguishes one sense modality from another.

Individuating the Senses When we ask how the different senses are distinguished from one another, there are two closely related questions we might be asking: • •

By what criteria are the senses distinguished from one another? How do we, as perceivers, distinguish between seeing something as opposed to hearing it, feeling it, and so on?

194  Perception and the Sciences of the Mind

The first of these questions is more metaphysical: it asks, what is it about the nature of the different senses that makes them distinct from one another? The second is more epistemological: it asks, when we have a sensory experience, how do we know which sense we are using? Suppose, for example, that as you are reading this, you smell coffee and hear the familiar hissing, spluttering sound of the coffee machine. How do you know that you smell the coffee and hear the coffee maker rather than the other way round? In an influential early paper, Grice (1962) suggests four possible answers to these questions – that each distinct sense modality corresponds to a particular sense organ (and associated downstream processing); that each distinct sense modality is sensitive to different external stimuli; that each distinct sense modality yields experiences with a specific kind of phenomenal character; and that each distinct sense modality is characterized by the distinctive features that it enables us to become aware of.

The Characteristic Sense Organ View Probably the most intuitively natural response would be to suggest that we individuate sense modalities by appeal to their respective sense organs. This would be to answer the first question by claiming that seeing is what we do with the eyes, hearing with the ears, smelling with the nose, tasting with the tongue, and feeling with the skin. It answers the second by claiming that we know which sense we are using because we know which sense organ we are using. Of course, to try and individuate the senses by means of their respective organs means that we need to begin by individuating the organs themselves. This may seem obvious: the nose is the protuberance in the middle of the face, the eyes are the orbs located in the eye sockets, and so on. However, things are not quite so straightforward. Consider, for example, that one can feel not only with the skin, but also with the eyes, ears, nose, and tongue. If seeing is what is done with the eyes, then feeling something sharp poking one’s eye would thereby be counted as a case of seeing; if hearing is what is done with the ears, then feeling the cold of ice water by having it dropped in the ear canal would count as hearing the coldness; and so on. This suggests that this basic attempt to individuate the senses by way of their respective organs will not work. One way of refining this view is to note that our everyday concept of a sense organ is not a purely physical notion; in fact, the sense organs are defined partly functionally: by what we do with them. As O’Dea puts it, the eyes “are characterized in perceptual experience not as biological entities but rather, as it were, as tools” (2011: 306). He suggests that when we visually perceive something, we are aware of more than just the thing that we see – we are also aware that we are using our eyes to do so: there is both a distinctive proprioceptive feeling and a distinctive set of exploratory behaviors that corresponds to the use of each particular sense organ. In a claim that echoes the enactive view of Noë and O’Regan, O’Dea suggests that to see is not merely to have an experience of a tract of one’s immediate environment, it is also: to know what to do to see what the world is like to one’s right or left. And this involves being aware that I am doing something in order to see exactly what I am seeing and that [this] is made possible by what I am doing with my eyes. (2011: 307)

Nonvisual Sense Modalities  195

This enables the view to claim not only that we individuate sense modalities by appeal to their respective sense organs, but to give some content to the suggestion that we know whether we are seeing or hearing (etc.) because we know whether we are using our eyes or ears.

The Characteristic Stimulus View Another way of developing the sense organ could be to say that when we talk about “the organs of sight”, we are not meaning to refer to the eyes qua biological object, but to the eyes qua organs that respond to a particular stimulus. Eyes are thus understood as the organ that responds to light, ears as the organ that responds to sound waves, and so. Feeling something as sharp by pressing it into your eyeball would therefore not qualify as a case of seeing, as in this case the eyes are responding to pressure. Feeling something as cold by dripping it into your ear canal would therefore not count as hearing coldness, because in such a case the ears would not be functioning by responding to sound waves. This takes us to the second of Grice’s possible answers: that what distinguishes one sense from another is a matter of the nature of the external stimuli to which it is responsive. This proposal runs into difficulties when we consider the case of touch. Touch has many different kinds of physical stimulus it responds to, including “heat, cold, pressure, irritation, itch and pain” as well as distinctively “low force, slowly moving mechanical stimuli” such as a light stroke (McGlone and Reilly 2010). Of course, this may be – and has been – taken as evidence that touch is not a single sense after all, but in addition to this, this proposal struggles to provide an adequate answer to the second, epistemological question of how a normal perceiver might come to know that they are seeing or hearing. Suppose we have a perceiver who presses something sharp into their eye and denies that they see the sharpness, or drips ice water into their ear and denies that they hear the coldness. This response would seem to require this perceiver to know, in each of these cases, that they were not responding to light or pressure waves in the atmosphere. This response therefore seems to require our perceiver to have specialized knowledge of a kind that, normally, we would not expect to be possessed by someone who could tell which sense they were using. Another interesting problem case for the view that seeing is the sense that responds to the energy in light can be found in what are known as sensory substitution systems (O’Dea 2011: 299–300). In the most famous case of a sensory substitution system – the tactile-visual substitution system (TVSS) – Paul Bach-yRita and colleagues (1969) installed a grid containing hundreds of small points on a blind subject’s back, and the feed from a camera worn on the subject’s head was then converted to an array of tactile stimulation using this grid. Using such a system, blind subjects could learn to detect and orient themselves to distal objects in their environment. In addition, Noë and Hurley report the “remarkable fact that subjects fitted with TVSS report perceptual experiences that are, to varying degrees, like vision” (2003: 195), which suggests that not only does TVSS provide information about distal objects, but does so in way that at least feels visual. This therefore appears to be a case in which subjects are sensing their environment and

196  Perception and the Sciences of the Mind

are doing so in a way that is responsive to the energy in light. The view that vision is the sense that responds to the energy in light would therefore appear to classify this case as one in which a blind person is seeing, or at least that they have partial sight, but O’Dea rejects this claim as “absurd” (2011: 300). Another response would be to hold that, despite what it makes the subject aware of and how, as the most proximal stimulation for this new sense is pressure, not light, it remains a case of touch. In an influential 2002 paper, Brian Keeley develops a view that reflects how this question is conceived of within the biological sciences, which combines aspects of both of these approaches. He argues that “to possess a genuine sensory modality is to possess an appropriately wired up sense organ that is historically dedicated to facilitating behaviour with respect to an identifiable class of energy” (2002: 6, my emphasis). As indicated by the italicized sections of the quote, this definition contains four key elements: each sense (i) involves a distinctive sense organ and downstream neural processing, (ii) is responsive to a distinctive physical stimuli, (iii) enables the organism to engage in behavioral discrimination of stimuli, and (iv) that this information gathering system has an evolutionary role. This final condition is in place to explain why, for example, we don’t consider the eye to be a detector of mechanical stimuli, even though pressing on your eye can produce visual experiences – because it evolved to be receptive to light, not pressure. It also enables Keeley to avoid the claim that TVSS subjects have a visual sense. Such subjects: should no more be said to have a visual modality than all of us should be said to have an electrical modality just because we can detect electricity with our tongues. It is true that they are getting visual information about the world, but they are getting that information via the tactile modality. (2002: 19) Interestingly, however, Keeley doesn’t address the concerns about touch, suggesting that this approach counts touch as a single sense on the basis that it involves a sense organ dedicated to a particular class of physical stimulation (2002: 26). For the reasons given above, it may not be this straightforward.

The Characteristic Experience View A third possible answer to the question of what distinguishes the senses from one another is the suggestion that each sense corresponds with a characteristic kind of experience. Grice outlines the suggestion as follows: It might be suggested that two senses, for example, seeing and smelling, are to be distinguished by the special introspectible character of the experiences of seeing and smelling; that is, disregarding the differences between the characteristics we learn about by sight and smell, we are entitled to say that seeing is itself different in character from smelling. (1962: 135)

Nonvisual Sense Modalities  197

What is important about this suggestion is that to distinguish it from the alternative suggestion that sight and smell give us information about different properties (the characteristic properties view, which we shall discuss shortly), Grice outlines it as claiming that seeing and smelling have a different character, even when we disregard the differences between the properties we learn about. Given that within each modality there are significant differences among the possible phenomenal properties we can enjoy – think about the difference between seeing a nearby pink sphere and a distant blue mountain, or between the sound of a quiet whisper and a loudly blown tuba – this proposal requires there to be some meta-character shared by all phenomenology in one modality that is not possessed by phenomenology from any of the other modalities. All of the other modalities will each have their own distinctive meta-character. (A sympathetic proposal, albeit cast in representational terminology, can be found in Chalmers, when he suggests that representational states might have, in addition to properties of representing that things are thus-and-so, the properties of representing this content in a certain manner, such as representing it visually or auditorily (2004: 342).) Now, this proposal can seem reasonable – after all, we can all tell the difference between visual phenomenal character and auditory phenomenal character, even if we find it difficult to articulate precisely what that difference is – but there are cases where it is less clear. For example, consider touch: do the experiences of feeling the warm sun on your face and having someone stand on your foot share a meta-character? The idea that there is some distinctive visual phenomenal character, common to all visual experiences, also appears to clash with the intuition (discussed in Chapter  3) that visual experience is transparent. To claim that an experience is transparent, recall, is to say that when we introspect, we “see through” the experience to the objects themselves; that we only find features of the objects of experience, not features of the experience itself. If this is so, then there is no additional “smelling character” and “seeing character” to the experiences – at least not one that we can discover through introspection. As Grice argues, “the attempt to describe the differences between seeing and feeling seems to dissolve into a description of what we see and what we feel” (1962: 144). In addition to this, a view that makes an ineliminable appeal to special modality-specific phenomenal character runs into difficulty when it comes to accounting for the possibility of distinct sensory modalities in other animals. For instance, Nagel famously wondered what it would be like to be a bat, given that bats have a sonar sense that humans lack (1974/1979). Other animals also appear to have senses that we don’t. For example, it is widely believed that sharks have the capacity to sense magnetic fields, that snakes have the capacity to sense infrared radiation, and that some fish can sense electric potential (Keeley 2002). If the characteristic experience view is correct, then this requires there to be a distinct phenomenal character associated with each distinct sense. But given the problems Nagel raises, claims that particular animals have senses that we lack would therefore have the status of “propositions to which one must subscribe without really understanding them” (1974/1979: 447). If an appeal to special meta-characters is problematic, we may be able to avoid this by appeal to the notion that experiences can be indiscriminable from one another.

198  Perception and the Sciences of the Mind

To see how this might work, consider the relations being exactly the same height as and looking as tall as. Focusing on the first relation first, note that we can legitimately infer that if Alysha is exactly the same height as Bismah and Bismah is exactly the same height as Cai, then Alysha is exactly the same height as Cai. This inference is legitimate because the relation being exactly the same height as is a transitive relation. The relation looking as tall as, however, is intransitive. As this relation is intransitive, the following inference will be illegitimate: if Alysha looks as tall as Bismah and Bismah looks as tall as Cai, then Alysha looks as tall as Cai. The reason this inference is illegitimate is because it leaves open the possibility that Alysha and Bismah might be unnoticeably different in height, as might Bismah and Cai, but that when you stand Alysha and Cai next to one another, you can just make out that they are not as tall as one another. This provides a way of developing a characteristic experience view that does not appeal to special modality-specific phenomenal characters (Clark 1993). The claim is this: Any two experiences belong to the same modality if and only if those experiences can stand at either end of a series of experiences that are connected at each stage by the indiscriminability relation. To see how this works, imagine three experiences: the first is an experience of an expanse of red; the second an experience of an expanse of green; and the third is an experience of C-sharp. In the case of the first two experiences, we could construct a series of experiences, such that the first experience is one of an expanse of red; the second is of a red with a hint of orange so slight as to be indiscriminable from the first; the third of a yet more orange-red indiscriminable from the second; and so on through yellow to green. So our first two experiences could stand at either end of a series of experiences, where the members of each pair of experiences in the series are indiscriminable from one another. They therefore qualify as experiences of the same modality. Yet we cannot construct such a sequence from either of the first two experiences to the experience of C-sharp. They therefore qualify as experiences from different modalities. In this way, it might be possible to separate out all experiences into modality-specific classes. There are more generic concerns with the characteristic experience view, however. Grice, for example, worries that it leaves it a contingent matter that seeing is the kind of experience we have in response to colors and shapes; that smell is the kind of experience we have in response to odors and scents; and so on. This concern can be highlighted by asking the following question: Would it be possible to have the sight-characteristic experience – where this is either an experience with the introspectible “seeing character” or an experience from a particular indiscriminability-based class – in response to scents and odors, whilst having the smellcharacteristic experience in response to colors and shapes? Our natural inclination is to answer this question in the negative, but it is not clear how this response justifies us in doing so. Moreover, cross-modal cases of synesthesia may also provide real-life examples of cases in which, say, the kind of experience typically associated

Nonvisual Sense Modalities  199

with seeing – a phenomenal color character experience – occurs in response to auditory stimuli. Given that such cases actually occur, it may give us reason to think that Grice’s worry could be well-founded. Another concern is that the characteristic experience view rules out the possibility of either fundamentally nonconscious sensory modalities or the possibility of nonconscious experiences within a modality that also supports conscious experiences. As to the first of these possibilities, not only does it seem to be a coherent possibility that we might have a fundamentally nonconscious modality, there is evidence to think that we might actually have such a modality (although of course being nonconscious, this may come as news to us!). Keeley (2002) suggests that there is evidence that humans possess a vomeronasal sense modality, physically located in pits on either side of the nasal septum, which enable us to sense pheromones – chemical compounds that are involved in social sexual interactions. In other animals, damage to the vomeronasal system leads to reduced sexual behavior, whilst artificial stimulation of the system leads to increases in such behavior. According to Keeley, there is a “growing list” of similar findings for humans, which is some evidence that humans have such a sense modality too (2002: 24). Yet if there is indeed a vomeronasal sense modality, it is one which has no characteristic experiences associated with it. So the characteristic experience view would be unable to distinguish this modality from others. When it comes to the second possibility – that there can be nonconscious experiences within a typically conscious modality – there has been a recent rise of defenses of the claim that there can be unconscious perception of this kind (e.g. Prinz 2010). In Chapter 3, we briefly discussed the phenomenon of blindsight, in which subjects who claim to lack phenomenal awareness of some area of their visual field nevertheless appear to be receiving and processing information from that area. Beatrice de Gelder and colleagues have even suggested that blindsight subjects can recognize facial expressions (de Gelder et al. 2005) and successfully avoid barriers and blockages while walking (de Gelder et al. 2008). This latter finding may also remind us of the two visual systems hypothesis we met in Chapter 6, and its suggestion that there is a nonconscious perceptual processing stream that can enable real-time visual guidance of action. In each of these cases, it looks as though information about the subject’s environment is coming in through the visual sense modality in a way that can be utilized by the subject, yet lacks the distinctively visual phenomenal character required by this approach. It may be that where philosophical theories of visual perception are concerned, the challenges of nonconscious perception could potentially be avoided simply by claiming (in the kind of way we saw in Chapter 5) that the conscious states that the philosopher of perception theorizes about are not to be identified with states of the perceptual system; so the presence of nonconscious states in perceptual processing does not entail that these states are nonconscious perceptual experiences (Phillips 2016b). When it comes to distinguishing the sense modalities, however, this cannot be dodged in the same way: if this is indeed visual information, as it appears to be, then we cannot distinguish it from information that arrives through other senses solely on the basis of its possessing a distinctively visual phenomenal character.

200  Perception and the Sciences of the Mind

The Characteristic Properties View The characteristic properties view aims to individuate the senses by means of the different external properties they give us access to. To adequately spell out this view, we need to begin with the sense-neutral claim that we perceive – have sensory access to – an array of properties. This claim is “sense-neutral” because it doesn’t specify which sense we perceive these properties by, merely that they can be perceived. Then, on this basis, we distinguish the senses from one another in terms of which of this array of properties each sense gives us access to. For example, seeing is the sense that gives us access to colors, shapes, and sizes; hearing is the sense that gives us access to volume, pitch, and timbre; and so on. In the terminology of Chapter 6, this claim could be interpreted as holding that each sense corresponds with a particular set of admissible contents, which suggests, in turn, that this approach would be a natural fit with representational theories of perception. Properties that are only accessible to a single sense are often called, following Aristotle, that sense’s proper sensibles. Yet not all perceptible properties are tied to a single sense in this way. There is also a range of common sensibles: properties that can be perceived by more than one sense. For instance, we can both see and feel that things are warm, or smooth, or square. Likewise we can both see and hear that someone is angry or that the sea is rough. Or we can both smell and taste that something is sweet, or both see and smell the burnt toast, and so on. Common sensibles thus constitute a prima facie problem for the characteristic properties view inasmuch as we cannot associate each sense with mutually exclusive lists of characteristic properties. In order to accommodate this, Grice turns to the idea that some properties are perceived directly, whilst others are perceived indirectly or in virtue of the direct perception of distinct properties. The idea is that we can then deal with these problem cases by stipulating that the property in question is only perceived directly by (at most) one of the two senses. On this approach, we directly feel that something is warm but only see that it is warm by seeing other properties, such as it being red, perhaps. Yet it is not clear that this response will work in all cases. Both smelling and tasting something as sweet, for example, or both seeing and feeling something to be square seem to be cases in which both properties are perceived equally directly. An alternative approach is to appeal to the fact that when we see an object to be square, we perceive far more properties than squareness alone; we see the object’s squareness as a member of a set of properties that includes its size, color, direction of motion, and so on. When we feel squareness, however, we perceive squareness as a member of a different overall set of properties, including heft, texture, temperature, and so on. On this view, we distinguish between the senses by distinguishing between the range of properties that sense enables you to perceive. As Grice points out, this proposal runs into a problem: Suppose a man to be resting a half-crown on the palm of one hand and a penny on the palm of another: he might (perhaps truthfully) say, ‘The half-crown looks to me larger than the penny, though they feel the same size’. (1962: 137–8)

Nonvisual Sense Modalities  201

According to the characteristic property view, our subject perceives an array of properties – where “perceives”, recall, is sense-neutral – which includes properties from both the visual list (colors, shapes, etc.) and the tactual list (texture, temperature, etc.). Our subject also perceives one coin to be larger than the other and the two coins to be the same size. The problem is that nothing in the characteristic property view’s presentation of the facts enables us to conclude that the perception of the coins as differing in size is through the sense of sight and that the perception of them as having the same size is through the sense of touch. In light of this problem, Roxbee-Cox (1970) presents a Key Feature version of the characteristic property view, which is similar in spirit to Aristotle’s claim that the senses can be individuated by their “special objects” (1984: 418a). Where the straightforward characteristic property view collects together the array of properties that can be seen or heard, and then claims that seeing or hearing just is the perception of those properties, the key feature view picks out specific properties from that collection and claims that seeing or hearing just is the perception of those key properties. Roxbee-Cox suggests that the key feature for sight is the property of having some color property, the key feature for hearing is having some loudness, pitch, and timbre, the key feature for taste is having some taste, the key feature for smell is having some odor, and the key feature for feeling is having some feel to the touch. The question of how a perceiver knows which sense he or she is using can then be answered by appeal to the perceiver’s knowing that he or she is hearing something, say, by knowing that the thing perceived has some loudness or timbre, or would know that he or she is smelling something by knowing that the thing perceived has some odor, and so on. In order to allow that other properties can be seen, heard, tasted, smelled, and felt, Roxbee-Cox presents the following suggestion. For non-key property, p, to be perceived by sense, S, the subject also needs to (directly) perceive that the object bearing p has S’s key feature. For instance, circularity is a non-key property. For it to be perceived by sight, this requires us to also directly perceive that the (circular) object has some color property. For it to be perceived by touch, this requires us to also directly perceive that the (circular) object has some feel to the touch.

The Components of Sound When it comes to the characteristic properties of sound, two of the three – pitch and loudness – are relatively intuitive. The pitch of a sound is how high or low it is: the difference between the different notes on a piano is a difference in pitch. Loudness is, well, loudness: the same note on a piano can be played softly and forcefully – here, the sounds have the same pitch but differ in loudness. What about timbre? Well timbre accounts for the leftovers. Whatever differs between two sounds that isn’t a difference in pitch or loudness is a difference in timbre. So the difference between a piano, a violin, and a clarinet playing the same note? That’s timbre. The difference between two voices singing the same note? Timbre again. On a physical level, differences in timbre will show up as differences in the shape of the waveform, but phenomenologically, it’s simply everything that is not pitch or loudness.

202  Perception and the Sciences of the Mind

Roxbee-Cox suggests that the key feature view can deal with Grice’s coin case as follows. Recall that the characteristic property view claimed that the subject both perceived the half-crown to be larger than the penny and perceived the coins to be the same size, but didn’t have the resources to account for the fact that the subject sees the coins to differ in size whilst feeling them to be the same size. The key feature view, however, can say the following: [T]he half-crown seems larger than the penny, and … the seeming-perception of this relationship involves the direct perception that the coins have some colour property; while … the half-crown and the penny seem equal in size, and … the seeming-perception of this relationship involves the direct perception that the coins have some feel to the touch. (1970: 539) There remain concerns with this approach, however. For one, the key feature for taste is the property of having some taste. Yet of course for this to work, we have to know what tastes (qua properties of objects) are independently of them being defined as whatever properties can be sensed by the modality of taste. Given the statements of the key features listed above, it would seem that the same objection might also be made to smell and, given some of the concerns about color realism raised in the previous chapter, even to the case of vision. In addition to this, we might also wonder whether the account of the perception of non-key properties is correct: is it really the case that perception of any non-key property at any time requires the subject to perceive the key property? For example, we saw in Chapter 6 that subjects can suffer from achromatopsia – the inability to see colors – yet still have and know that they have visual experiences. Even where normal subjects are concerned, the change blindness experiments we discussed might be argued to show that on some occasions, it is possible to see an object’s shape, say, without seeing that object to have some color property. Relatedly, Treisman and Gelade’s experiments on feature binding (Treisman 1996), where they suggest that in visual processing, properties such as shape and color are processed separately and only bound together under attention, also raises the possibility that “shape might be sensed independently of colour” (Matthen 2015: 576). The final objection to consider turns on the case of touch. What is the key feature that is associated with the sense of touch? Think of all the different kinds of feeling there are. There is an active kind of contact feeling, in which we probe the object to feel its shape or the texture of its surface; there is a kind of passive feeling, when we feel an object press or push against us; there is feeling an object to be warm or cold, feeling the air to be warm or cold, feeling water in which we are immersed to be warm or cold, the feeling of cold when alcohol evaporates on the skin; there is feeling one’s heart beating inside one’s chest or an oyster slipping down one’s throat; there is feeling the sun on your back and the wind on your face; there is feeling an itch or a tickle; there is feeling chemicals burning one’s skin; and so on. When all of the different kinds of feeling are listed in this way, we can

Nonvisual Sense Modalities  203

see how difficult it will be to identify one unique key feature common to all cases of touch, especially when we recall that on the key feature view, the perception of non-key properties requires the perception of the key feature. In a recent discussion, de Vignemont and Massin make a case for the claim that pressure is the characteristic property of touch, and that many of the other things that we can feel are felt in virtue of our feeling pressure (2015: 297–300). For example, feeling the texture of a surface is a matter of feeling the ways the pressure in your fingertips changes as you move your fingers over it; feeling that an object is hard involves feeling that, as the pressure between it and your body increases, the object does not deform. However, de Vignemont and Massin do acknowledge that this approach cannot accommodate every type of perception that we mentioned above – in particular, it implies that “the perception of temperatures … needs to be excluded from touch proper” (2015: 299). Yet this suggestion faces a challenge in explaining the fact that pressure and heat experiences can blend together phenomenologically. For example, Matthew Ratcliffe describes the experience of grasping a warm cup, and suggests that the “the touch of the object and the perception of its warmth seem to knit together seamlessly; what is perceived is not an object plus heat, but a hot object” (2012: 418).

Other Approaches Although each of these four approaches has appealing features, each picking out something that seems important for the question of what differentiates the senses, none of them has been generally accepted. What are the prospects for developing a more widely accepted method for individuating the senses? Well, one possibility might be to reject the claim (implicit in the above) that our individuation process ought to answer both the metaphysical and epistemological questions together. Those questions, recall, asked what distinguished sensory modalities from one another and how we know that we are experiencing with one rather than another. To see why this might be plausible, note that the very fact that there can be disagreement over whether or not humans have a vomeronasal sense relies on its being accepted that it is possible that there could be a sense organ that we are not conscious of using. If we have to know that we have a sense, then there can be no dispute. So if we do accept that there could be sensory modalities we don’t know we have, this suggests that the questions of what makes something a sense modality of a particular kind is importantly distinct from the question of how we might know which sense modality we are using on any given occasion. This would allow us to develop an account of what individuates the senses from one another on purely scientific grounds, say, without worrying about how a subject might know which sense they are using (Keeley 2002). Other theorists have attempted to combine elements of the theories presented above, with the hope of thereby avoiding the problematic objections. For example, Fiona Macpherson argues that all four key criteria are important when it comes to thinking about the senses: representation (i.e. the features of the world that the sense

204  Perception and the Sciences of the Mind

gives us information about), proximal stimulus (the nature of the energy that is transduced in the sense’s operations), phenomenal character, and sense organ. Macpherson suggests that these four criteria can be used to define a multidimensional space in which any possible sense would be located: from the traditional five Aristotelian senses, to nearby cases such as the vision of dichromatic and tetrachromatic animals, to more distant cases of animal perception that involve sensitivity of a kind not possessed by humans (such as bat echolocation, shark electroreception, and pit viper infrared detection) and beyond. She then suggests that when all actual senses are plotted in this multidimensional space, “we might find that the actual senses are to be found in clusters in this space. For we will find, for example, that human vision and bee ‘vision’ are closer together in this space than human vision and bat echolocation” (2011a: 38). This would enable us to accept that there is no unique defining feature of the folk concept of any one sense, whilst maintaining that nonetheless the folk were on to something, in that the traditional senses identify significant clusters. An alternative approach focuses on what we do with our perceptual systems as opposed to what they make us sensitive to. For example, Mohan Matthen begins from the idea that perception typically involves an array of different activities in which creatures use their senses to explore their environment. With this in mind, consider some environmental property, F, and note that different types of perceptual activity might provide the creature with information about this property: for instance, if the property is shape, then the activities of attentively looking at an object and exploring it with our hands can both give us information about this property. If two types of perceptual activity give us access to a property in this way, Matthen calls them “mutually certifying” for that property. This notion is then employed to group perceptual activity types into sets: “A set of activity-types S is a perceptual system if every member of S certifies every other for all properties detected by each” (2015: 580). As he notes, the strength of this condition – every activity type has to certify all of the properties certified by all of the activities in the set – means there couldn’t be overlap between sets: if there was one activity (call it A) that appeared to belong to two different sets, then as this would mean that A shares all its mutually certifying properties will all members of both sets, all members of both sets would share mutually certifying properties with each other, thus showing that in fact this was a single set. As there can be no overlap, Matthen then suggests that this can be used to define a perceptual modality, as “a faculty that gathers environmental information by means of a perceptual system. Two perceptual modalities are different if their perceptual systems are different” (2015: 581). By focusing on what we do with our perceptual systems in this way, Matthen argues that we end up with a taxonomy of perceptual modalities that is closer to the original Aristotelian five.

Hearing Given that the majority of this book has been given over to discussing and evaluating theories of visual perception, we might think that we would be well placed to simply extrapolate theories of the nonvisual senses from these extant theories

Nonvisual Sense Modalities  205

of vision. However, it turns out that things are not quite that simple. Let’s begin by asking about the direct objects of hearing: What do we hear, when we hear? Everyday language might make us think we can offer the same theory for hearing as we did for vision: we hear things, such as lions, refrigerators, and guitars. Yet as natural as this is, people will also spontaneously talk about hearing the sounds that objects make: the roar of the lion, the hum of the fridge, or the jangling sound of the guitar. Some philosophers have argued that this shows the claim that we hear things like lions and guitars is, strictly speaking, mistaken: on this view, our auditory perception of objects is indirect, mediated by our immediate perception of the sounds they produce (Batty 2010). This claim might seem to be supported by the fact that when we discussed the characteristic property approach to individuating the senses above, we identified pitch, loudness, and timbre as the characteristic properties of hearing, but unless we are speaking metaphorically, these are not properties of things, but of the sounds they make. Yet unlike the indirectness we examined when discussing the sense datum theory, here the direct objects of perception – the sounds – also appear to be public entities of some kind: when driving in the car and hearing a strange rattle, you and I can both attend to, discuss, and theorize about the sound that we both hear. So there appears to be an extra level in the case of hearing that is not present in the case of vision. Where in visual perception we perceive objects by perceiving their visible properties, in auditory perception we hear objects by hearing the sounds they make and we hear sounds by hearing their acoustic properties. This suggests that the closest parallel to vision appears to be at the level of sounds rather than at the level of objects that produce sounds. This raises two important and interrelated questions: What are sounds, and how do we hear objects by hearing sounds? Taking the first question first, we might think that sounds could be given a purely subjective reading, along the lines of the subjectivism about color we met in the previous chapter. Yet such a view would conflict with the intuition that sounds are public objects (although of course an account that paralleled a dispositionalist account of color could bridge this gap). If we accept that sounds are public objects, then what are they? Knowledge of the science of sound might make us think that as objects produce sounds when they vibrate, then the sounds themselves must be identical to what is produced by a vibrating object – that is, compression waves in the surrounding air. This would also fit with the science of auditory perception as it is compression waves that stimulate our auditory sense organs. However, the suggestion that we perceive compression waves in air is no more plausible than the suggestion that we perceive light: yes, we perceive by being sensitive to compression waves/light, but that is not what we perceive in virtue of this sensitivity (Ratcliffe 2012: 416). Moreover, as Pasnau argues, this theory threatens to make “our perception of sound illusory. We do not hear sounds as being in the air; we hear them as being at the place where they are generated” (1999: 311). Pasnau himself claims that “sounds, like colours, are … properties of the objects we perceive” (1999: 318). Which properties? Well to acknowledge the importance of the scientific findings, he suggests that sounds should be identified

206  Perception and the Sciences of the Mind

with an object’s property of vibrating in a certain way. If such a view were correct, then it would provide a straightforward answer to the question of how we hear objects by hearing sounds – in the same way that we see an object by seeing its color (Leddington 2014). However, O’Callaghan (2007: 17) argues that there are reasons to be dissatisfied of the claim that sounds are properties of objects. In particular, he argues that there are a range of features of sounds – that sounds have duration, they stand in similarity and difference relations to one another based on their pitch, loudness, and timbre, and can persist through changes in these qualities – that are difficult to square with a view of sounds as properties of objects. Moreover, he suggests, the analogy to color just gets the relationship between objects and their sounds wrong: objects are not qualified by their sounds, they produce sounds, and the sounds they produce are not repeatable in the way properties are (2011: 378). In addition to these considerations, we might also note that sounds are essentially temporally extended entities. Kalderon (2019) draws a useful analogy between color and sound. We can see that colors are essentially spatially extended, because whilst we can imagine ever smaller things being colored, we cannot imagine anything being colored while lacking extension altogether. Similarly, whilst we can imagine briefer and briefer bursts of a particular pitch, the idea of something having a pitch without duration is inconceivable. Given these considerations, a natural proposal is that sounds must be a kind of temporally extended particular that instantiates acoustic properties. On the basis of phenomenological observations, O’Callaghan goes on to suggest that the best theory of the nature of sounds is that they are event-like particulars that are spatially located at or near their sources (2007: 30). He ultimately opts for a theory of sound, according to which sounds are identified with the event of a vibrating object disturbing its surrounding medium (2007: 59). This view is very similar to that of Casati and Dokic (2005), who argue that a sound is identical to the event of an object’s vibrating. Although these theories are obviously near neighbors, there are important differences between them. One concerns whether there can be sounds in a vacuum. For Casati and Dokic, as objects can vibrate in a vacuum, sounds can exist in a vacuum; for O’Callaghan, as there is no surrounding medium in a vacuum, then even where there is an event of an object’s vibrating, there will not be an event of the vibrating object’s disturbing its surrounding medium, and hence no sound in a vacuum. Another significant difference is the way the two theories answer the question of how hearing sounds enables us to hear objects. Suppose I tread on a floorboard, which squeaks. According to O’Callaghan (2011: 376), the squeaky sound – the event of vibrations in the floorboard disturbing the surrounding air – is a proper part of the event of my treading on the floorboard. This approach enables O’Callaghan to maintain that we hear audible events by hearing the sounds that are parts of those events. Casati and colleagues (2013), however, hold that the sound event is in fact identical to the source event: the event of the floorboard’s vibrating just is the event of my treading on it.

Nonvisual Sense Modalities  207

Temporal Perception If we perceive sounds (or indeed any other event), and events are essentially temporally extended entities, then our perceptual experiences themselves would appear to have a temporal structure. This is often overlooked in philosophical theories of perception, probably in part because of the emphasis on vision, and the ease with which we can assume that a visual experience of a scene has similarities to a photograph or snapshot of a scene. Some theories of temporal experience do attempt to build temporal phenomenology out of snapshots, in the way a movie combines still images to form a moving picture; others retain the commitment to momentary perceptual experiences, but allow that they have extended periods of time as (possibly intentional) objects; yet others allow that perceptual experiences themselves are temporally extended (see Kon and Miller (2015) for an overview of a range of different positions that have been taken with regard to temporal consciousness). Matthew Nudds (2015) argues that both events-based theories of sound face a problem in individuating multiple sounds when these are produced by a single object. He gives the example of simultaneously scratching and tapping a reverberant rough surface, which leads to the production of two sounds – a scratch and a tap (where these name vibration events that are identified with the sounds, not the two events (involving fingers) that produced the sounds). Yet there is only one object that is vibrating; so how can there be two vibration events to identify with the two distinct sounds? After considering and rejecting some attempts to resolve this problem, Nudds argues that “we can’t individuate sounds events in terms of objects’ vibrations … as such; but we can individuate sound events in terms of what caused or excited the different modes of vibration of an object” (2015: 283). The scratching sound can be identified with the vibrations caused by the (finger-involving) scratching event, and likewise for the tapping sound. What is more, he suggests, this approach conforms to what we know about how the auditory system works. A key task for the auditory system is to take the compression wave that reaches the ears, which has a particular set of characteristics on account of all the sound events occurring in the subject’s environment, and try and reconstruct the individual sounds that led to this particular proximal stimulation. In a discussion that echoes our discussions about vision in Chapter 6, Nudds suggests the auditory system does this by making use of implicit assumptions about how the different frequency components in the proximal stimulus would have been produced and transmitted in order to allocate each element of the stimulus to the particular event that (probably) produced it. According to this representational theory of sound, the “auditory system does not represent sounds and then determine what produced those sounds; it determines what sound-producing events there are and that process results in an experience of sounds corresponding to those events” (2015: 288).

208  Perception and the Sciences of the Mind

These theories of sound suggest that hearing can give us information both about objects and about the sounds they make. But we might also ask whether hearing, like vision, can give us information about the layout of space – that is to say, is hearing a spatial sense? Famously, P.F. Strawson argued that the phenomenology of auditory experience could be fully characterized just by appeal to sounds: that if experience were “exclusively auditory in character, there would be [no] place for spatial concepts at all” (1959: 65). This can seem a surprising claim: whilst it is true that we cannot locate sounds in space in the same kind of way we can locate objects in space using vision, it still seems that we can hear sounds coming from particular directions, and possibly even at different distances. These appear to be spatial concepts, even if they don’t involve a spatial field or frame of reference in the way vision does. Strawson allows that we can assign directions to sounds based solely on auditory experience (1959: 66), yet nonetheless insists that hearing is not intrinsically spatial. Nudds argues that this claim makes sense if we see the claim that hearing is not intrinsically spatial as claiming that audition does not present space itself. In vision, we are aware of areas of space in which objects could be located, but are not; in hearing, however, there is no auditory equivalent – no auditorily presented space where a noise could be but is not: “there’s no difference between not experiencing a sound at some place, and experiencing no sound there” (2001: 213). Casey O’Callaghan, however, insists that it is an empirical question whether audition alone could suffice for a conception of space and argues that even though it would be somewhat impoverished, empirical studies demonstrate that at a minimum, “audition has a prominent directional component that depends upon, for instance, differences in arrival time, phase, and levels of sounds waves at the two ears” (2010: 135). Once we have a fix on what the objects of hearing are, we can then ask about the nature of our awareness of those objects. This is the call for a philosophical theory of hearing. Perhaps, one might think, we can develop a theory of hearing by simply extrapolating our preferred theory of vision. Yet, of course, things are not that simple as the different features of hearing may make the different approaches more or less plausible. For instance, a sense datum theory of hearing might hold that when we hear a sound, we are directly acquainted with/sense a sense datum that possesses the right combination of auditory properties of loudness, pitch, and timbre. The plausibility of this approach will depend, in part, on how applicable you find the Phenomenal Principle to be for the case of hearing: for example, suppose you hallucinate a baby crying, how forceful is the intuition that despite this being a hallucination, there nevertheless must be some thing that you hear which bears the relevant auditory properties? Similarly, when we consider the case for a broadly adverbialist/qualia theoretic/ phenomenal intentionalist treatment of audition, as the theory itself would seem to be able to be applied to the case of hearing – so long as we allow that there are either hearing-based ways in which sensing can be modified or a range of audible qualia to go alongside the visual qualia – we will need to ask whether the arguments that seemed to support it for the case of vision carry over. Take the inverted spectrum, for example: the intuition that a green apple could look to you just like a

Nonvisual Sense Modalities  209

red apple looks to me (and vice versa) can be used to argue for the claim that both of our experiences of the apple’s surface have a different qualitative property, and that this cannot be reduced to or identified with any property of the apple itself. In the case of hearing, is there a comparable intuition that could be used to motivate this approach in the same way? As we saw for the case of color, representationalists and naïve realists will likely need to endorse a realist view of what the objects of hearing are, but with this caveat in place, essentially the same considerations seem to apply. For representationalists, for example, it would seem reasonably straightforward to hold that auditory experiences represent sounds as having loudness, timbre, pitch, and possibly location properties, for example. Things are slightly less clear-cut in the case of naïve realism, however. The naïve realist would claim that auditory experiences are episodes of being acquainted with sounds and their properties. As we saw in the case of color, however, theories of sound that attempt to reduce sound to some physical property can challenge the intuition that the sounds really possess the phenomenal qualities they need to for the naïve realist explanations to work. Suppose, for illustration, that Casati and Dokic are right that the sound of a violin just is the event of the violin string’s vibrating. The naïve realist not only needs to say that we are acquainted with these events in auditory perception, but that the phenomenal character of auditory perception is inherited from the phenomenal character of the sound itself. In the case of color, we saw reasons to be skeptical that a reductive physicalist theory of colors could supply this; if this is also the case for audition (as Casati and Dokic (2005) appear to suggest when describing their theory as “non-phenomenalist”), then maybe the naïve realist needs a sound-based sibling of color primitivism. Perhaps the theory developed by Scruton (2009) – on which sounds are events that are defined solely in terms of the audible phenomenal properties they instantiate and which, whilst being caused by events in the physical world, cannot be reduced to any such events – might be a better fit.

Touch As we saw in our discussions of how to distinguish the senses, it is not entirely clear how many senses fall under the broad umbrella of what we call “touch”, and answering this question isn’t made any easier by exploring the underlying physiology. The skin contains a number of distinct types of mechanoreceptors – transducers that convert a mechanical stimulus into electrical signals – that respond to different types of mechanical stimulus. Gallace and Spence (2014) list six mechanoreceptors, which respond to stimuli such as vibrations of different frequencies, pressure, skin deformation, temperature change, and tissue damage (20). In addition to the mechanoreceptors, there are also thermoreceptors, nociceptors (which themselves can be classified into thermal, mechanical, and chemical), and “unspecialized free ending terminals” (23). In addition to these cutaneous receptors, there are also a range of kinesthetic and proprioceptive mechanoreceptors located in the muscles, tendons, and joints that also play important roles in signaling information about the things that we touch (Lederman and Klatzky 2009: 1439). We have already discussed

210  Perception and the Sciences of the Mind

the wide array of different properties that this suite of sensory receptors can give us access to, but in addition, Matthew Ratcliffe argues that when these different types of receptors work together, they can facilitate the perception of properties that cannot be perceived by the activity of a single type of receptor. As an example, he suggests that properties such as wetness and oiliness are properties that can only be perceived by the interaction of pressure and temperature sensitive systems (2012: 419), and that in the absence of either of these systems, properties such as these cannot be perceived. When we think back to Grice’s criteria for individuating senses, these considerations can suggest that touch is not actually a single sense after all, but is rather an array of distinct senses. Physiologically, touch involves a range of sensory receptors and subsequent processing, and is receptive to a range of different stimuli. And as we saw in our discussions above, these differences mean not only that there are a range of different properties that touch gives us access to – and often our sole source of access to – but also that there is not obviously a unique proper sensible that is involved in all touch experiences. As for the question of whether touch involves a characteristic experience, the answer to this will depend on our willingness to group sensations of heat, pressure, wetness, and itchiness together as similar experiences which, if we endorse Clark’s indiscriminability-based approach for grouping experiences, are able to stand at points in a series of experiences where each pair of experiences in the series are indiscriminable from one another. For these reasons, theorists have often concluded that what we call touch is in fact “a legitimate form of multisensory perception” (Jones and Lederman 2006: 76). Yet as we saw above, this approach struggles to accommodate the fact that these distinct sensory systems not only interact to yield phenomenologically unified experiences, but also to enable the perception of properties that could not otherwise be perceived. In his recent book, Matthew Fulkerson argues that this unifying activity can actually serve as the basis of a unifying account of the sense of touch. What this shows, Fulkerson suggests, is that all of the constituent physiological systems discussed above are unified by a common goal: to identify sensory features and bind them to a single perceptual object. The sense of touch achieves this, he argues, by recruiting a unique cluster of exploratory mechanisms. Consider, for example, exploring a small metal sphere using touch: This involves reaching out and picking it up, rolling it around in our hands, squeezing it, supporting it on our palm, pressing against it with our fingers, tracing its outline. Through these actions, a number of tangible features come to be predicated of the object – solidity, smoothness, coolness, hardness, spherical shape, weight and size. (2013: 39–40) Through this array of physiological systems being recruited by a single exploratory activity – by being used, we might say, to actively touch the sphere – a range of tangible properties are identified and attributed to the single object. This,

Nonvisual Sense Modalities  211

­

The Bodily Senses and Bodily Awareness Kinesthesia and proprioception name two closely related capacities. Proprioception refers to our awareness of how our body is positioned. If you have ever tried to see whether you can, with your eyes closed, bring the tips of your index fingers together in front of your face, you have called on your proprioceptive ability to know how your limbs are located. Proprioception doesn’t require movement, however – even when still we typically know how our limbs are located in space. Kinesthesia, however, refers to our awareness of how our body is moving. Another related system is the vestibular system, which gives us our sense of balance. The existence of these capacities also raises interesting questions about whether they should be considered as involving a distinctive form of bodily perception, and hence possibly as sense modalities in their own right (e.g. O’Shaughnessy 1995). To enable us to explore these questions, it is first necessary to say a little more about the notion of bodily awareness. We are typically aware of the positions of our limbs, that we are balanced, and whether we are moving or still, but it is less clear whether this is a kind of phenomenal awareness of the sort we have been considering in our explorations of perception. Moreover, the best examples of such phenomenal bodily awareness seem to include awareness of something else in addition – for example, right now I can feel my legs, but this seems to be in part due to my feeling the points at which the chair I am sitting is pressing on my legs. Other examples

212  Perception and the Sciences of the Mind

include feeling a dull ache in my ankle or that my hands are cold. However, Brian O’Shaughnessy argues that, even though bodily awareness is typically “attentively recessive to a high degree” (1995: 175), if we concentrate our attention on a particular body part, we not only become aware of the body part to which we attend, but also to the rest of the body which functions as a background to this “figure” (1995: 183). This suggests that when we experience the properties of a surface by touch, we will also undergo certain sensations of bodily awareness, even if these are not sensations to which we would typically attend. So, for example, when I run my fingers over a rough surface, there will be certain sensations in my fingertips of irregular decreases and increases of pressure, and a sensation of my skin being dragged or pulled. O’Shaughnessy concludes on this basis that “in absolutely every instance of tactile perception an awareness of one’s body stands between one and awareness of the tactile object” (1995: 176, emphasis in original). This suggests that the sense of touch, conceived of as a sense that enables us to perceive the tangible properties of objects, is fundamentally mediated by a sense of bodily awareness: that we become aware of properties of objects by being aware of our bodies and how they are affected by those objects. However, as we saw when we discussed the sense datum theory, claims of mediated perception of this kind can seem to threaten to draw a “veil of perception” – in this case, a veil of bodily awareness – across the world of surfaces and their properties. However, Mike Martin suggests that bodily awareness, conceived of as a sense that enables us to become aware of our own bodies, and touch, conceived of as a sense that enables us to become aware of objects, are in fact two sides of the same coin. He asks us to imagine touching the rim of a glass with your fingertip: The sensation one feels in one’s fingertip feels to be within one’s body and at the limits of one’s body, at the skin. One also feels one’s fingertip to be pressing against an object … The place where one feels the sensation to be shares certain spatial properties with the object which impedes one’s movement; it is in the same place in space. (1992: 204) Importantly, though, Martin is not intending to claim that there are two experiences – an experience of bodily awareness and an experience of the rim of the glass – that are co-located, but that there is one experience that can be attended to either as a bodily experience or as a tactile perception. As Fulkerson puts it, we are able to “shift between the two experiences by selectively attending to an external object or to our bodies directly. … This shift is similar to what occurs in viewing a Necker cube” (2013: 102–3). As A.D. Smith argues, however, if my fingertip aches and I place this fingertip on the rim of a glass, this bodily sensation will be located at the same place in space as the edge of the glass, but it does not thereby give me a tactile perception of the glass’s rim (2002: 168). So not every bodily sensation will be Janus-faced in the kind of way Martin suggests. This means that we need to explain the difference between those bodily sensations that can be attended to as a tactile experience and

Nonvisual Sense Modalities  213

those that cannot. To account for this, A.D. Smith (2002) focuses on the particular tactile case of the Anstoss: the experience of there being “a check or impediment to our active movement: an experienced obstacle to our animal striving, as when we push or pull against things” (2002: 154). Smith argues that bodily sensation can support the awareness of external objects in situations where that bodily sensation is accompanied by this feeling that our exploration is countered by an external body that resists our probing.

The Anstoss and Anti-Skepticism This feature of touch experiences – of feeling that our movements are resisted by things external to us – has long been appealed to as an important element of our conviction in the existence of an external world. As an example, consider William James’s claim that “resistance to our muscular effort is the only sense which makes us aware of a reality independent from ourselves” (1890: 29). Although James does not explicitly mention the sense of touch in this passage, it seems plausible to also claim that touch is the sensory modality through which we both experience physical effort and resistance to it; indeed, as de Vignemont and Massin argue (2015: 301), we only experience physical effort inasmuch as we are aware of attempting to overcome forces that oppose the motions we produce. So the experience of resistance to activity and the experience of exerting effort to overcome that resistance are two sides of the same coin. In addition to potentially explaining our conviction in the existence of an external reality, Smith also argues that the “unique non-sensory nature of the Anstoss” serves to refute the claim that we are always directly aware of mediating sense data: “In the case of the Anstoss … it is just such focal sensations that are absent. There is simply no such sensuous item to interpose itself between us and the external physical force that we experience.” (2002: 165).

Louise Richardson (2013a) raises two further objections to the attempt to explain the relationship between touch and bodily awareness on the model of a single experience located in space at the point at which body meets world, which can then be attended to as either a bodily or tactile experience. First, it seems that the tactile sensations we experience when we feel certain qualities, such as wetness and oiliness, do not correspond to any particular bodily sensation (which is perhaps unsurprising given that, as we saw above, our perception of these qualities is mediated by multiple distinct physiological mechanisms). Second, there are cases of distal touch in which the bodily sensations that accompany touch are not found in the same spatial location as the surfaces that we feel. For example, as I stir the pot with a wooden spoon, I can feel the viscosity of its contents and, if things haven’t gone to plan, the burned bits stuck to the bottom. In such a case, whilst it is plausible

214  Perception and the Sciences of the Mind

that I can only feel these things because I have certain bodily sensations, it is much less plausible that where I feel them to be is at the very edge of my body – on the contrary, I feel the burned food to be stuck to the bottom of the pan. To deal with cases such as these, Richardson argues that we should adopt a different view of the relationship between bodily awareness and touch. She argues that in a typical case of touch, like Martin’s example of touching the edge of a wineglass, whilst it is possible to attend to the sensation qua bodily experience, this is not equivalent to attending to the feeling in your finger “independently from any relation it seems to bear to anything else” (2013a: 142, emphasis in original). In such a case, the only way to adequately describe the bodily sensation is by reference to something external – say, as having a cool, smooth edge pressing into one’s finger – and if you were to try to express the sensation without mentioning some extra-bodily object, you would end up describing a different sensation. The difference between the bodily sensations that support tactile perception and the ones that do not, such as Smith’s aching finger, is thus that they cannot be given an adequate characterization without mentioning anything external in this way. Similarly, this approach can accommodate distal touch inasmuch as it doesn’t require spatial co-location of bodily sensation and the sensation of touch: so long as the bodily sensation can only be adequately described as, say, the sensation of resisting the pressure caused by a horizontally sweeping tool that is being drawn over a rough, sticky surface at its distal end, then this bodily sensation can enable us to feel the surface at the bottom of the pan. These discussions raise interesting questions about the spatiality of touch. We might have quite naturally thought that, to the extent that bodily awareness and touch could give us information about things in space, it would be limited to the egocentric space of our bodies and their surfaces. If I am holding a glass, then what I know about the spatial location of the glass seems to be limited to its relation to my body – I know that the glass is in my right hand, but not (without using any other senses) where that is in objective physical space. As my body provides the spatial frame of reference in which touch localizes objects, this is sometimes known as the body map theory. To accommodate distal touch on the body map theory, it would seem that the tools we use would have to be incorporated into the body map, and there is indeed evidence that in monkeys, the receptive fields of neurons that encode the hand schema can extend to include tools (Iriki et al. 1996). Moreover, it has been argued that tool users, particularly experienced ones, can feel sensations in their tools (de Vignemont and Massin 2015: 306). However, there is another sense in which touch and bodily awareness can locate things outside the body: think about the proprioceptive test we discussed earlier, which involved closing your eyes and attempting to bring the tips of your index fingers together. In the process of moving your fingers toward each other, but before they touch, you have a sense both of the positions of your hands in external space and of the space between them (Martin 1992: 202). The space surrounding the body – the space which can be easily reached, and which thereby is the space available for active exploration – is called peripersonal space. It is an intermediate spatial frame between the body map and objective physical space

Nonvisual Sense Modalities  215

that, like the body map, is centered on the body, even though it extends beyond its boundaries. By now, the reader will probably already be thinking about if and how touch could be accommodated by the different theories of perception explored earlier, so I won’t spend too much time on this here. After all, we have already seen A.D. Smith arguing that the Anstoss shows that a sense datum theory cannot be true of touch, and even Brian O’Shaughnessy, who accepts a sense datum theory for vision, agrees. He suggests that touch “involves the use of no mediating field of sensation. There is in touch no analogue of the visual field of visual sensations which mediates the perception of the environment” (1989: 38). This claim – that touch is an immediate awareness of objects without the mediation of sensations – suggests that it may also be unclear how adverbialism and qualia theories would accommodate this phenomenon. This is not to say that all is well with the more direct realist approaches of representationalism and naïve realism however. For example, consider our experiences of the temperature of objects (Gray 2012). The phenomenology of thermal experience is as of departures from a neutral point: at this neutral point, we find temperatures that we don’t really experience thermally at all – they feel neither hot nor cold – but as temperatures rise or fall on either side of this point, we have distinctive thermal experiences: as temperatures rise, we feel things to be warm, then hot, then very hot; as they fall, our experiences go to cool, to cold, to very cold. There are two problems here: first is that these experiences seem to represent two distinct properties – hotness and coldness – when there is only objective physical property that it would be most natural to think that our thermal experiences represent: heat, understood as mean kinetic energy. The second is that across the temperatures that correspond to the phenomenal transition from very cold to neutral to very hot, we just see a steady increase in the mean kinetic energy of the system – there is nothing that corresponds to a decrease in a quantity as a neutral point is approached, followed by a departure in the opposite direction. And although the representationalist may be able to respond in the way Gray suggests – by claiming that thermal experiences detect not heat itself, but the exchange of heat between the body and the felt object – it is not clear whether this would be amenable to a naïve realist treatment.

The Chemical Senses Of the original Aristotelian five senses, we have now explored vision extensively, and have spent time exploring some of the distinctive questions raised by both hearing and touch above. That leaves smell and taste – our chemical senses – which both involve “contact between chemical stimuli and chemosensory receptors in the nose and mouth” (Smith 2015: 315): when chemicals stimulate the receptors in the nose, we call that olfaction; when they stimulate the receptors in the mouth and on the tongue, gustation. However, as Smith notes, there is a further type of receptor in the eyes, nose, and mouth that responds to chemical stimuli – the trigeminal nerve. When the trigeminal nerve is irritated by chemicals, this gives rise to the sensations of tingling, stinging, burning, and cooling. The stimulation is known as chemesthesis.

216  Perception and the Sciences of the Mind

Olfaction – smelling – is (relatively) straightforward: it involves molecules being drawn in through the nose when inhaling or sniffing and reaching the chemoreceptors on the roof of the nasal cavity known as the olfactory epithelium. Taste, however, is somewhat more complex. This is because there is a second route by which chemical stimuli can travel to the olfactory epithelium. In addition to the orthonasal route just described, chemical stimuli can also travel retronasally: from the mouth to the olfactory epithelium via the throat. This process is implicated in the sense of taste in addition to the purely gustatory processes involving the receptors on the tongue. If the retronasal pathway is impeded, say by a head cold or holding one’s nose, then our taste experiences are diminished. Louise Richardson describes the experience of tasting fruit-flavored sweets: with the retronasal pathway blocked, the sweets just taste sweet (and maybe slightly sour) – they lose their distinctively fruity flavors (2013b: 322). In this case, we can still taste the simple flavor components of sweetness and maybe sourness because receptors for these stimuli are found on the tongue and in the mouth. These purely gustatory receptors are thought to correspond to five basic tastes – sweet and sour, alongside salt, bitter, and umami (savory). However, Spence et al. suggest that purely gustatory taste experiences “are seldom if ever experienced on their own” (2014: 260). In the normal scheme of things, purely gustatory processes are almost always working in tandem with retronasal olfaction to produce the rich phenomenology of taste. And just to add to the complication, not only is the trigeminal mechanism also involved in taste – when irritated by chemicals, it gives rise to sensations such as the hotness of chili and the coolness of peppermint – but so is the sense of touch, where the feel of food in the mouth contributes to aspects of flavor perception involving texture, such as the creaminess, chewiness, and crunchiness of foods (Smith 2015: 317), and a thermal sense. To avoid confusion, then I will use the word “taste” to describe the everyday understanding of the sense that yields perceptions of rich, multifaceted “flavor” and “gustation” to describe the tongue-based sense that yields perception of the five “basic tastes”. One of the most distinctive features about the chemical senses is that, the particular case of color aside, they seem to be the best candidates for a purely subjective account. For example, consider the following passage from Bill Lycan where he talks about smell: Phenomenally speaking, a smell is just a modification of our consciousness, a qualitative condition or event in us. … A smell is just a quale, whether we take the quale strictly to be a first-order property of a phenomenal individual or an “adverbial” modification of an event or state of sensing. Conventional wisdom has it that qualia merely linger on uselessly in the mind, and do not represent. (2000: 281) Although Lycan does not ultimately endorse a purely subjective view of smell, he repeatedly asserts that, prima facie, smell seems to be a purely qualitative experience. Similarly, while A.D. Smith does think that we can taste things in our mouths, he contends that this is only because we have already individuated the object in our

Nonvisual Sense Modalities  217

mouths through touch: “in the absence of suitable tactile experience, experiences of taste … are mere sensations” (2011: 345). Why do the chemical senses seem so amenable to a subjectivist treatment? Well, in the case of taste, A.D. Smith holds that it is because taste “is a bodily sensation, one entirely lacking the kind of spatiality we are concerned with here. Tastes are in the mouth as headaches are in the head” (2002: 139). Similarly, Lycan says that smell is “aspatial. Phenomenologically, an odor is just with us, happening right in the center of our minds” (2000: 278). To see the sense in which this is the case, consider the smell of freshly brewed coffee drifting down the hall to my office. When I smell the coffee, my smell experience doesn’t place the coffee in the kitchen, like vision of the coffee maker does; it doesn’t even tell me that it’s coming from a certain direction, as its hissing, spluttering sound seems to. As far as the phenomenological experience of smelling the coffee is concerned, the smell just seems to be here, where I am (Matthen 2005: 284), in the vicinity of the nose (Richardson 2013b: 417). And whilst I may be able to follow the smell of the coffee to the kitchen, this is not because the smell indicates the direction of its source, but because by sampling the smells at different locations, I can determine a chemical gradient – places where the smell is more intense – and then move in that direction. Without exploration and the temporal extension this entails, though, there is no directional component to a smell experience.

Modifications of Taste Another familiar reason to think that the chemical senses should be given a subjectivist treatment comes from the Berkelian observation that changes in the perceiver can make the same object taste (and, as it turns out, smell) different. For instance, if I have a cold or have recently tasted orange juice, pineapple tastes quite different to me. This can suggest that what we typically think of as “the taste of pineapple” is not actually something in the pineapple, but a purely phenomenal feature of the way I respond to the pineapple on a particular occasion. We will return to this later when we discuss the possibility of taste and smell illusions.

Another way in which olfactory experience seems aspatial is that it doesn’t seem to individuate objects: if there are two things in my vicinity that smell similar, say garlicky, I can’t tell this from momentary olfactory experience alone. Rather, both of these garlicky-smelling objects contribute to the same garlicky odor I can smell (Matthen 2005: 285). Something similar would also seem to be true of taste: if I place two different flavored substances into my mouth, I would not be able to individuate the two distinct flavors, but would experience a commingling of the two. What would subjectivist theories of smell and taste look like? Well, as can be seen in the quote above, Lycan suggests that either an adverbial or qualia approach would be most natural (he suggests that a sense datum account of smell is less

218  Perception and the Sciences of the Mind

intuitive on the grounds that a smell experience does not obviously appear to present a smelly thing, or alternatively that the Phenomenal Principle does not seem intuitive for cases of smell). Drawing on the discussions in earlier chapters, we can see that there are a couple of ways an adverbialist could go here: they might claim that smelling and tasting are distinct modes of sensing and then distinguish between experiences within that mode adverbially – thus we smell coffee-ly when we drink the coffee and taste bacon-ly when we eat the bacon – or hold to only one mode of sensing and to distinguish between all experiences from all modalities by the adverbs alone. Thus we sense coffee-smell-ly when we smell the coffee and sense bacon-taste-ly when we taste the bacon. This approach is most similar to that of the qualia theory, save for the fact that in qualia terminology, instead of taking about adverbially modified episodes of sensing, we would talk about the olfactory qualia possessed by the experience of smelling coffee, and how that would be different from the gustatory qualia possessed by the experience of tasting bacon, and so on. As Lycan points out, however, accepting a subjectivist account of the chemical senses is likely to be the thin end of the wedge (2000: 275–6). Although it is possible that different philosophical theories of perception will be true of different senses, once we accept qualia or adverbial modes of sensing for one case, many of the reasons we have for resisting them in others would be made redundant. Suppose you aim to resist nonphysical qualia in the case of vision on the grounds that their introduction would be metaphysically excessive – well, if we have to posit them for smell and/or taste anyway, then this undercuts the motive to resist them in the case of vision. What this suggests is that for those who are inclined toward what Batty calls a “world-directed” theory of perception (2011: 164) for other senses, it is not an option to just shrug one’s shoulders when it comes to theorizing about the chemical senses and think, what will be will be. Some have also argued that a world-directed view is to be preferred to a subjectivist view on the grounds that a world-directed view is needed to capture the fact that smell and taste play important roles as sources of information about our immediate environment. Consider, for example, smelling smoke or rotten food; in both cases, smell can give us a way of accessing critical information that can then be used to ensure that our behavior is responsive to the environment we find ourselves in (Batty 2011: 170). Louise Richardson argues that this is a point at which the visuocentric nature of mainstream philosophy of perception misleads us. By focusing on the visual case as a paradigm of an exteroceptive sense – a sense that phenomenologically seems to give us access to an external world – it can seem that what grounds our taking an experience to be a perception of an external reality is its presenting us with objects that are at a distance from us and in a particular direction. She argues, however, that in the case of smell, the act of sniffing or breathing in through the nose – which involves actively bringing something into the body from outside – is a part of the overall conscious experience of smelling, which can underpin smell’s claim to also be an exteroceptive experience (2013b: 410). A similar claim could also be made for the case of taste – that the act of introducing something into the mouth and manipulating it with the lips, tongue, and teeth is a way of engaging with something external to us.

Nonvisual Sense Modalities  219

When it comes to the task of developing an appropriate world-directed theory of the chemical senses, this has progressed further in the case of smell than the case of taste. For example, Clare Batty develops a representational account of smell experience (although she suggests the view “could be appropriately translated for other world-directed views” (2011: 164)). On this view, smell experiences present us with objects – call them, for present purposes, olfactory objects – and attributes properties to them. However, to accommodate the sense in which smell is aspatial, as discussed above, she denies that representing objects in this way requires the objects to be individuated or placed in space. She draws an analogy to the visual experience of being placed so close to a colored wall that you cannot see any boundaries. In this case, the experience you have does not enable you to discriminate the object you see (the wall) from any other objects. In such a case, she suggests, your visual experience will attribute a color property to an object without being able to specify which particular object in space it attributes the property to. In a case like this, then “the content of your experience will be unquestionably abstract: there is something or other before you that is colored” (2011: 172 emphasis in original). The case of smell is similar, save for the fact that the inability to individuate and spatially locate olfactory objects is intrinsic to smell as it is not to vision: smell experience represents that there is something or other that is broadly “here” – where you are – that is smoky, or lemony, or has some other olfactory property. To fill out this philosophical theory of smell, we need to say something about what olfactory objects are. Everyday language can suggest that olfactory objects are just the familiar everyday objects that we can see and touch: after all, we talk about smelling the coffee, or the wet dog, or the burnt toast. However, there are problems with this view. For one, smells can often persist after the smelly object has ceased to exist: where certain foods are concerned, we can still smell them long after they have been consumed, and the pots and pans washed up. For another, according to the theory just presented, the olfactory objects are represented to be here – where the subject is when they have the experience. As in the coffee case earlier, this may well not be at the location that the smelly object or substance is. We might be tempted to say that in such a case, I must be subject to an illusion: that my smell experience represents the coffee as being here (where I am – in the office) when it is in fact in the kitchen. Yet to treat such cases as illusory seems implausible. In the vast majority of cases in which we might talk about my smelling an everyday object, that object is not going to be located precisely where I am. Moreover, we don’t tend to get confused by smell experiences by thinking that when I smell the popcorn at the movie theater, the popcorn is going to be right in front of me. Instead, it seems like what we are smelling, in such a case, is what we would naturally call the smell of coffee or the smell of popcorn, although to avoid potentially confusing the olfactory object and the olfactory sense, we will call the objects of smell odors. What are odors? Well, to accommodate the fact that odors seem to occupy volumes of space, and can be detached from its source object such that the odor can sometimes even outlast the object, odors would need to be treated as a kind of particular that has the capacity to fill a volume of space. For this reason, the standard approach is to identify odors with the clouds of airborne molecules that are typically given off by smelly objects

220  Perception and the Sciences of the Mind

or substances (Lycan 2000: 281; Batty 2011: 532), and which possess olfactory qualities in virtue of the particular chemicals that comprise them (Young 2020: 4). Given this, we might think that the structure of smell experience mirrors that of hearing, where we hear the car by hearing the sound it makes rather than vision or touch, where we can perceive the car directly: in the case of smell, the thought would be, we smell the coffee by smelling the coffee’s nutty, caramelly odor (Tye 2002: 142; Budek and Farkas 2014). However, Batty seems to resist this view on the grounds that odors can outlast their sources – if I smell an odor and the experience I have represents its source object, then if that source object has ceased to exist it would appear that my experience must be non-veridical (Batty 2010; for further discussion, see Cavedon-Taylor 2018). As with the cases discussed above, however, treating this case as illusory seems implausible, suggesting that olfactory experiences are of odors alone. Vivian Mizrahi objects that this austere view fails to explain the ways in which smell leads to object-directed behavior: “If smell were directed only toward the air surrounding us, our actions based on olfactory experiences and directed towards objects would be quite mysterious” (2014: 245). To accommodate this, she develops a theory of odors as properties, but instead of being properties of objects, she suggests that odors are properties of stuff, where stuff refers to the matter of which objects are composed. (If you want to see the difference, she suggests, look at whether we refer to something using a mass noun or a count noun: objects can be counted, stuff can’t.) Instead of being a property of the rose then, its odor is a property of the stuff of which the rose is composed. But when this stuff is volatile, a portion of the stuff will be evaporated into the air, at which point it can be drawn into the nose and smelled. So odors are found both in the air and in the objects themselves: and by smelling an airborne odor, we are smelling properties of the stuff of which an object is comprised – just that portion of the stuff that has been evaporated into the air. Once again, these broadly reductive theories of odors raise particular challenges for naïve realist theories and their claim that the phenomenology of olfactory experiences are grounded in acquaintance with the phenomenal qualities of odors themselves. Not only does this require a robust realism about olfactory properties – holding that the properties that characterize the odor of coffee are objective properties of the odor, and not merely a feature of our idiosyncratic responses to it – but also that these phenomenal properties are literally possessed by clouds of molecules. However, there may be primitivist alternatives available, such as the view recently developed by Louise Richardson (2018). This theory blends features of Wilson and Stevenson’s (2006) view, which we met in Chapter 5, with Mizrahi’s theory of odors as stuffs, and argues that whilst odors are not stuffs per se, they are parcels of qualitative stuff. In describing stuff as qualitative in this way, Richardson insists that: a stuff’s being vanillary [should not] be understood as being the physical property of the stuff that triggers … the perception of vanillariness. … [A] quality like vanillariness should be understood as supervening on, but not being identical to this physical property. This means that the stuff of which an odour is a parcel can be physically heterogenous. What makes the stuff what it is, is its olfactory quality or olfactory qualitative character. (2018: 108)

Nonvisual Sense Modalities  221

When it comes to the case of taste, much less work has been done on the nature of the objects of taste (flavors). As mentioned above, there is of course the view that taste is purely subjective, but in recent work a world-directed view of flavors is suggested. For example, in a 2020 paper, Barry Smith hints at a view of “flavors as emergent properties dependent on but not reducible to the underlying chemical properties … which we track by means of integrated multisensory flavor perceptions” (2020: 46). This has strong echoes of Richardson’s theory of odors, and indeed Richardson herself has recently suggested something extremely similar: to think of flavours as emergent properties, had by substances (often food stuffs), that are made up of component substances – odorants, tastants, trigeminal irritants – that on their own lack this property type and which it therefore requires multiple sensory systems to perceive. (forthcoming) It will be interesting, however, to see whether this kind of view can be developed to accommodate the fact that the way substances taste can be significantly affected by such things as temperature (Spence et al. 2014: 253): should the temperature of substances be treated as a component of flavor, such that things literally have different flavors at different temperatures, or should temperature be treated as one factor that can facilitate or thwart attempts to veridically taste a substance’s flavor? For this latter possibility to be an option, it would need to be possible that our smell and taste experiences mislead us as to the real nature of the odors and flavors in the world (Lycan 2018: 31). And there certainly seem to be cases where we are wrong, or at the very least inconsistent, about the flavor and/or basic taste that substances have: for example, Spence (2017) describes empirical studies that show that when served in a white pot, strawberry mousse is rated 10% sweeter than when served in a black one; that when weights are added to a yogurt container, people take it to be richer and creamier; and that when the same liquid is colored with red or green dye, subjects describe the green liquid as sourer and the red liquid as sweeter. Similar results are also found in the case of smell. For instance, Richard Stevenson (2011) describes an experiment in which participants were exposed to either several citrus-smelling odors or several woody-smelling odors, and then asked about the smell of the chemical dihydromyrcenol: subjects who had been exposed to the citrus context described it as woody, whereas subjects who had been exposed to the woody context described it as citrusy. The subjectivist of course has a ready explanation of these findings: we would expect differences such as these, because the experienced smell or flavor is a construct of the idiosyncratic ways we process sensory information on any given occasion (Shepherd 2011; Small 2012). (Although it is worth noting that Barry Smith (2020: 40) argues that this is actually a problem for the subjectivist view in that it gives up on the possibility of there being illusions of taste/flavor or smell.) What explanation can the defender of a world-directed view offer of such cases? One response is that these cases are illusory: “Spence’s carefully constructed experimental setups are cases of perceptual illusions: ways in which our perception of flavor is

222  Perception and the Sciences of the Mind

being influenced or distorted by the impact of other sensory impacts” (Smith 2020: 42). Clare Batty (2014: 7) rejects the description of the smell-based cases as illusions, on the grounds that they don’t involve misrepresentation of a particular object, but she nonetheless accepts that they are non-veridical (2014: 7). If they are treated non-veridically in this way, then as was the case with illusions in the other sense modalities, this line will be somewhat easier for representationalists, who ground the phenomenology of such a case in the way the world is (mistakenly) represented to be, than it is for naïve realists, who ground the phenomenology in acquaintance with a phenomenal particular. Yet as Richardson points out, a Brewer-style approach seems available in such cases: “Perhaps the hot chocolate in the orange cup tastes exactly as chocolatey as that which arrives in a white cup, and the colour of the cup affects just what the subject thinks and says about how it seems” (Richardson forthcoming, my emphasis). Before we conclude our discussion of the chemical senses, I want to briefly return to the question with which we began this chapter – how we individuate the senses – by way of an important discussion in the philosophical literature on taste, which is how should we conceive of the sense of taste itself. The view of many scientists working in this area is that the sense of taste and gustation are one and the same thing – that the sense of taste is based around the receptors on the tongue and in the mouth, and has only the basic tastes as proper objects. In the terms of our discussions at the beginning of this chapter, this looks to be characterizable as a sophisticated version of the characteristic sense organ view, just at the level of receptors. This approach begins from the start point that taste is what happens in the mouth when we eat or drink. Then, science investigates what goes on in the mouth when we eat or drink and discovers that there are a range of receptors in the mouth and on the tongue. These are thus identified as the taste receptors, which then fixes what we mean by taste: taste is the mode of sensing that involves the taste receptors. It also fixes the characteristic stimuli of taste: the sense of taste is stimulated by the tastants that correspond to the five basic tastes. Given this, the rich multifaceted experience of flavor must depend on “the multi-modal combining of inputs from different sense modalities into a unified flavor percept” (Smith 2015: 317), involving not only gustation and retronasal olfaction, but also touch in its thermal, trigeminal, and pressure-based forms. This means that “flavor experiences are always defined in a multisensory space” (Spence et al. 2014: 254). A concern with this approach is that it has the consequence that we are often just wrong about features of our pre-theoretic understanding of the chemical senses. For example, when we eat pineapple or cherry, it seems to us as though the flavor of the fruit is located where the fruit is – that is, in the mouth. Yet the particular fruity tastants of pineapple and cherry – over-and-above the sweet and sour tastants that bind onto the receptors in the mouth – are actually transduced in the olfactory epithelium. So one way in which we are wrong is in where we take the flavor to be: we thought it was in the mouth, but really it is the nasal cavity. “This location illusion is due to [the phenomenon of] oral referral” (Smith 2015: 324). In addition, there also appears to be a “constant error made, when eating, of attributing to taste what really belongs to the sense of smell” (Auvray and Spence 2008: 1023).

Nonvisual Sense Modalities  223

We thought we were tasting the pineapple-y flavor of the pineapple, but we were wrong – as those stimuli are transduced on the olfactory epithelium, and the olfactory epithelium hosts the receptors dedicated to olfaction, it turns out we were smelling it after all. Some philosophers have thought that the violence this does to our everyday conception of the senses is too great. In the previous section, we discussed Mohan Matthen’s activity-based approach to distinguishing the senses. This suggests the possibility of an alternative approach on which we start from the same point – that taste is what happens in the mouth when we eat or drink – but then rather than look for the receptors that are located in the mouth and identify these as “the” taste receptors, we can focus on the fact that these activities – eating and drinking – give us access to flavor. Science can then investigate the range of sub-personal processes that are involved in the process of tasting flavors when eating and drinking. And while science did discover that this involved the taste buds in the mouth, subsequent discoveries found that taste also involves the olfactory epithelium, the trigeminal nerve, and so on. These can then be identified as the sub-personal underpinnings of a single sense. This approach suggests that what individuates taste as a unique sense is not something that science can in principle discover but is something that is fixed conventionally – by our social practices of explaining and understanding behavior (Nudds 2003, 2011; Richardson 2013b). On this view, we identify taste as a unique sense because it matters to us, as human beings, whether we perceive an object by tasting it as opposed to touching it, seeing it, smelling it, and so on. This is not to say of course that science is irrelevant to our understanding of the senses: on the contrary, science can and does contribute to our understanding of these senses; it just cannot “correct (or confirm) any commitments we have about what individuates the senses” (Richardson 2013b: 328). Among the things that science can’t show, on this view, is that flavors are really perceived in the nose or that we can only really taste sweet, sour, salty, bitter, and umami. What science can show us, from this perspective, is what the sub-personal mechanisms that underpin our senses are and how they function. And where the chemical senses are concerned, among the things that science shows are that the sense of taste is supported by numerous distinct physiological mechanisms, and that surprisingly, two senses – smell and taste – actually share a sub-personal mechanism (Richardson 2013b: 337).

Questions •

•

•

What do you think is the key difference between the different senses? To what extent do you think science can correct our everyday conception of the senses? Clare Batty suggests that her representational account of smell experience could be appropriately translated for other world-directed views, such as naïve realism. How might the naïve realist translate this view? Given the differences between the different senses, what are the prospects for a unified philosophical theory of perception?

224  Perception and the Sciences of the Mind

Further Reading Fiona Macpherson’s 2011 collection The Senses: Classic and Contemporary Philosophical Perspectives is a great start point for thinking about individuating the senses. It contains many of the papers discussed in this chapter, including the classic papers by Grice and Roxbee-Cox, as well as more recent papers by many of the other philosophers we have discussed, including Keeley, Nudds, Martin, O’Dea, and A.D. Smith. Stokes, Matthen and Biggs 2015 collection Perception and Its Modalities is also an excellent resource. It doesn’t cover the classics, as Macpherson’s collection does, but it makes up for it with an excellent range of up-to-date articles that covers a lot of the topics and authors found in this and the next chapter. Casey O’Callaghan’s Sounds (2007) is an excellent discussion of auditory perception, and Matthew Fulkerson’s The First Sense: A Philosophical Study of Human Touch (2013) covers the key issues concerning touch. A.D. Smith’s The Problem of Perception (2002) contains his argument that the Anstoss shows that sense datum theories of perception cannot accommodate touch as well as discussions of all the other nonvisual senses. For smell, explore Clare Batty’s “Smelling Lessons” (2011), Vivian Mizrahi’s (2014) “Sniff, smell, and stuff”, and Louise Richardson’s (2018) “Odors as olfactabilia” For taste, Spence, Auvray, and Smith’s (2014) “Confusing tastes with flavours” is a good introduction to the science-inspired view of taste as involving multiple modalities, and Louise Richardson’s (2013) “Flavour, taste and smell” is a good counterpoint.

9

Multimodality

Overview In this chapter, we will build on our discussions about the different sense modalities to ask how the different senses interact in the construction of our rich perceptual experiences of the world. We begin by discussing the claim that, while the experiences corresponding to the different modalities are distinct, perceptual experience itself could be considered multimodal on the grounds that the modality-specific experiences are all co-conscious. We then move on to investigate stronger ways in which perceptual experience could be said to be multimodal. The chapter concludes by considering how the discussions in this chapter might impact upon the philosophical theorizing about perception that comprised the first part of this book.

As we have seen throughout this book, a lot of philosophical work on perception has not only tended to focus on vision, but has also tended to assume that the senses are relatively distinct from one another. As we began to see in the previous chapter, however, this has recently begun to change, with more research being undertaken into the nonvisual senses. However, even this research has tended to operate on a sense-by-sense basis, as though assuming that all we need to do to provide a complete theory of perception is to provide a theory of each individual perceptual sense. In this final chapter, we will explore this assumption. Tim Bayne calls the picture that underlies the view that theories of each individual perceptual sense can simply be compiled to provide a complete theory of perception the decomposition thesis (2014: 15) and Casey O’Callaghan calls it the composite snapshot conception (2008: 320). This conception of perceptual experience assumes not only that our total perceptual experience at any given time is constituted by the experiences corresponding to the different sense modalities at that time, but also that these modality-specific experiences are all there is. So our total perceptual experience at a time is both constituted by, and exhausted by, our modality-specific experiences. Moreover, there is also an underlying assumption that the experiences

226  Perception and the Sciences of the Mind

we enjoy in one modality do not depend, in any important way, on the experiences we have in other modalities. According to this understanding, one’s total perceptual experience at a time is an assemblage or composite of modality-specific experiences. Perceptual experience comprises discrete, modality-specific components or ‘snapshots’. Each such modality-specific experience has its own recognizable and distinctive character. (O’Callaghan 2008: 321) As is made clear in O’Callaghan’s quote, a component of this picture is the idea that each sense modality is associated with a distinctive kind of phenomenal character, which motivated the suggestion met in the previous chapter that we might distinguish the senses from one another on the basis of their characteristic phenomenology. As we saw when we discussed this characteristic experience view in the previous chapter, one motivation for this view – and consequently for the composite snapshot conception/decomposition thesis – is the idea that the difference between the phenomenal characters of different perceptual modalities is just something that is made obvious to us in introspection, even if we do find it difficult to articulate exactly what the differences between the phenomenal characters are. Regardless, the phenomenal character of, say, a touch experience just could not be the phenomenal character of a hearing experience or of a visual experience. Another reason to think that our total perceptual experience at a time is made up of distinct modality-specific strands is that we know that it is possible to lack or to lose one or more of those senses. If somebody is totally deaf, for example, then they would completely lack auditory experience; yet this does not preclude the deaf person from enjoying vision or indeed sensory experiences from any of their other sense modalities (O’Callaghan 2012: 94). This can suggest that each sense is an explanatorily independent, parallel stream of information and phenomenal character that could, in principle, be stripped off with no loss to the other streams (although for reasons given in the previous chapter, the sense of smell is a counterexample to this intuitive line of thought). As we saw in the opening chapter, however, this is just the kind of assumption that can play a critical role in shaping both the scope and limits of the philosophical and empirical theorizing that then takes place. In this case, once the decomposition thesis paradigm is in place, a self-sustaining philosophical-empirical ecosystem arises around it. When we try to empirically investigate one of our sensory systems, like the good experimentalists we are, we try to hold all other variables constant to avoid unwittingly allowing in confounding factors. In a laboratory context, this means that we try to present and vary stimuli only to the sense we are studying, and to ensure that the stimuli to all the other senses are held constant. Yet this experimental methodology all but guarantees that we will not find any interesting interactions between the senses. This, in turn, can seem to corroborate the decomposition thesis paradigm and so on.

Multimodality 227

It is in this context that phenomena like synesthesia can serve to make us question these assumptions. As we saw in Chapter 7, synesthesia can involve stimulation to one sense modality inducing idiosyncratic experiences in a different modality. Although synesthesia is relatively rare, it does nonetheless suggest that there might be interesting and important connections between what the decomposition thesis paradigm tells us are distinct sensory channels. As we shall see, the intermodal phenomena to be explored in this chapter are not idiosyncratic in the way synesthesia is. On the contrary, they are both far more robust in their operation and far more common in the population. Indeed, as we shall see, there are reasons to think that interactions between what we thought were distinct sense modalities may even underpin a great deal of everyday experience. This can, in turn, lead us to question the decomposition thesis: are our everyday perceptual experiences really both constituted by and exhausted by our modality-specific experiences?

Minimal Multimodality and Phenomenal Unity The basic idea of the decomposition thesis, as initially presented by Bayne, is that one’s total perceptual experience at a time is made up of modality-specific experiences. Yet is that really all there is to say about one’s total perceptual experience at a time? Michael Tye, for example, suggests that it just doesn’t feel like we undergo five different simultaneous perceptual experiences; instead, “it is phenomenologically as if I were undergoing one” (2003a: 17–18). Similarly, Bayne himself asks us to consider: what it’s like to hear a rumba playing on the stereo while seeing a bartender mix a mojito. … these two experiences possess a conjoint experiential character. … Any description of one’s overall state of consciousness that omitted the fact that these experiences are had together as components, parts, or elements of a single conscious state would be incomplete. (2010: 11) This phenomenon – that the streams of phenomenal consciousness that correspond to the different sense modalities are experienced as being phenomenally unified – suggests that the decomposition thesis alone fails to completely capture what it is like to be a perceiver with multiple senses. Fiona Macpherson (2011b) attempts to explain this by contrasting our case with one way we might think about what it is like for a “split-brain” patient (Gazzaniga 1998). Split-brain patients have undergone a surgical procedure known as a commissurotomy, in which the corpus callosum – the bundle of nerve fibers that connects the two hemispheres of the brain – is severed. In laboratory contexts, it is possible to set up tests on such patients in which visual images are presented to a single eye, which is, in turn, processed by a single hemisphere of the brain. By eliciting appropriate behavioral responses, it can be determined that the left eye – information from which is processed by the right hemisphere, which is not

228  Perception and the Sciences of the Mind

involved in linguistic activity – has seen and processed the presented images, but because the connections between the hemispheres have been severed, the subject (strictly, the subject’s language-controlling left hemisphere) knows nothing about them, and is only aware of the images that have presented to the right eye. A standard philosophical theory about what is going on here is that such a subject has two independent streams of conscious experience (Macpherson 2011: 452; although see Bayne 2010: 209–14 for an alternative understanding). As the decomposition thesis suggests that we have five independent streams of conscious experience delivered by each sense, one possibility would therefore be that these are fully independent, like the two streams of consciousness in the split-brain patient. Yet that is manifestly not what it is like for most of us. For this reason, we might think that Tye is right and that, even if each sense delivers a relatively independent stream of phenomenal character, they are nonetheless still unified into a single perceptual experience that is made up of multimodal phenomenal character. This gives us what Briscoe says is a minimal sense in which “perceptual consciousness might be multisensory…: experiences in different modalities are sometimes simultaneously co-conscious” (2016: 121). Explaining this co-consciousness is not straightforward, however. For this reason, O’Callaghan adds a rider to the composite snapshot conception to give us the thesis of minimal multimodality: “The phenomenal character of each perceptual episode is exhausted by that which could be instantiated by a corresponding merely visual, merely auditory, merely tactual, merely gustatory, or merely olfactory experience, plus whatever accrues thanks to simple co-consciousness” (O’Callaghan 2015: 562, my emphasis). By something that “accrues thanks to simple co-consciousness”, it appears O’Callaghan has in mind additional (and therefore non-modality-specific) phenomenal character introduced to capture what it is like for one’s experiences to feel unified in this way. Exploring whether or not this does indeed require additional phenomenal components would take us too far from our present purposes, but it is worth noting that Tye (2003a: 22) explicitly argues against the view that phenomenal unity can be explained by supplementing the five streams of sensory phenomenal character with an additional unifying phenomenal character. When it comes to positive accounts of the unity of consciousness, both Bayne and Tye endorse a view that explains unity in terms of the subject being in a single conscious state (Bayne 2010: 16) or a single experience (Tye 2003b: 97). The disagreement between Tye and Bayne seems to turn on whether this unifying experience has modality-specific experiences as proper parts: Bayne says yes, Tye thinks not. For our purposes, however, the key thing to note is that, to the extent that the unity of consciousness is explained by a subject being in a single conscious state or having a single experience that accounts for all aspects of the subjects phenomenal life at that time, these experiences/conscious states will be fundamentally multimodal. There is, however, a skeptical alternative to the claim that perception is multimodal, even in this limited sense. Spence and Bayne argue that our introspective confidence in phenomenal unity may not be entirely reliable. After all, phenomenal unity does not claim merely that perception is multimodal over a period of time – certainly, it seems, introspection can deliver a reliable positive answer to that – but

Multimodality 229

that perception is multimodal at any single moment in time. As we saw when we discussed change and inattentional blindness in Chapter 6, introspection can be misled by the fact that when we attend to something to check that it is there, we thereby become aware of it. In the change/inattentional blindness cases, this gave rise to the sense that we were aware of all of the rich detail of the scene in front of us – because whenever we looked to check if we were aware of some item of detail, we were aware of it – despite the fact that empirical evidence suggests that this may not be the case. A similar explanation could be given of our introspective confidence in phenomenal unity. Right now, I’m conscious of visually perceiving the words that are appearing on the screen as I type. Am I also perceiving auditorily? Let me check: yes, I can hear the birds roosting outside my office and the dog barking at a passing car. Am I also perceiving tactually? Let me check: yes, I can feel the heels of my hands on the table and the keyboard keys under my fingertips. Perhaps, Spence and Bayne suggest, something similar to this but much faster is happening to give us the introspective sense of phenomenal unity: that my conscious experience is in fact rapidly switching between inputs from the different sense modalities as my attention switches (2014: 104–5).

Intermodal Feature Binding and Object Unity O’Callaghan’s thesis of minimal multimodality provides a very basic sense in which we might say that perception is multimodal: that the deliveries of all of our different sense modalities are somehow brought together into a single multimodal experience or conscious episode. However, we might think that minimal multimodality is multimodality in letter only and not in spirit. After all, as far as this position is concerned, there need be no interaction between the senses: each can work on its own, independently from the other senses, so long as it delivers its results into a unified experience. But when we ask whether perception is multimodal, I think we may have a more robust thesis than this in mind. One way in which perception has been argued to be slightly less minimally multimodal than this is in the sense that the deliveries of the different sense modalities are not just co-conscious, but that different features found in the modality-specific streams can be bound together. For example, consider the following passage from Ophelia Deroy concerning our experience of a whistling kettle: As I hear the whistle of the kettle while I am writing this line, I am conscious of the kettle’s whistle. … As I turn my head toward the stove, I am still conscious of the sounds, but I am also conscious of the colored shape of the steaming kettle. … I experience these auditory and visual features as belonging to one and the same object: the whistling, gray kettle. (2014a: 105) Deroy goes on to call this phenomenon object unity: the phenomenal conjunction of features perceived by different modalities in a unique perceptible object. To

230  Perception and the Sciences of the Mind

achieve object unity, features that are accessed through different sense modalities must be bound together by a process that is often called intermodal feature binding. These phenomenological considerations in favor of intermodal feature binding can also be supplemented by findings from perceptual science. O’Callaghan (2014: 80–2) describes a range of empirical evidence for the claim that not only do the different sensory systems interact and share information, but that they integrate information from different senses into the same “object file”, where an object file is a name for a “short-lived perceptual representation that functions to keep track of an object across time and space and to store information about its potentially changing properties” (Briscoe 2017: 3).

The Binding Problem Intermodal feature binding is a cross-modal example of a phenomenon that has been well studied in the intramodal case. In vision, for example, we have reason to think that the different visible features of an object – color, shape, motion, size – are processed in different parts of the brain, so given this, we might wonder how these features are brought back together again. Anne Treisman (1996) refers to the challenge of discovering how the visual system succeeds in grouping the right properties together as the “binding problem” (Treisman and Gelade 1980). As Deroy (2014a) notes, it is important to keep the difference between the notion of feature binding and the notion of object unity clear: binding is a sub-personal operation that results in distinct features being attributed to a particular object in visual processing; object unity, however, is a conscious, personal-level phenomenon that occurs when distinct properties are phenomenally perceived to be instantiated by the same object. In addition, there is also the neuroscientific phenomenon of multisensory processing, in which information from different sense organs is combined in the brain, and multisensory integration, in which this combination of information yields “genuinely new information that differs from the information deriving from each sense organ” (Macpherson 2011: 430). Both Deroy and Macpherson go on to explore the relationships between these different level processes in detail.

Again, however, Spence and Bayne contend that evidence of intermodal feature binding in the sub-personal perceptual system cannot straightforwardly be taken to be evidence of object unity. The fact that the perceptual system binds features from different senses together into a single object file might show that those features are simultaneously represented to be possessed by a single object, they suggest; it does not without additional argument or assumptions show that these features will be simultaneously conscious. As we saw earlier, it remains possible that an experience that appears to exhibit object unity could in fact be

Multimodality 231

“structured in terms of multiple instances of unimodal perceptual objects” (2014: 119), with the apparent object unity of the experience being explained as a cognitive or inferential phenomenon. In the kind of case Deroy describes, then, the thought would be that this experience can be explained in a way that is consistent with the thesis of minimal multimodality as follows: we visually perceive the steaming kettle at a particular location, and simultaneously auditorily perceive a whistling noise emanating from the same location. Each of these perceptual experiences is entirely contained within its own modality-specific stream, but because we can switch attention between these streams extremely rapidly, they are experienced (essentially) simultaneously, or at least in such a way that makes it compelling to judge that the steaming kettle we see is the source of the whistling noise that we hear. In response to this kind of objection, Deroy argues that there is an important phenomenal difference between a situation in which we perceive a sound we hear to be produced by an event we see and a situation in which we merely hear and see them to occur at the same place and time. She describes the experience of watching a movie as the audio is slowly desynchronized from the video: If the lip movement and sounds get badly out of synchrony, say by 200 milliseconds or more, I will suddenly experience the auditory event and the visual event as distinct. Object unity has suddenly been lost and phenomenal unity is all that I am left with. (2014: 112) The phenomenal difference between these two cases – mere phenomenal copresence and binding – can seem significant. This, together with recognition that we cannot bring about the phenomenal switch simply by judging that the offset voice we hear is supposed to be produced by the speaker, suggests that this is a truly perceptual phenomenon. O’Callaghan also argues that this kind of phenomenal switch has some interesting consequences for the intuitive idea that each sense modality has its own distinctive phenomenal character. First, phenomenal switches suggest that not all perceptual phenomenal character will be modality-specific: if there is a phenomenal difference between perceiving that something is round and something is loud at a particular location, and perceiving that some one thing is both round and loud at that location (Macpherson 2011: 449), then this overall phenomenal difference will not be reducible to a difference in either the visual or auditory phenomenal character. As there is no reason to think this phenomenal difference would be grounded in some other modality-specific phenomenal character, then it must be grounded in a phenomenal feature that is not associated with a particular modality. Not only does this begin to loosen the grip of the intuition that we just know, on the basis of introspection, that perceptual phenomenal character can be decomposed without remainder into modality-specific components, it also challenges the intuition that for any given modality, we know what its distinctive modality-specific phenomenology is.

232  Perception and the Sciences of the Mind

Suppose that I am perceiving an object through both vision and touch. It may be that I can know that I am having both a distinctively visual and a distinctively tactile experience, because I am aware of the object’s color and texture. I am also aware that I can perceive the object’s shape, but O’Callaghan argues that as not every phenomenal feature must belong to a particular modality-specific phenomenal character, I “may not be able to tell what modality a phenomenal feature is associated with on an occasion” (2014: 89). Given this, why should we expect there to be distinct phenomenology corresponding to both a visual and a tactile experience of the object’s shape? This means that: there may be no clear verdict concerning which phenomenal features, among many candidates, belong to the distinctive overall character that is associated with a given modality [and that] the boundaries of the phenomenal character associated with a modality on an occasion may not be introspectable. (2014: 89–90)

Molyneux’s Question The question of the relationship between our phenomenal awareness of the same object across different sense modalities is connected to a historically important question in the philosophy of perception known as Molyneux’s Question. The original Molyneux’s question concerned the relationship between tactile and visual experiences of common sensibles such as shape, which are available to both senses. It is named after William Molyneux, who wrote to Locke in 1688 asking whether someone who was born blind, and had learned to distinguish a globe from a cube by touch, would be able to distinguish them by sight alone should their sight be restored. As well as being an intriguing question in its own right, this possibility has held to be particularly pertinent to the philosophy of perception on a number of fronts. For example, John Campbell suggests that Molyneux’s question bears on the topic of “whether there is any difference between visual and tactual experience of shape itself; [that is,] whether the shape perception itself has a different phenomenal character in sight than in touch” (1996: 301–2). Campbell argues that different answers to Molyneux’s question correspond to different views of phenomenology. On what he calls an internalist view – the kind of theory that holds our perception of properties is not direct, but is mediated by awareness of something else such as qualia – “the primitive consciousness of shape might be quite different in the different sensory modalities, which will all have their own phenomenal geometries” (1996: 302). If, as the decomposition thesis maintains, each sense modality has its own distinctive phenomenology, then the phenomenology associated with a cube in one modality (touch) would be different to the phenomenology associated with a cube in any other modality (vision). If this is so, then we would not expect someone newly sighted to be able to associate these two phenomenal characters; instead, this association must be something learned. Given this, “a negative answer to [Molyneux’s] question usually implies a theory of perception in which access to a meaningful external world is not direct but mediated” (Gallagher 2005: 154).

Multimodality 233

Campbell contrasts this with what he calls an externalist view, which holds that where different sense modalities pick up on one and the same property, they will be phenomenally similar. Of course, as each sense will pick up on other properties besides shape, there will be significant phenomenal differences between the two experiences, but so far as their experience of shape goes, “shape perception will be amodal, since the different senses will be picking up the very same properties of the objects around one” (1996: 303). This is essentially the view that we saw O’Callaghan defend above. If the phenomenal characters of tactile and visual experiences of the same shape are the same in this way, then – all else being equal – we would expect someone who encounters the visual phenomenal character for the first time to be able to recognize that it is a phenomenal character that they have encountered before. If the radical externalist view is correct, then we would expect a positive answer to Molyneux’s question to be forthcoming. Schwenkler (2013: 87) provides a useful way of getting to grips with these two alternatives by drawing an analogy to our perception of the temporal features of events. Suppose someone born blind were to develop the ability to know which of two touches on the skin happened first or which was longer, and now suppose that person was enabled to see. We would expect this capacity from touch perception to carry over to visual perception, such that the subject would immediately be able to tell which of two pulses of light happened first or which was longer – no additional learning would be required. This suggests that the phenomenology of succession is the same in sight as it is in touch. In contrast, suppose that someone born blind were to hear the sound of a bell, and then given sight. We would not expect this person to be able to recognize, without further investigation, a bell: the association between the look and the sound of a bell is something which has to be learned. In this light, Molyneux’s question is then: is the relationship between visual and tactual perceptions of shape more like the first or the second case? A closely related way of viewing the import of Molyneux’s question concerns the way we are aware of space across different sense modalities. We can bridge from shape to space by interpreting Molyneux’s question as a question concerning our concepts of shape – and in particular, whether the concepts of shape we exercise in vision and touch are the same or different – and then noting that our concepts of shape are what they are because of the relationship they bear to space (Evans 1985: 384). Under this interpretation, Molyneux’s question becomes that of whether the different spatial frames of reference – the egocentric body map for touch and public physical space for vision – are themselves modality-specific, having their own distinctive way of representing space. If this is the case, then we would have to learn correlations between the two. This was Russell’s view: space as we see it is not the same as space as we get it by the sense of touch; it is only by experience in infancy that we learn how to touch things we see, or how to get a sight of things which we feel touching us. (1912/1967: 29) If on the other hand the answer is no – that spatial frames of reference are not modality-specific – then there could be a single, shared representation of

234  Perception and the Sciences of the Mind

space with each different sense modality mapping the features it discovers into this shared space (Matthen 2014). Under this interpretation of Molyneux’s question, our ideas of shape will count as modality-specific just in case our ideas of space are modality-specific (Matthen and Cohen 2020: 50): so a negative answer to Molyneux’s question would suggest distinct modality-specific spaces, and a positive answer to a common shared space. If the foregoing is correct, and we can establish correlations between philosophical theories and answers to Molyneux’s question, then this may merely provide a way of highlighting the consequences of different philosophical theories. This approach views Molyneux’s question as just a thought experiment. However, if Molyneux’s question could actually be answered by empirical investigation, then the preceding considerations suggest that we may be able to convert some knotty philosophical problems into claims that are, at least in principle, empirically testable. For example, if Campbell is right that different responses to Molyneux’s Question would support/undermine internalist and externalist theories of phenomenology, then all we need to do to resolve these philosophical questions is to find blind subjects, give them sight, and ask them. And this is exactly what Richard Held and colleagues (2011) did: they took a group of subjects with either dense congenital bilateral cataracts or bilateral congenital corneal opacities and restored their sight. Shortly after surgery – all within 48 hours – they then presented these subjects with a matching task, in which one object was presented – either visually or haptically – before being removed and replaced with two objects: one which matched the original object (the target) plus a distractor. This second presentation could be in the same modality as the first, creating touch-to-touch and vision-to-vision conditions, or it could be a touch-to-vision (Molyneux) condition. Held et al. found that in the touch-to-touch and vision-to-vision conditions, subjects performed well, with success rates of 98% and 92%, respectively. In the Molyneux condition, however, subjects’ performance was significantly worse at a near-chance 58%. Held et al. claim that these results “suggest that the answer to Molyneux’s question is likely negative” (2011: 552). Do these experiments constitute an empirical resolution of a philosophical question? Well, as we saw in our discussions of both naïve realism and color science, empirical results are always open to philosophical critique and reinterpretation. Given the nature of these particular experiments, a positive result could be explained away on account of the subjects potentially having had the ability to learn certain associations earlier in life if the blindness were incomplete or of sufficiently late onset; and a negative result as due to “neuronal deterioration or transformation in the visual cortex of the congenitally blind person” (Gallagher 2005: 167). Another possibility would be to suggest that the experimental methodology was flawed in some way. In the case of Held et al.’s experiments, Schwenkler argues that the nature of the visual discriminations required in the vision-to-vision task were significantly different to the visual discriminations that were required in the touch-to-vision task (2013: 91–2). He argues that success in the vision-to-vision task only required the subjects to be able to match low-level features such as color,

Multimodality 235

shadow, and approximate overall contours – they didn’t need to see the objects as three-dimensional at all, and later experiments gave reasons to think that the subjects were failing to do so. (Schwenkler (2013: 93) puts this down to a flaw in the methodology: although subjects were allowed to engage in actively exploring the shapes through touch, they were far more limited in their ability to explore the shapes through vision.) Yet when it comes to an empirical test of Molyneux’s question, we are interested in whether the visual phenomenology of a three-dimensional object would be recognized as a reencounter with a shape that had already been met through touch. In addition, another way of explaining away the apparent negative result would be to hold that something else confused the issue. As we noted above, neither sense picks up on shape properties alone, so even when a subject is both seeing and touching the same shape, there will be phenomenal differences between the two experiences. And when we consider that Molyneux subjects are newly sighted, many of the other aspects of visual phenomenology will be new to them. Given this, it is at least possible that these features of visual phenomenology might distract the subject from the similarities between their new visual experiences and their past tactile ones. This means that one possible interpretation of Held et al.’s results is that the: Molyneux subjects’ visual and tactual representations of circles and squares are sufficiently similar to count as instances of the same type of perceptual experience (or as giving rise to the same type of perceptual concept), but that irrelevant phenomenal differences such as differences in color, shading, or edge detail prevent the subjects from recognizing, at least until further reflection, that this is so. (Levin 2018: 596) At this point, then Molyneux’s question remains a way of asking a question about whether or not our perceptual access to both common sensibles and public space has a modality-specific phenomenology or an amodal phenomenology. As far as the defender of the decomposition thesis is concerned, where two sense modalities access a single type of external property, each modality-specific experience of this property will be accompanied by a distinctive modality-specific phenomenal character, which could not accompany the experience of that property via a different modality. The defender of a more substantive form of multimodality must therefore suggest that this interpretation is mistaken. And although the phenomenal evidence concerning the “phenomenal switch” we overviewed in the previous section is powerful, particularly when coupled with empirical evidence of intermodal feature binding, it nevertheless remains possible for the committed skeptic to resist it, and hold that the phenomenology of these scenarios can be accounted for in terms of co-conscious, or at least rapidly alternating, modality-specific experiences, together with judgments that aspects of different elements of these experiences are grounded in common external objects.

236  Perception and the Sciences of the Mind

Cross-Modal Illusions In order to try and further undermine the skeptical interpretation of what is going on in intermodal feature binding, we might turn our attention to the closely related phenomenon of cross-modal illusions. These illusions occur when information from one sense modality affects the processing in a second sense modality in such a way that the outputs of the second modality become illusory. Consider the effect of listening to someone “throw their voice”, a phenomenon known as the ventriloquism effect (Bertelson 1999). This occurs when the ventriloquist – the voice thrower – speaks whilst moving the mouth of a dummy and keeping their own mouth and lips still, which leads the audience to experience the voice as emanating from the dummy. Even though the voice does not actually come from the dummy’s direction, it is nonetheless experienced as so doing. To the extent that the subsequent auditory experience mislocates the real source of the sound, we can call this a cross-modal illusion: an illusion in which information from one modality affects the outputs of a distinct modality in such a way that it leads to error. The McGurk effect (McGurk and MacDonald 1976) is another cross-modal auditory illusion, in which our perception of a subject’s mouth movements affects what we hear. To create this illusion, participants are shown a (silent) video of a speaker making the sound /ga/. If you try and make this sound yourself, you will see that it is produced with the back of the tongue on the roof of the mouth, and with the lips remaining open throughout. With this visual stimulus playing, a synchronized audio stimulus is played of a subject making the sound /ba/. Again, if you try this yourself, it is produced quite differently, primarily through the meeting and parting of the lips leading to the distinctive bilabial plosive burst of air. Intriguingly, however, subjects of this experiment hear the sound /da/. If you try to make this sound, you will find that, in terms of mouth movements, it is similar to the /ga/ sound: although the tongue’s role is different – with the tip going behind the teeth  – the lips remain open throughout. However, auditorily, /da/ is quite similar to /ba/ (because the burst of air involved in the alveolar [d] and [t] plosives is more similar to that involved in the bilabial [b] and [p] plosives, as they are all produced at the front of the mouth, than it is to the release of air at the back of the mouth involved in the velar [g] and [k] plosives). The accepted explanation of the McGurk effect is that information from vision – in particular that the seen lips remain open – conflicts with information from hearing (the nature of the plosive burst), and so our perceptual systems reconcile this by concluding that the sound heard must be /da/, which is the best “match” of both the visual and auditory inputs. Importantly, this is clearly a perceptual phenomenon, not merely a phenomenon of judgment: we hear a /da/ sound (try it yourself – there are many examples available on YouTube. If you do, try closing your eyes so that there is no longer any visual information to conflict with what is being heard: you will find the sound reverts back to /ba/ immediately). Although both the ventriloquism and McGurk effects involve visual information affecting what we hear, this is not only a one-way street. In the sound-induced flash illusion (Shams et al. 2002), a black disk is flashed on a screen in front of the

Multimodality 237

subject, accompanied by audio of one or two short beeps. If the flash is accompanied by a single beep, participants report a single flash; if it is accompanied by two beeps, two flashes are reported (again, examples can be found online). In another set of studies, Vroomen and de Gelder (2000) found that if subjects watching a rapidly changing visual display hear an abrupt sound, the sound appears to “pin” the visual display and cause it to freeze. In these experiments, information from hearing affects what we see. Does this case also involve reconciling a conflict between what we hear and what we see? That it can be described in this way is not as obvious as it was in the previous two cases, but if we remember that, outside of the psychologist’s laboratory, the sudden generation of audible events often corresponds with conspicuous visible events (such as the sight and the sound of a dropped object hitting the floor and bouncing), then the illusion could be explained as the perceptual system’s attempt to ensure that a somewhat ambiguous collection of sensory inputs accords to this everyday principle of audio-visual correlation. In addition to these experiments, there are a range of others that suggest that all of our different senses are either subject to or instigate cross-modal illusions. In the previous chapter, we saw examples in which the color of a drink, the color of the pot that a mousse was served in, and even the weight that a yoghurt was felt to be could all affect the way the substances tasted. In the parchment skin illusion (Jousmäki and Hari 1998), subjects rubbed their hands together, but rather than hearing the sound this produced directly, it was relayed through headphones. It was found that by altering features of the sounds that subjects heard, the way their hands felt could be altered – in particular, the louder the sound and/or the more prominent the high (>2 kHz) frequencies were, the smoother and drier that subjects felt their hands to be, with some describing their hands feeling almost like parchment paper. This suggests that information from audition can modify touch, such that a subject’s touch experience becomes more appropriate to the sounds that their touching activities make. There is also a touch version of the sound-induced flash illusion, in which a double beep causes subjects who are tapped once to feel that they have been tapped twice in quick succession (Hötting and Röder 2004). Zampini and Spence (2004) showed that potato chips that tasted stale to participants could be made to taste fresh if they were wearing headphones that relayed the sounds they were making, but modified to emphasize the “crunchiness”, suggesting that auditory information can modify taste. Dematté and colleagues (2006) asked blindfolded subjects to rate the texture of fabrics and found that a single fabric was rated to be softer when it smelled pleasant (of lavender), and rougher when it had an animal smell, and studies showing that coloring a white wine red can change the way even wine-tasters say it smells (Morrot et al. 2001). Even our bodily awareness can be involved in cross-modal illusions. For example, in the rubber hand illusion (Botvinick and Cohen 1998), subjects rested their left arm on a table to the side, which was screened from their view, a fake arm was placed in front of them, and paintbrushes were then used (synchronously) to gently stroke both the real and the fake arm. Not only did subjects report “feeling” the stroke on the rubber hand, when they were asked to identify where the index finger of their left hand was (using their right hand which was out of sight beneath the

238  Perception and the Sciences of the Mind

table), this was displaced toward the fake hand, suggesting that visual input resulted in a proprioceptive illusion of where their left hand was located. In a slightly different variant of the experiment (Petkova et al. 2012), participants were blindfolded, so never saw the rubber hand, and instead of simultaneously stroking both the rubber hand and the real hand with paintbrushes, the experimenters used the participant’s left index finger to stroke the rubber hand, while themselves simultaneously stroking the participant’s right hand. Once again, participants described feeling that they were stroking their own hand, and when asked to identify where the index finger of the stroked hand was, this was again displaced toward the fake hand. (Intriguingly, blind participants were not subject to this illusion; see de Vignemont (2014: 1006) for discussion). On the face of it, these examples suggest that information from one sensory modality can causally affect the phenomenal character of an experience in a distinct modality. What explains these cases? Well, as in the case of many of the traditional optical illusions, the orthodox interpretation is that these illusions involve perceptual processing biases that are advantageous under normal, everyday conditions being exploited in laboratory conditions. Think about the ventriloquism effect: normally, there is a strong correlation between the visible location of a moving mouth and the audible location of spoken sounds. Given this, it makes sense that a perceptual system whose job it is to determine the arrangement of an environment in which there is typically high levels of correlation between features that are available through vision and features that are available through hearing (or indeed, between any pair of modalities) should use clues from one modality to resolve ambiguities in another. The standard explanation is thus that “crossmodal illusions result from intermodal biases and recalibrations that are not mere accidents … [and] in fact help to improve the accuracy of our perceptual responses given information from multiple sensory sources” (O’Callaghan 2014: 100). If this is right, then we would expect these cross-modal interactions to be happening far more often than just in those situations that lead to illusion. For example, it is probable that whenever we hear someone speak, we will be using visual information about their mouth and lip movements to refine our auditory classification of the sounds they make; it is just that we don’t notice this unless it takes place in the kind of contrived experimental context that yields an illusion like the McGurk effect. Given this, even if cross-modal illusions are relatively rare, cross-modal interactions will be pervasive. Moreover, O’Callaghan argues that cross-modal illusions can form part of a response to the objection from Spence and Bayne we saw above, which suggested that the appearance of object unity could in fact be a purely cognitive, inferential effect (2014: 84). By focusing on illusions, he suggests, we can remove the potential influence of cognitive states such as belief and judgment. Just as we saw in Chapter 3, when we discussed the persistence of the Müller-Lyer illusion, the fact that an illusion endures even when we know that it is misleading gives us reason to think that the illusion is perceptual, not cognitive. Similarly, in the case of the ventriloquism effect, we know that the dummy is not speaking, but it still sounds as though that is where the voice is coming from. The McGurk effect is even more striking: even

Multimodality 239

when the situation is set up in such a way to emphasize its illusory aspects – say by displaying two faces side-by-side, one making /ba/ mouth movements and the other making /ga/ movements – the illusion is incredibly persistent. Even though you can tell it is the same audio in each case, it still sounds different depending upon which face you attend to. This is strong evidence that the phenomenal experiences in these cases are not grounded in cognition or inference. Cross-modal illusions thus hint at a form of multimodality that is both common and relatively significant: if this line of thinking is correct, then there will be extensive cross-modal interactions that serve to refine, enhance, or make precise our perception of features, leading to greater accuracy than could have been achieved had each sense modality been working independently. For this reason, Briscoe calls this phenomenon “optimizing multisensory integration” (2016: 123). It also gives us reason to think that our overall perceptual experience at any given time would be quite different were it not for the interactions between sense modalities. However, to the extent that these cross-modal interactions only serve to optimize our perception of features that can be perceived by sense modalities working independently, there is still a sense in which it remains possible to describe our total perceptual experience at a time along unimodal lines. Even with the kinds of cross-modal effects we have been discussing, our total perceptual experience could nonetheless still be an assemblage or composite of modality-specific experiences that can be characterized in ways that only make use of familiar modality-specific terms, it is just that the precise deliveries of each modality-specific experience is influenced by information derived from the other modalities (O’Callaghan 2008: 327–8). This is often characterized as a “purely causal” view of crossmodal interactions, as opposed to the kind of ‘constitutive’ view that “gives rise to experiences that are inherently multisensory” (Spence and Bayne 2014: 99).

Multimodality Reflection on cross-modal illusions therefore only appears to take us so far. They show that there do appear to be causal interactions between the different sense modalities, but we are not yet at a point where the skeptical interpretation – that the phenomenal characters of the experiences associated with the different modalities remain entirely unimodal – can be dismissed. When we discussed the case of taste in the previous chapter, however, we saw a way in which we might argue for distinctively multimodal phenomenal character. On one interpretation of the physiological evidence, flavor properties can be argued to be properties that can only be accessed by multiple senses – primarily gustation and olfaction – working together. If you lost your sense of smell, you would lose your ability to perceptually access flavors; all that would be left would be the five basic tastes alongside the experiences that result from trigeminal stimulation; similarly, if you lost the gustatory sense associated with the tongue, you would also be unable to perceive flavors. This can suggest that the phenomenology of flavor is a fundamentally multimodal phenomenology – phenomenal character that can only be enjoyed when multiple sense modalities work in concert (Briscoe 2016: 124–5).

240  Perception and the Sciences of the Mind

Other candidates for features that can only be perceived multimodally could be argued to include properties such as wetness and oiliness as “pressure plus cold can facilitate perception of wetness, whereas pressure plus warmth is often perceived as oiliness” (Ratcliffe 2012: 418). If thermal perception is a different perceptual modality to tactile perception, then perception of these properties would also seem to only be available to multimodal perception. If this is compelling, then perhaps haptic touch itself could also be a candidate inasmuch as it relies on the combination of cutaneous touch and kinesthesis/proprioception (Briscoe 2019). Briscoe calls this generative multisensory integration to distinguish it from the optimizing multisensory integration found in cross-modal illusions. Optimizing multisensory integration optimizes our perception of features that can nonetheless be perceived unimodally; generative multisensory integration, however, generates the ability to perceive novel features that could not be perceived by any sense modality working alone. As we saw in the previous chapter, however, Louise Richardson (2013b) disputes the claim that flavor perception is multisensory simply because it involves different physiological pathways. If we think of the different senses along broadly functional lines, then rather than taking the involvement of the olfactory epithelium in the perception of flavor to show that we perceive flavor, in part, by using the sense of smell, we could take it instead to show that the olfactory epithelium is part of the sub-personal underpinnings of two different senses. When it is stimulated by chemicals arriving orthonasally as a result of the activities of inhaling or sniffing, these receptors function as part of a sense of smell; when the stimuli arrive retronasally, as a result of the activities of chewing and swallowing, the same receptors can function as part of a different sense: the sense of taste. This approach also looks to be extendable to the touch-based cases too. As we saw in the previous chapter, Matthew Fulkerson (2013) argues that the array of underlying touch-related physiological systems – cutaneous touch and pressure perception, thermal perception, proprioception, and kinesthesis – can also all be unified as the sub-personal underpinnings of a single sense of touch when we consider it from a functional perspective. In both of these cases, then the perception of features that might initially appear to be a result of the generative multisensory integration of distinct senses working in concert may be able to be reconceptualized as features that are available to single senses, understood functionally. This might suggest that the question of whether or not there are any features that can only be perceived multimodally will not be able to adequately be addressed until we have agreed on how different senses are to be distinguished from one another. Another phenomenon that has been argued to require generative multisensory integration is the intermodal perception of meter – the recurring patterns in a piece of music. Now we might think that this would not count as generative, given that musical meter can already be accessed by hearing, and maybe also by cutaneous touch (patterns in taps on the skin) and vision (patterns in pulses of lights), but the key to understanding how this might nonetheless count as generative is to see meter as

Multimodality 241

a property of the audible music that we can access. Normally, we access meter by listening to a sequence of sounds; in touch, we can access the meter in a sequence of touches. What Huang et al. (2012) found, however, was that if we break a meter across a pattern of sounds and touches, so that neither the sounds themselves nor the taps themselves exemplify the relevant meter, subjects can still access the meter formed by the combination of sounds and taps. This appears to be generative, in that it is not a property that is possessed by the sounds alone nor by the taps alone, but a property of a pattern spread over sounds and taps. This is a property that cannot be perceived by any individual sense modality working alone, so perception of this property seems to be an example of intermodal perception giving us access to a novel property. A final case that has been argued to require generative multisensory integration is outlined by Briscoe, who argues that the perceptible feature of being at a certain location in egocentric space is a “strongly novel type of perceptible feature” – that is, a feature that can only be perceived multimodally (2019: 17). His argument begins by noting that, initially, each sense encodes information about the location of objects relative to its own sense-organ-based frame of reference: for example, vision encodes information retinocentrically, audition using a head-centered frame of reference, and touch on a hand-centered basis. As Driver and Spence point out, however, given that “the receptors for each modality can move freely relative to external objects … and can also move relative to each other …, a single modality alone cannot provide a stable representation of external space” (1998: 254). If this is correct, then no single sense working alone can locate an object at a particular point in external space. In order to bring these modality-specific spatial frames of reference together to enable objects to be located at a particular point in space (specified egocentrically), the perceptual system needs to engage in a process of coordinate transformation, which relies on information from bodily awareness: to move from a retina-centered to a head-centered frame of reference, you need proprioceptive information about where the eyes are pointing; to move from a head-centered frame of reference to a torso-centered frame of reference, you need proprioceptive and kinesthetic information about the location and movement of the head, shoulders and back, and so on. To head off the possibility of responding to this by arguing that we cannot reliably draw conclusions about personal-level phenomenal experience from data about sub-personal mechanisms, and that a compatible unimodal explanation of spatial phenomenology is available, Briscoe begins by noting the strong introspective evidence that the “locations of which we seem most directly and non-inferentially to be perceptually aware [are] locations in a single, crossmodally accessible system of places and distances” (2019: 19). To try and account for this in terms of a co-conscious awareness of objects in, say, a retinocentric frame of reference alongside awareness of the attitudes of the head and eyes, he argues, would not only seem phenomenologically spurious, it would also require a counterintuitive analysis of experiences of subject-relative motion, such as those we enjoy when visually tracking a moving object (2019: 22-3).

242  Perception and the Sciences of the Mind

Is a Multimodal Space Needed for Intermodal Feature Binding? As Briscoe notes, the ability to locate objects accessed through a single sense in a multimodal space is a precursor of the kind of intermodal feature binding we discussed above: when the perceptual system makes a decision about which streams of incoming information have their source in a single object, spatial congruence is probably the most important consideration. Yet the only way in which information from different modalities could be deemed to be spatially congruent would be if the different modalities were locating objects using a common spatial frame of reference. Given this, it looks as though any instances involving intermodal feature binding do turn out to involve generative multisensory integration after all.

Moreover, the claim that spatial phenomenology can be explained unimodally also struggles to accommodate the empirical findings of vibration-induced illusions of object motion. It is well known that vibrating the tendons of a limb can lead to an illusory sensation of movement; intriguingly, however, inducing movement illusions using vibration has also been found to induce associated visual illusions. For example, it was found that vibrating the muscles in a subject’s neck can result in both the bodily awareness illusion that the head is moving upwards and the visual illusion of a stationary target moving upwards (Roll et al. 1991). If the normal visual experience of a stationary target were a conjunction of a purely visual experience of an object located at a certain point relative to the eyes alongside a bodily awareness experience of the head and eyes being still, then we might expect the vibrations to induce an illusion of the body moving, but not the target. The fact that the disruption in bodily awareness also yields a visual illusion suggests that, in fact, the experience of visual location is actually a multimodal experience involving both visual and proprioceptive/kinesthetic mechanisms. However, it still seems possible that cases such as these could be analyzed in the way suggested above for haptic touch and flavor perception. In this context, it may be significant that bodily awareness is not one of the traditional five Aristotelian senses. In this light, consider that the analysis of unimodal flavor perception discussed above relied upon the possibility that a single physiological system could underpin more than one sense modality, understood functionally; consider also that the analysis of haptic touch suggested that proprioception and kinesthesis could actually form part of the sub-personal basis of touch. This suggests that it is at least possible to hold that, whether or not we want to consider bodily awareness to be a distinct sense in its own right, our proprioceptive and kinesthetic systems could also be a part of the sub-personal underpinnings of vision (and maybe even other senses too). If this is correct, then although bodily awareness would be intimately involved in the ability of sight to locate objects in egocentric space, this could nonetheless still be an achievement of vision alone.

Multimodality 243

Looking Forward The question of whether or not perception is fundamentally multimodal is thus still open and depends in no small part on the question with which we began the previous chapter of how the senses are distinguished from one another. When it comes to looking at how these discussions impact upon philosophical theorizing about perception, we might notice that the decomposition thesis is a natural ally of views that provide a negative answer to Molyneux’s question and maintain modality-specific theories of phenomenal character, such as qualia theories, and the kind of impure versions of representationalism that hold that contents are represented in different ways or manners that correspond to the different sense modalities. What about allies of the multimodal view that rejects the decomposition thesis? Well, if we are convinced by the foregoing considerations that there is distinctively multimodal phenomenology in perception, then at a minimum this would require rejecting the element of the decomposition thesis that claims that perceptual phenomenology is exhausted by the modality-specific experiences. It would thus allow for distinctive phenomenologies corresponding to each sensory modality, but would require an additional theory to accommodate the extra multimodal phenomenology. A more thoroughgoing response to the issues discussed in this chapter, however, might be to think that the heart of the decomposition thesis – that a person’s overall perceptual experience is just the sum of their modality-specific experiences – is mistaken. Perhaps, we might think, it’s time to move beyond perceptual modalities (Briscoe 2017: 9, see also Speaks 2015: 185) and answer Molyneux’s question in the affirmative. On such a view, we might think of the entire perceptual system as a single complex system that responds to an array of different stimuli and results in a unified multimodal perceptual experience of the world. Such a view would still have room for a distinction between senses – understood as either physiological systems or as functional ways of engaging with the world – but these distinctions between senses would not be reflected in distinctions between types of phenomenology (save in the minimal sense that if the preferred understanding of the senses has certain senses as having proper sensibles, then we might be able to talk about the phenomenal appearance of those proper sensibles as “modality-specific” phenomenology). To this extent then, we might think that this approach meshes well with austere versions of representationalism and naïve realism that attempt to ground perceptual phenomenology in the pure representation of, or acquaintance with, the external environment. We might also think that there are other reasons why empirical realist theories align well with multimodal approaches. For example, when we explored crossmodal illusions, we saw that from the perspective of perceptual processing, the occurrence of illusions is often explained on the basis of the perceptual system making use of short cuts – tacit assumptions about the way the world is – which can then be manipulated to lead the system into error. We might therefore ask what implicit assumptions would the perceptual system be having to make in order for cross-modal illusions to occur. Well, the answer would seem to be that in certain situations – in particular, when the information being accessed via different sense modalities comes from (roughly) the same place at the same time – the perceptual

244  Perception and the Sciences of the Mind

system operates on the assumption that this incoming information is information about the same thing. This assumption explains why the perceptual system attempts to integrate the multimodal information, and it is this assumption that explains why certain cross-modal reports are experienced as conflicts that are in need of resolution. If it weren’t because the perceptual system assumes that the lip movements and the audible sounds in the McGurk effect are being produced by the same object, then there would be no reason to think that these streams of information were in conflict, which in turn would mean that there was no need to resolve this conflict in the kind of way that results in a cross-modal illusion. As O’Callaghan emphasizes, reconciliation “requires a conflict in information, and genuine conflict requires disagreement. Disagreement requires a common subject matter. Even apparent disagreement requires the presumption of a common subject matter” (2014: 105). So multimodalism meshes well with the view that the role of the perceptual system, understood as a single complex system, is to give us information about objects in an external world. Altogether, these considerations suggest there is still much important work to be done in the philosophy of perception. Although the vast majority of philosophical work has focused on vision, there remains no widely accepted theory of sight. What is more, we have also seen that there are reasons to think that theories of the other senses may follow a different pattern. So even if a theory of vision does gain wide acceptance, there is no guarantee that adequate theories of hearing, touch, smell, and taste will follow the same structure. Finally, in this chapter, we have seen reason to question the assumption that the right way to think about perception is by breaking it down into distinct senses and dealing with them separately. Perhaps, instead, a thorough philosophical understanding of our capacity to perceive the world will require a united theory of all our senses. How all of this plays out remains to be seen.

Questions •

•

•

What is the relationship between sub-personal and personal-level phenomena when it comes to multimodality? Does empirical evidence of intermodal feature binding give us any reason to think that phenomenal perceptual experience will be multimodal? Could there be an empirical resolution of Molyneux’s question? Does the phenomenal evidence appealed to in cases of cross-modal illusion show us that perception is multimodal, at least in part, or is there a plausible skeptical alternative? Is perception fundamentally multimodal? What impact does your answer to this question have on your thinking about philosophical theories of perception more generally?

Multimodality 245

Further Reading A great deal of work outlining and defending multimodality has been done by Casey O’Callaghan. His papers “Seeing What You Hear: Cross Modal Illusions and Perception”, “Intermodal Feature Binding Awareness”, and “The Multisensory Character of Perception” are good starting points, and his 2019 book A Multisensory Philosophy of Perception brings a lot of these important discussions together. Other useful introductions include Tim Bayne’s “The Multisensory Nature of Perceptual Consciousness”, which introduces the decomposition thesis, and Fiona Macpherson’s paper “Cross-Modal Experiences”. Charles Spence and Tim Bayne’s 2014 article “Is Consciousness Multisensory” is a good skeptical counterpoint to O’Callaghan. This paper can be found in the Stokes, Matthen, and Biggs’ collection mentioned in the previous chapter. Parts I and II of Robert Briscoe’s “Multisensory Processing and Perceptual Consciousness” also contain good overviews of both the issues and a range of the relevant empirical evidence. Briscoe’s 2019 paper “Bodily awareness and novel multisensory features” contains the important argument that location in egocentric space is a feature that can only be accessed multimodally.

References

Akins, K. and M. Hahn (2015) “Colour Perception” in M. Matthen (ed.) The Oxford Handbook of Philosophy of Perception (Oxford: Oxford University Press), 422–40. Ali, R. (2018) “Does Hallucinating Involve Perceiving?” Philosophical Studies 175, 601–27. Allen, K. (2013) “Blur” Philosophical Studies 162, 257–73. Allen, K. (2015) “Hallucination and Imagination” The Australasian Journal of Philosophy 93:2, 287–302. Allen, K. (2016) A Naïve Realist Theory of Colour (Oxford: Oxford University Press). Alston, W.P. (1999) “Back to the Theory of Appearing” Philosophical Perspectives 13, 181–203. Alter, T. (2006) “Does Synesthesia Undermine Representationalism?” Psyche 12:5, 1–11. Antony, L. (2011) “The Openness of Illusions” Philosophical Issues 21, 25–44. Aristotle (1984) “De Anima” reprinted in J. Barnes (ed.) The Complete Works of Aristotle, vol. 1, (Princeton: Princeton University Press), 641–92. Armstrong, D.M. (1961) Perception and the Physical World (London: Routledge and Kegan Paul). Armstrong, D.M. (1968) A Materialist Theory of the Mind (London: Routledge and Kegan Paul). Armstrong, D.M. (1973) Belief, Truth and Knowledge (Cambridge: Cambridge University Press). Austin, J.L. (1962) Sense and Sensibilia (Oxford: Clarendon Press). Auvray, M. and O. Deroy (2015) “How Do Synaesthetes Experience the World?” in M. Matthen (ed.) The Oxford Handbook of Philosophy of Perception (Oxford: Oxford University Press), 640–58. Auvray, M. and C. Spence (2008) “The Multisensory Perception of Flavour” Consciousness and Cognition 17, 1016–31. Bach-y-Rita, P., C.C. Collins, F. Saunders, B. White, and L. Scadden (1969) “Vision Substitution by Tactile Image Projection” Nature 221:5184, 963–4. Balcetis, E. and D. Dunning (2010) “Wishful Seeing: More Desired Objects Are Seen as Closer” Psychological Science 21, 147–52. Banerjee, P., P. Chatterjee, and J. Sinha (2012) “Is It Light or Dark? Recalling Moral Behaviour Changes Perception of Brightness” Psychological Science 23, 407–9. Barnes, W. (1965) “The Myth of Sense-Data” in R.J. Swartz (ed.) Perceiving, Sensing and Knowing (Berkeley: University of California Press), 138–67. Baron-Cohen, S., M.A. Wyke, and C. Binnie (1987) “Hearing Words and Seeing Colours: An Experimental Investigation of a Case of Synaesthesia” Perception 16:6, 761–7. Batty, C. (2010) “A Representational Account of Olfactory Experience” Canadian Journal of Philosophy 40:4, 511–38. Batty, C. (2011) “Smelling Lessons” Philosophical Studies 153, 161–74. Batty, C. (2014) “The Illusion Confusion” Frontiers in Psychology 5, 1–11.

248  References Bayne, T. (2009) “Perception and the Reach of Phenomenal Content” The Philosophical Quarterly 59, 385–404. Bayne, T. (2010) The Unity of Consciousness (Oxford: Oxford University Press). Bayne, T. (2014) “The Multisensory Nature of Perceptual Consciousness” in D.J. Bennett and C.S. Hill (eds.) Sensory Integration and the Unity of Consciousness (Cambridge, MA: The MIT Press), 15–30. Beck, O. (2019) “Rethinking Naïve Realism” Philosophical Studies 176, 607–33. Bennett, J. (1971) Locke, Berkeley, Hume: Central Themes (New York: Oxford University Press). Berkeley, G. (1709/1910) A New Theory of Vision and Other Writings (London: J.M. Dent and Sons Ltd.). Berkeley, G. (1734/1982) A Treatise Concerning the Principles of Human Knowledge (Indianapolis: Hackett). Bermúdez, J-L. (1995) “Nonconceptual Content: From Perceptual Experience to Subpersonal Computational States” Mind & Language 10:4, 333–69. Bertelson, P. (1999) “Ventriloquism: A Case of Crossmodal Perceptual Grouping” Advances in Psychology 129, 347–62. Bhalla, M. and D.R. Proffitt (1999) “Visual-motor recalibration in geographical slant perception” Journal of Experimental Psychology: Human Perception and Performance 25, 1076–96. Billock, V.A., G.A. Gleason, and B.H. Tsou (2001) “Perception of Forbidden Colors in Retinally Stabilized Equiluminant Images: An Indication of Soft-Wired Cortical Color Opponency? Journal of the Optical Society of America 18, 2398–403. Blake, R., T.J. Palmeri, R. Marois, and C-Y. Kim (2005) “On the Perceptual Reality of Synaesthetic Color” in L.C. Robertson and N. Sagiv (eds.) Synaethesia: Perspectives from Cognitive Neuroscience (New York: Oxford University Press), 47–73. Block, N. (1990) “Inverted Earth” in J.E. Tomberlin (ed.) Philosophical Perspectives Volume 4: Action Theory and Philosophy of Mind (Atascadero: Ridgeview), 53–79. Block, N. (1996) “Mental Paint and Mental Latex” Philosophical Issues 7, 19–49. Block, N. (1999) “Sexism, Racism, Ageism and the Nature of Consciousness” Philosophical Topics 26:1–2, 39–70. Block, N. (2007) “Consciousness, Accessibility and the Mesh between Psychology and Neuroscience” Behavioral and Brain Sciences 30, 481–548. Block, N. (2008) “Consciousness and Cognitive Access” Proceedings of the Aristotelian Society 108, 289–317. Block, N. (2010) “Attention and Mental Paint” Philosophical Issues 20, 23–63. Boghossian, P.A. and J.D. Velleman (1989) “Colour as a Secondary Quality” Mind 98, 81–103. Botvinick, M. and J. Cohen (1998) “Rubber Hands ‘Feel’ Touch that Eyes See” Nature 391, 756. Bourget, D. (2010) “Consciousness Is Underived Intentionality” Noûs 44:1, 32–58. Bourget, D. (2017) “Why Are Some Phenomenal Experiences ‘Vivid’ and Others ‘Faint’? Representationalism, Imagery, and Cognitive Phenomenology” Australasian Journal of Philosophy 95:4, 673–87. Brentano, F. (1874/1995) Psychology from an Empirical Standpoint (London: Routledge). Brewer, B. (2005) “Perceptual Experience has Conceptual Content” in E. Sosa and M. Steup (eds.) Contemporary Debates in Epistemology (Oxford: Blackwell), 217–30. Brewer, B. (2006) “Perception and Content” European Journal of Philosophy 14, 165–81. Brewer, B. (2008) “How to Account for Illusion” in A. Haddock and F. Macpherson (eds.) Disjunctivism: Perception, Action, Knowledge (Oxford: Oxford University Press), 168–80. Brewer, B. (2011) Perception & Its Objects (Oxford: Oxford University Press). Brewer, B. (2013) “Attention and Direct Realism” Analytic Philosophy 54:4, 421–35.

References 249 Brewer, B. (2017) “The Object View of Perception” Topoi 36:2, 215–27. Briscoe, R. (2016) “Multisensory Processing and Perceptual Consciousness: Part I” Philosophy Compass 11:2, 121–33. Briscoe, R. (2017) “Multisensory Processing and Perceptual Consciousness: Part II” Philosophy Compass 12, e12423. Briscoe, R. (2019) “Bodily Awareness and Novel Multisensory Features” Synthese. https://doi. org/10.1007/s11229-019-02156-2 Broakes, J. (2011) “Where Do the Unique Hues Come From?” Review of the Philosophy of Psychology 2, 601–28. Brogaard, B. (2012) “Colour Eliminativism or Colour Relativism?” Philosophical Papers 41:2, 305–21. Brooks, R.A. (1991) “Intelligence without Representation” Artificial Intelligence 47, 139–59. Bruner, J.S. and C.C. Goodman (1947) “Value and Need as Organizing Factors in Perception” Journal of Abnormal and Social Psychology 42, 33–44. Buccella, A. (2017) “Naturalizing Qualia” Phenomenology and Mind 12, 152–61. Budek, T. and K. Farkas (2014) “Which Causes of an Experience are Objects of the Experience?” in B. Brogaard (ed.) Does Perception Have Content? (Oxford: Oxford University Press), 351–69. Burge, T. (1979) “Individualism and the Mental” reprinted in T. Burge ed.) Foundations of Mind (Oxford: Clarendon Press), 100–50. Burge, T. (1986) “Cartesian Error and the Objectivity of Perception” reprinted in T. Burge (ed.) Foundations of Mind (Oxford: Clarendon Press), 192–207. Burge, T. (1991) “Vision and Intentional Content” in E. LePore and R. vanGulick (eds.) John Searle and His Critics (Oxford: Basil Blackwell), 195–214. Burge, T. (2005) “Disjunctivism and Perceptual Psychology” Philosophical Topics, 33:1, 1–78. Burge, T. (2011) “Disjunctivism Again” Philosophical Explorations, 14:1, 43–80. Butchvarov, P. (1980) “Adverbial Theories of Consciousness” Midwest Studies in Philosophy 5, 261–80. Butterfill, S. (2009) “Seeing Causings and Hearing Gestures” The Philosophical Quarterly 59, 405–28. Byrne, A. (2001) “Intentionalism Defended” The Philosophical Review 110, 199–240. Byrne, A. (2002) “Don’t Panic: Tye’s Intentionalist Theory of Consciousness” in A Field Guide to the Philosophy of Mind Symposium on Tye’s Consciousness, Color, and Content http:// host.uniroma3.it/progetti/kant/field/tyesymp_byrne.htm. Byrne, A. (2009) “Experience and Content” The Philosophical Quarterly 59, 429–51. Byrne, A. and D. Hilbert (2003) “Color Realism and Color Science” Behavioral and Brain Sciences 26, 3–21. Byrne, A. and D. Hilbert (2007) “Color Primitivism” Erkenntnis 66, 73–105. Byrne, A. and H. Logue (2008) “Either/Or” in A. Haddock and F. Macpherson (eds.) Disjunctivism: Perception, Action, Knowledge (Oxford: Oxford University Press), 57–94. Byrne, A. and H. Logue (2009) Disjunctivism: Contemporary Readings (Cambridge, MA: The MIT Press). Campbell, J. (1993) “A Simple View of Colour” in J. Haldane and C. Wright (eds.) Reality, Representation and Projection (Oxford: Oxford University Press), 257–68. Campbell, J. (1996) “Molyneux’s Question” Philosophical Issues 7, 301–18. Campbell, J. (2002) Reference and Consciousness (Oxford: Clarendon Press). Campbell, J. (2010) “Demonstrative Reference, the Relational View of Experience, and the Proximality Principle” in R. Jeshion (ed.) New Essays on Singular Thought (Oxford: Oxford University Press), 193–212.

250  References Campbell, J. (2014) “A Straightforward Solution to Berkeley’s Puzzle” in J. Campbell and Q. Cassam (eds.) Berkeley’s Puzzle: What Does Experience Teach Us? (Oxford: Oxford University Press), 25–49. Campbell, J. (2020) “Does that Which Makes the Sensation of Blue a Mental Fact Escape Us?” in D.H. Brown and F. Macpherson (eds.) The Routledge Handbook of Philosophy of Colour (London and New York: Routledge), 408–20. Carrasco, M., S. Ling, and S. Read (2004) “Attention Alters Appearance” Nature Neuroscience 7, 308–13. Casati, R., E. Di Bona, and J. Dokic (2013) “The Ockhamization of the Event Sources of Sound” Analysis 73:3, 462–66. Casati, R. and J. Dokic (2005) “Sounds” in E. Zalta (ed.) The Stanford Encyclopedia of Philosophy https://plato.stanford.edu/entries/sounds/ Cassam, Q. (2014) “Berkeley’s Puzzle” in J. Campbell and Q. Cassam (eds.) Berkeley’s Puzzle: What Does Experience Teach Us? (Oxford: Oxford University Press), 99–117. Cavedon-Taylor, D. (2018) “Odors, Objects, and Olfaction” American Philosophical Quarterly 55:1, 81–94. Chalmers, D.J. (1995) “Facing Up to the Problem of Consciousness” in J. Shear (ed.) Explaining Consciousness: The Hard Problem (Cambridge, MA: The MIT Press), 9–32. Chalmers, D.J. (1996) The Conscious Mind (New York: Oxford University Press). Chalmers, D.J. (1997) “Moving Forward on the Problem of Consciousness” Journal of Consciousness Studies 4:1, 3–46. Chalmers, D.J. (2004) “The Representational Character of Experience” in B. Leiter (ed.) The Future of Philosophy (Oxford: Oxford University Press), 153–81. Chalmers, D.J. (2006) “Perception and the Fall from Eden” in T.S. Gendler and J. Hawthorne (eds.) Perceptual Experience (Oxford: Oxford University Press), 49–125. Chemero, A. (2009) Radical Embodied Cognitive Science (Cambridge, MA: The MIT Press). Child, B. (1994) Causality, Interpretation and the Mind (Oxford: Oxford University Press). Chirimuuta, M. (2015) Outside Color: Perceptual Science and the Puzzle of Color in Philosophy (Cambridge, MA: The MIT Press). Chisholm, R.M. (1942) “The Problem of the Speckled Hen” Mind 51, 368–73. Chomsky, N. (1959) “A Review of B.F. Skinner’s Verbal Behavior” Language 35:1, 26–58. Clark, A. (1993) Sensory Qualities (Oxford: Clarendon Press). Clark, A. (2012) “Dreaming the Whole Cat: Generative Models, Predictive Processing, and the Enactivist Conception of Perceptual Experience” Mind 121, 753–71. Clark, A. (2013) “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science” Behavioral and Brain Sciences 36:3, 181–204. Clark, A. (2017) “Predictions, Precision, and Agentive Attention” Consciousness and Cognition 56, 115–19. Coates, P. (2000) “Deviant Causal Chains and Hallucinations: A Problem for the AntiCausalist” Philosophical Quarterly 50, 320–31. Coates, P. (2007) The Metaphysics of Perception: Wilfrid Sellars, Perceptual Consciousness and Critical Realism (New York: Routledge). Cohen, J. (2009) The Red and the Real: An Essay on Color Ontology (Oxford: Oxford University Press). Cohen, J. (2015) “Ecumenicism, Comparability, and Color, Or: How to Have Your Cake and Eat It, Too” Minds and Machines 25, 149–75. Cornman, J.W. (1971) Materialism and Sensations (New Haven: Yale University Press). Crane, T. (1992) “The Nonconceptual Content of Experience” in T. Crane (ed.) The Contents of Experience: Essays on Perception (Cambridge: Cambridge University Press), 136–57.

References 251 Crane, T. (1998) “Intentionality as the Mark of the Mental” in A. O’Hear (ed.) Contemporary Issues in the Philosophy of Mind (Cambridge: Cambridge University Press), 229–52. Crane, T. (2001) “The Origins of Qualia” in T. Crane and S. Patterson (eds.) The History of the Mind-Body Problem (London: Routledge), 169–94. Crane, T. (2006) “Is There a Perceptual Relation?” in T. Gendler and J. Hawthorne (eds.) Perceptual Experience (Oxford: Oxford University Press), 126–46. Crane, T. (2009) “Is Perception a Propositional Attitude?” The Philosophical Quarterly 59, 452–69. Crane, H.D. and T.P. Piantanida (1983) “On Seeing Reddish Green and Yellowish Blue” Science 221, 1078–80. Cytowic, R.E. (1993) The Man Who Tasted Shapes (London: Abacus). D’Ambrosio, J. (2019) “A New Perceptual Adverbialism” The Journal of Philosophy 116:8, 413–46. Dancy, J. (1985) Introduction to Contemporary Epistemology (Oxford: Blackwell). Davidson, D. (1980) “The Logical Form of Action Sentences” in his Essays on Actions and Events (Oxford: Clarendon Press). Davies, M. (1991) “Individualism and Perceptual Content” Mind 100, 461–84. Davies, A.A. (2004) “Disorders of Spatial Orientation and Awareness: Unilateral Neglect” in J. Ponsford (ed.) Science and Practice of Neuropsychology. Cognitive and Behavioral Rehabilitation: From Neurobiology to Clinical Practice (New York: Guilford Press), 175–223. Delk, J.L. and S. Fillenbaum (1965) “Differences in Perceived Color as a Function of Characteristic Color” The American Journal of Psychology 78:2, 290–93. De Gelder, B., J.S. Morris, and R.J. Dolan (2005) “Unconscious Fear Influences Emotional Awareness of Faces and Voices” Proceedings of the National Academy of Sciences of the United States of America 102:51, 18682–7. De Gelder, B., M. Tamietto, G. van Boxtel, R. Goebel, A. Sahraie, J. van den Stock, B.M. Stienen, L. Weiskrantz, and A. Pegna (2008) “Intact Navigation Skills after Bilateral Loss of Striate Cortex” Current Biology 18:24, R1128–9. Dematté, M.L., D. Sanabria, R. Sugarman, and C. Spence (2006) “Cross-Modal Associations between Olfaction and Touch” Chemical Senses, 31, 291–300. Dennett, D.C. (1969) Content and Consciousness (London: Routledge and Kegan Paul). Dennett, D.C. (1988) “Quining Qualia” in A.J. Marcel and E. Bisiach (eds.), Consciousness in Contemporary Science (Oxford: Oxford University Press), 42–77. Dennett, D.C. (1991) Consciousness Explained (London: Penguin). Deroy, O. (2014a) “The Unity Assumption and the Many Unities of Consciousness” in D.J. Bennett and C.S. Hill (eds.) Sensory Integration and the Unity of Consciousness (Cambridge, MA: The MIT Press), 105–24. Deroy, O. (2014b) “Synesthesia: An Experience of the Third Kind?” in R. Brown (ed.) Consciousness Inside and Out: Phenomenology, Neuroscience, and the Nature of Experience (Dordrecht: Springer), 395–407. Deroy, O. and M. Auvray (2014) “A Crossmodal Perspective on Sensory Substitution” in D. Stokes, M. Matthen and S. Biggs (eds.) Perception and Its Modalities (Oxford: Oxford University Press), 327–49. Descartes, R. (1637/2001) “Optics” in P.J. Olscamp (trans.) Discourse on Method, Optics, Geometry and Meteorology (Indianapolis: Hackett). Descartes, R. (1644/1985) “Principles of Philosophy” in J. Cottingham, R. Stoothoff, and D. Murdoch (trans.) The Philosophical Writings of Descartes: Volume I (Cambridge: Cambridge University Press). De Vignemont, F. (2014) “A Multimodal Conception of Bodily Awareness” Mind 123, 989–1020.

252  References De Vignemont, F. and O. Massin (2015) “Touch” in M. Matthen (ed.) The Oxford Handbook of Philosophy of Perception (Oxford: Oxford University Press), 294–313. Dinges, A. (2015) “The Many-Relations Problem for Adverbialism” Analysis 75:2, 231–7. Dixon, M.J., D. Smilek, and P.M. Merikle (2004) “Not All Synaesthetes Are Created Equal: Projector versus Associator Synaesthetes” Cognitive, Affective, and Behavioral Neuroscience 4, 335–43. Dokic, J. and J-R. Martin (2012) “Disjunctivism, Hallucinations, and Metacognition” WIREs Cognitive Science 3, 533–43. Drayson, Z. (2018) “Direct Perception and the Predictive Mind” Philosophical Studies 175, 3145–64. Dretske, F. (1995) Naturalizing the Mind (Cambridge, MA: The MIT Press). Dretske, F. (2003) “Experience as Representation” in E. Villanueva (ed.) Philosophical Issues 13: Philosophy of Mind (Atascadero: Ridgeview), 67–82. Driver, J. and C. Spence (1998) “Attention and the Crossmodal Construction of Space” Trends in Cognitive Sciences 2:7, 254–62. Ducasse, C.J. (1942) “Moore’s Refutation of Idealism” in P.A. Schlipp (ed.) The Philosophy of G.E. Moore (first volume) (New York: Tudor Publishing Company), 225–51. Durie, B. (2005) “Future Sense: Doors of Perception” New Scientist 185:2484, 33–6. Eagleman, D.M., A.D. Kagan, S.S. Nelson, D. Sagaram, and A.K. Sarma (2007) “A Standardized Test Battery for the Study of Synesthesia” Journal of Neuroscientific Methods 159:1, 139–45. Elias, L.J., D.M. Saucier, C. Hardie, and G.E. Sartry (2003) “Dissociating Semantic and Perceptual Components of Synaesthesia: Behavioural and Functional Neuroanatomical Investigations” Cognitive Brain Research 16, 232–7. Evans, G. (1982) The Varieties of Reference (Oxford: Oxford University Press). Evans, G. (1985) “Molyneux’s Question” in J. McDowell (ed.) Collected Papers: Gareth Evans (Oxford: Clarendon Press), 344–99. Farkas, K. (2008) “Phenomenal Intentionality without Compromise” The Monist 91:2, 273–93. Firestone, C. and B. Scholl (2016) “Cognition Does Not Affect Perception” Behavioral and Brain Sciences, 6, 1–19. Firth, R. (1949/1965) “Sense-Data and the Percept Theory” reprinted in R.J. Swartz (ed.) Perceiving, Sensing and Knowing (Berkeley: University of California Press), 204–70. Fish, W. (2009) Perception, Hallucination, and Illusion (New York: Oxford University Press). Fish, W. (2013) “Perception, Hallucination, and Illusion: Reply to My Critics” Philosophical Studies 163, 57–66. Fish, W. (ed.) (2016) Perception: Critical Concepts in Philosophy (New York: Routledge). Fodor, J. (1992) “A Theory of Content II: The Theory” in his A Theory of Content and Other Essays (Cambridge, MA: The MIT Press). Fodor, J. (2000) The Mind Doesn’t Work That Way: The Scope and Limits of Computational Psychology (Cambridge, MA: The MIT Press). Foster, J. (2000) The Nature of Perception (Oxford: Oxford University Press). Fodor, J. and Z. Pylyshyn (1981) “How Direct Is Visual Perception? Some Reflections on Gibson’s ‘Ecological Approach’” Cognition 9, 139–96. French, C. (2014) “Naïve Realist Perspectives on Seeing Blurrily” Ratio 27, 393–413. Fulkerson, M. (2011) “The Unity of Haptic Touch” Philosophical Psychology 24:4, 493–516. Fulkerson, M. (2013) The First Sense: A Philosophical Study of Human Touch (Cambridge, MA: The MIT Press). Fumerton, R. (2006) “Brewer, Direct Realism, and Acquaintance with Acquaintance” Philosophy and Phenomenological Research 63:2, 417–22.

References 253 Gallace, A. and C. Spence (2014) In Touch with the Future: The Sense of Touch from Cognitive Neuroscience to Virtual Reality (Oxford: Oxford University Press). Gallagher, S. (2005) How the Body Shapes the Mind (Oxford: Oxford University Press). Gantman, A.P and J.J. Van Bavel (2014) “The Moral Pop-Out Effect: Enhanced Perceptual Awareness of Morally Relevant Stimuli” Cognition 132, 22–9. Gazzaniga, M.S. (1998) The Mind’s Past (Berkeley: University of California Press). Genone, J. (2014) “Appearance and Illusion” Mind 123, 339–76. Genone, J. (2016) “Recent Work on Naïve Realism” American Philosophical Quarterly 53:1, 1–25. Gert, J. (2017) Primitive Colors: A Case Study in Neo-Pragmatist Metaphysics and Philosophy of Perception (New York: Oxford University Press). Gibson, J.J. (1966) The Senses Considered as Perceptual Systems (Boston: Houghton Mifflin). Gibson, J.J. (1972) “A Theory of Direct Visual Perception” in J. Royce and W. Rozenboom (eds.) The Psychology of Knowing (New York: Gordon & Breach), 77–89. Gibson, J.J. (1979) An Ecological Approach to Visual Perception (Boston: Houghton Mifflin). Gilbert, C.D. and W. Li (2013) “Top-Down Influences on Visual Processing” Nature Reviews Neuroscience 14:5, 350–63. Glüer, K. (2009) “In Defence of a Doxastic Account of Experience” Mind & Language 24:3, 297–327. Gray, R. (2012) “What Do Our Experiences of Heat and Cold Represent?” Philosophical Studies 166, 131–51. Grice, H.P. (1961) “The Causal Theory of Perception” Proceedings of the Aristotelian Society 35, 121–68. Grice, H.P. (1962) “Some Remarks about the Senses” in R.J. Butler (ed.) Analytical Philosophy (Oxford: Basil Blackwell), 133–53. Grossenbacher, P.G. and C.T. Lovelace (2001) “Mechanisms of Synaesthesia: Cognitive and Physiological Constraints” Trends in Cognitive Science 5, 36–41. Grush, R. (2007) “A Plug for Generic Phenomenology” Behavioral and Brain Sciences 30, 504–5. Grzankowski, A. (2018) “The Determinable-Determinate Relation Can’t Save Adverbialism” Analysis 78:1, 45–52. Haffenden, A.M. and M.A. Goodale (1998) “The Effect of Pictorial Illusion on Prehension and Perception” Journal of Cognitive Neuroscience 10, 122–36. Hardcastle, V.G. (1996) “The Why of Consciousness: A Non-Issue for Materialists” in J. Shear (ed.) Explaining Consciousness: The Hard Problem (Cambridge, MA: The MIT Press), 61–8. Hardin, C.L. (1988) Color for Philosophers: Unweaving the Rainbow (Indianapolis: Hackett Publishing Co.). Hardin, C.L. (2003) “A Spectral Reflectance Doth Not a Color Make” The Journal of Philosophy 100:4, 191–202. Hardin, C.L. (2014) “More Color Science for Philosophers” in D. Stokes, M. Matthen, and S. Biggs (eds.) Perception and Its Modalities (Oxford: Oxford University Press), 379–90. Harman, G. (1987) “(Non-solipsistic) Conceptual Role Semantics” in E. LePore (ed.) New Directions in Semantics (London: Academic Press). Harman, G. (1990) “The Intrinsic Quality of Experience” in J. Tomberlin (ed.) Philosophical Perspectives 4: Action Theory and the Philosophy of Mind (Atascadero: Ridgeview Publishing), 31–52. Harrison, J.E. and S. Baron-Cohen (1997) “Synaesthesia: An Introduction” in S. BaronCohen and J.E. Harrison (eds.) Synaesthesia: Classic and Contemporary Readings (Oxford: Blackwell), 3–16.

254  References Heck, R.G. (2000) “Nonconceptual Content and the “Space of Reasons”” The Philosophical Review 109, 483–523. Held, R., Y. Ostrovsky, B. de Gelder, T. Gandhi, S. Ganesh, U. Mathur, and P. Sinha (2011) “The Newly Sighted Fail to Match Seen with Felt” Nature Neuroscience 14:5, 551–3. Hellie, B. (2007) “Factive Phenomenal Characters” Philosophical Perspectives 21, 259–306. Hellie, B. (2013) “The Multidisjunctive Conception of Hallucination” in F. Macpherson and D. Platchais (eds.) Hallucination: Philosophy and Psychology (Cambridge, MA: The MIT Press), 149–74. Hering, E. (1964) Outlines of a Theory of the Light Sense (Cambridge, MA: The MIT Press). Hilbert, D.R. (1987) Color and Color Perception: A Study in Anthropocentric Realism (Stanford: CSLI). Hill, C. (2014) Meaning, Mind, and Knowledge (Oxford: Oxford University Press). Hinton, J.M. (1967) “Visual Experiences” Mind 76, 217–27. Hinton, J.M. (1973) Experiences: An Inquiry into Some Ambiguities (Oxford: Clarendon Press). Hobson, K. (2013) “In Defense of Relational Direct Realism” European Journal of Philosophy 21:4, 550–74. Hohwy, J. (2007) “Functional Integration and the Mind” Synthese 159:3, 315–28. Hohwy, J. (2013) The Predictive Mind (Oxford: Oxford University Press). Hohwy, J. (2020) “New Directions in Predictive Processing” Mind & Language 35:2, 209–23. Horgan, T.E. and J.L. Tienson (2002) “The Intentionality of Phenomenology and the Phenomenology of Intentionality” in D.J. Chalmers (ed.) Philosophy of Mind: Classical and Contemporary Readings (Oxford: Oxford University Press), 520–33. Horgan, T.E., J.L. Tienson, and G. Graham (2004) “Phenomenal Intentionality and the Brain in a Vat” in R. Schantz (ed.) The Externalist Challenge: New Studies on Cognition and Intentionality (Amsterdam: de Gruyter), 297–318. Hötting, K. and B. Röder (2004) “Hearing Cheats Touch but Less in the Congenitally Blind Than in Sighted Individuals” Psychological Science 15:1, 60–4. Howard, J.D., J. Plailly, M. Grueschow, J.D. Haynes, and J.A. Gottfried (2009) “Odor Quality Coding and Categorization in Human Posterior Piriform Cortex” Nature Neuroscience 12, 932–9. Huang, J., D. Gamble, K. Sarnlertsophon, X. Wang, and S. Hsiao (2012) “Feeling Music: Integration of Auditory and Tactile Inputs in Musical Meter” PLoS ONE 7, e48496. Hume, D. (1748/2011) “An Enquiry Concerning Human Understanding” in S. Buckle (ed.) Hume: An Enquiry Concerning Human Understanding: And Other Writings (Cambridge: Cambridge University Press), 1–144. Hurvich, L.M. and D. Jameson (1957) “An Opponent-Process Theory of Color Vision” Psychological Review 64, 384–403. Iriki, A., M. Tanaka, and Y. Iwamura (1996) “Coding of Modified Body Schema during Tool Use by Macaque Postcentral Neurones” Neuroreport 7:14, 2325–30. Jackson, F. (1975) “On the Adverbial Analysis of Visual Experience” Metaphilosophy 6:2, 127–35. Jackson, F. (1977) Perception: A Representative Theory (New York: Cambridge University Press). James, W. (1890) The Principles of Psychology (New York: Henry Holt). Johnston, M. (1992) “How to Speak of the Colors” Philosophical Studies 68:3, 221–63. Johnston, M. (2004) “The Obscure Object of Hallucination” Philosophical Studies 120, 113–83. Johnston, M. (2006) “Better that Mere Knowledge? The Function of Sensory Awareness” in T.S. Gendler and J. Hawthorne (eds.) Perceptual Experience (Oxford: Oxford University Press), 260–90.

References 255 Jones, L.A. and S.J. Lederman (2006) Human Hand Function (Oxford: Oxford University Press). Jousmäki, V. and R. Hari (1998) “Parchment-Skin Illusion: Sound-Biased Touch” Current Biology 8:6, R190–1. Kalderon, M.E. (2007) “Color Pluralism” The Philosophical Review 116:4, 563–601. Kalderon, M.E. (2011a) “Color Illusion” Noûs 45:4, 751–75. Kalderon, M.E. (2011b) “The Multiply Qualitative” Mind 120, 239–62. Kalderon, M.E. (2019) “Sound and Image” in C. Limbeck-Lilienau and F. Stadler (eds.) The Philosophy of Perception (Amsterdam: de Gruyter), 189–96. Keeley, B.L. (2002) “Making Sense of the Senses: Individuating Modalities in Humans and Other Animals” The Journal of Philosophy 99:1, 5–28. Keeley, B.L. (2013) “What Exactly Is a Sense?” in J. Simner and E. Hubbard (eds.) Oxford Handbook of Synethesia (New York: Oxford University Press), 941–58. Kind, A. (2008) “How to Believe in Qualia” in E. Wright (ed.) The Case for Qualia (Cambridge, MA: The MIT Press), 285–98. Kon, M. and K. Miller (2015) “Temporal Experience: Models, Methodology and Empirical Evidence” Topoi 34:1, 201–16. Kriegel, U. (2002) “PANIC Theory and the Prospects for a Representational Theory of Phenomenal Consciousness” Philosophical Psychology 15, 55–64. Kriegel, U. (2007) “Intentional Inexistence and Phenomenal Intentionality” Philosophical Perspectives 21, 307–40. Kriegel, U. (2008) “The Dispensibility of (Merely) Intentional Objects” Philosophical Studies 141, 79–95. Kriegel, U. (2011a) The Sources of Intentionality (New York: Oxford University Press). Kriegel, U. (2011b) “The Veil of Abstracta” Philosophical Issues 21, 245–67. Langsam, H. (2017) “The Intuitive Case for Naïve Realism” Philosophical Explorations 20:1, 106–22. Leddington, J. (2014) “What We Hear” in R. Brown (ed.) Consciousness Inside and Out: Phenomenology, Neuroscience, and the Nature of Experience (Dordrecht: Springer), 321–34. Lederman, S.J. and R.L. Klatzky (2009) “Haptic Perception: A Tutorial” Attention, Perception, & Psychophysics 71: 1439–59. Levin, D.T. and M.R. Banaji (2006) “Distortions in the Perceived Lightness of Faces: The Role of Race Categories” Journal of Experimental Psychology: General 135, 501–12. Levin, J. (2018) “Molyneux’s Question and the Amodality of Spatial Experience” Inquiry 61:5–6, 590–610. Lewis, C.I. (1952) “The Given Element in Empirical Knowledge” Philosophical Review 61, 168–75. Lewis, D. (1980) “Veridical Hallucination and Prosthetic Vision” Australasian Journal of Philosophy, 239–49. Lewis, D. (1986) “Causation” in his Philosophical Papers, Volume II (New York: Oxford University Press). Livitz, G.,A. Yazdanbakhsh, R.T. Eskew Jr. and E. Mingolla (2011) “Perceiving Opponent Hues in Color Induction Displays” Seeing and Perceiving 24, 1–17. Loar, B. (2003) “Phenomenal Intentionality as the Basis of Mental Content” in M. Hahn and B. Ramberg (eds.) Reflections and Replies: Essays on the Philosophy of Tyler Burge (Cambridge, MA: The MIT Press), 229–58. Locke, J. (1690/1961) An Essay Concerning Human Understanding (London: J.M. Dent and Sons). Loewer, B. (1997) “A Guide to Naturalizing Semantics” in C. Wright and B. Hale (eds.) A Companion to the Philosophy of Language (Oxford: Blackwell), 108–26.

256  References Logue, H. (2012a) “What Should the Naïve Realist Say About Total Hallucinations?” Philosophical Perspectives 26, 173–99. Logue, H. (2012b) “Why Naïve Realism?” Proceedings of the Aristotelian Society 112:2, 211–37. Lycan, W.G. (1996) Consciousness and Experience (Cambridge, MA: The MIT Press). Lycan, W.G. (2000) “The Slighting of Smell (with a Brief Word on the Slighting of Chemistry)” in S.M. Rosenfeld and N. Bhushan (eds.) Of Minds and Molecules: New Perspectives on Chemistry (New York: Oxford University Press), 273–89. Lycan, W.G. (2018) “What Does Taste Represent?” Australasian Journal of Philosophy 96:1, 28–37. Lyons, J.C. (2005) “Perceptual Belief and Nonexperiential Looks” Philosophical Perspectives 19, 237–56. Lyons, J.C. (2011) “Circularity, Reliability, and the Cognitive Penetrability of Perception” Philosophical Issues 21, 289–311. Mack, A. and I. Rock (1998) Inattentional Blindness (Cambridge, MA: The MIT Press). Macpherson, F. (2006) “Ambiguous Figures and the Content of Experience” Noûs 40:1, 82–117. Macpherson, F. (2007) “Synaesthesia, Functionalism and Phenomenology” in M. Marraffa, M. De Caro, and F. Ferretti (eds.) Cartographies of the Mind: Philosophy and Psychology in Intersection (Dordrecht: Kluwer), 65–80. Macpherson, F. (2011a) “Introduction: Individuating the Senses” in F. Macpherson (ed.) The Senses: Classic and Contemporary Philosophical Perspectives (Oxford: Oxford University Press), 3–43. Macpherson, F. (2011b) “Cross-Modal Experiences” Proceedings of the Aristotelian Society 111, 429–68. Macpherson, F. (2012) “Cognitive Penetration of Colour Experience: Rethinking the Issue in Light of an Indirect Mechanism” Philosophy and Phenomenological Research 84:1, 24–62. Macpherson, F. (2013) “The Philosophy and Psychology of Hallucination: An Introduction”, in F. Macpherson and D. Platchais (eds.) Hallucination: Philosophy and Psychology (Cambridge, MA: The MIT Press), 1–38. Margot, C. (2009) “A Noseful of Objects” Nature Neuroscience 12, 813–4. Marr, D. (1982) Vision: A Computational Investigation into the Human Representation and Processing of Visual Information (San Francisco: W. H. Freeman and Company). Martin, M.G.F. (1992) “Sight and Touch” in T. Crane (ed.) The Contents of Experience: Essays on Perception (Cambridge: Cambridge University Press), 196–215. Martin, M.G.F. (1994) “Perceptual Content” in S. Guttenplan (ed.) A Companion to the Philosophy of Mind (Oxford: Blackwell), 463–71. Martin, M.G.F. (1997) “The Reality of Appearances” in M. Sainsbury (ed.) Thought and Ontology (Milan: FrancoAngeli), 81–106. Martin, M.G.F. (2002) “The Transparency of Experience” Mind and Language 17:4, 376–425. Martin, M.G.F. (2004) “The Limits of Self-Awareness” Philosophical Studies 120:1–3, 37–89. Martin, M.G.F. (2006) “On Being Alienated” in T.S. Gendler andJ. Hawthorne (eds.) Perceptual Experience (Oxford: Oxford University Press), 354–410. Matthen, M. (1999) “The Disunity of Color” The Philosophical Review 108:1, 47–84. Matthen, M. (2005) Seeing, Doing, and Knowing: A Philosophical Theory of Sense Perception (Oxford: Clarendon Press). Matthen, M. (2014) “Active Perception and the Representation of Space” in D. Stokes, M. Matthen, and S. Biggs (eds.) Perception and Its Modalities (Oxford: Oxford University Press), 44–72. Matthen, M. (2015) “The Individuation of the Senses” in M. Matthen (ed.) The Oxford Handbook of Philosophy of Perception (Oxford: Oxford University Press), 567–83.

References 257 Matthen, M. and J. Cohen (2020) “Many Molyneux Questions” Australasian Journal of Philosophy 98:1, 47–63. Mattingley, J.B., J.M. Payne, and A.N. Rich (2006) “Attentional Load Attenuates Synesthetic Priming Effects in Grapheme-Colour Synaesthesia” Cortex 42, 213–21. McDowell, J. (1982/1998) “Criteria Defeasibility and Knowledge” in his Meaning, Knowledge and Reality (Cambridge, MA: Harvard University Press). McDowell, J. (1986/1998) “Singular Thought and the Extent of Inner Space” in his Meaning, Knowledge and Reality (Cambridge, MA: Harvard University Press). McDowell, J. (1994) Mind and World (Cambridge, MA: Harvard University Press). McDowell, J. (2008) “The Disjunctive Conception of Experience as Material for a Transcendental Argument” in A. Haddock and F. Macpherson (eds.) Disjunctivism: Perception, Action, Knowledge (Oxford: Oxford University Press), 376–89. McDowell, J. (2010) “Tyler Burge on Disjunctivism” Philosophical Explorations, 13:3, 243–55. McDowell, J. (2013) “Tyler Burge on Disjunctivism (II)” Philosophical Explorations, 16:3, 259–79. McGinn, C. (1982) The Character of Mind (Oxford: Oxford University Press). McGinn, C. (1996) “Another Look at Color” The Journal of Philosophy 93:11, 537–53. McGlone, F.P. and D. Reilly (2010) “The Cutaneous Sensory System” Neuroscience & Biobehavioural Reviews 34, 148–59. McGurk, H. and J. MacDonald (1976) “Hearing Lips and Seeing Voices” Nature 264, 746–48. McKitrick, J. (2005) “Are Dispositions Causally Relevant?” Synthese 144, 357–71. McLaughlin, B. (2003) “Colour, Consciousness, and Colour Consciousness” in Q. Smith andA. Jokic (eds.) Consciousness: New Philosophical Perspectives (Oxford: Clarendon Press), 97–154. Meier, B. and N. Rothen (2009) “Training Grapheme-Colour Associations Produces a Synaesthetic Stroop Effect, But Not a Conditioned Synaesthetic Response” Neuropsychologia 47, 1208–11. Mendelovici, A. (2018) The Phenomenal Basis of Intentionality (New York: Oxford University Press. Michotte, A. (1963) The Perception of Causality (London: Methuen). Mills, C.B., E.H. Boteler, and G.K. Oliver (1999) “Digit Synaesthesia: A Case Study Using a Stroop-type Test” Cognitive Neuropsychology 16:2, 181–91. Milner, A.D. and M.A. Goodale (1995) The Visual Brain in Action (Oxford: Oxford University Press). Mizrahi, V. (2014) “Sniff, Smell, and Stuff” Philosophical Studies 171, 233–50. Moran, A. (2018) “Naïve Realism, Hallucination, and Causation: A New Response to the Screening Off Problem” Australasian Journal of Philosophy 97:2, 368–92. Moran, A. (2019) “Naïve Realism, Seeing Stars, and Perceiving the Past” Pacific Philosophical Quarterly 100, 202–32. Morrot, G., F. Brochet, and D. Dubourdieu (2001) “The Color of Odors” Brain and Language 79, 309–20. Müller, J. (1837) Elements of Physiology (London: Taylor and Walton). Nagel, T. (1974/1979) “What Is It Like to Be a Bat?” in his Mortal Questions (Cambridge: Canto), 165–80. Neisser, U. (1967) Cognitive Psychology (New York: Appleton Century Crofts). Newen, A., L. De Bruin, and S. Gallagher (2018) The Oxford Handbook of 4E Cognition (Oxford: Oxford University Press). Noë, A. (2001) “Experience and the Active Mind” Synthese 129, 41–60.

258  References Noë, A. (2005) Action in Perception (Cambridge, MA: The MIT Press). Noë, A. (2008) “Reply to Campbell, Martin, and Kelly” Philosophy and Phenomenological Research 76:3, 691–706. Noë, A. and S. Hurley (2003) “The Deferential Brain in Action: Response to Jeffrey Gray” TRENDS in Cognitive Sciences 7:5, 195–6. Nudds, M. (2001) “Experiencing the Production of Sounds” European Journal of Philosophy 9, 210–29. Nudds, M. (2003) “The Significance of the Senses” Proceedings of the Aristotelian Society 104, 31–51. Nudds, M. (2011) “The Senses as Psychological Kinds” in F. Macpherson (ed.) The Senses: Classic and Contemporary Philosophical Perspectives (Oxford: Oxford University Press), 311–40. Nudds, M. (2015) “Audition” in M. Matthen (ed.) The Oxford Handbook of Philosophy of Perception (Oxford: Oxford University Press), 274–93. Nunn, J.A., L.J. Gregory, M. Brammer, S.C.R. Williams, D.M. Parslow, M.J. Morgan, R.G. Morris, E.T. Bullmore, S. Baron-Cohen, and J.A. Gray (2002) “Functional Magnetic Resonance Imaging of Synesthesia: Activation of V4/V8 by Spoken Words” Nature Neuroscience 5:4, 371–75. O’Callaghan, C. (2007) Sounds (New York: Oxford University Press). O’Callaghan, C. (2008) “Seeing What You Hear: Cross Modal Illusions and Perception” Philosophical Issues 18, 317–38. O’Callaghan, C. (2010) “Perceiving the Locations of Sounds” Review of Philosophy of Psychology 1, 123–40. O’Callaghan, C. (2011) “Hearing Properties, Effects or Parts?” Proceedings of the Aristotelian Society 61:3, 375–405. O’Callaghan, C. (2012) “Perception and Multimodality” in E. Margolis, R. Samuels, and S.P. Stich (eds.) The Oxford Handbook of Cognitive Science (Oxford: Oxford University Press), 92–117. O’Callaghan, C. (2014) “Intermodal Binding Awareness” in D.J. Bennett and C.S. Hill (eds.) Sensory Integration and the Unity of Consciousness (Cambridge, MA: The MIT Press), 73–103. O’Callaghan, C. (2015) “The Multisensory Character of Perception” Journal of Philosophy 112:10, 551–69. O’Callaghan, C. (2017) “Synesthesia vs. Crossmodal Illusions” in O. Deroy (ed.) Sensory Blendings: New Essays on Synaesthesia (Oxford: Oxford University Press), 45–58. O’Callaghan, C. (2019) A Multisensory Philosophy of Perception (Oxford: Oxford University Press). O’Dea, J. (2011) “A Proprioceptive Account of the Sense Modalities” in F. Macpherson (ed.) The Senses: Classic and Contemporary Philosophical Perspectives (Oxford: Oxford University Press), 297–310. O’Regan, J.K. (1992) “Solving the “Real” Mysteries of Visual Perception: The World as an Outside Memory” Canadian Journal of Philosophy 46:3, 461–88. O’Regan, J.K. and A. Noë (2001) “A Sensorimotor Account of Vision and Visual Consciousness” Behavioral and Brain Sciences 24:5, 939–73. O’Shaughnessy, B. (1989) “The Sense of Touch” Australasian Journal of Philosophy 67, 37–58. O’Shaughnessy, B. (1995) “Proprioception and the Body Image” in J-L. Bermúdez, N. Eilan, and A. Marcel (eds.) The Body and the Self (Cambridge, MA: The MIT Press), 175–203. Orlandi, N. (2013) “Embedded Seeing: Vision in the Natural World” Noûs 47:4, 727–47. Palmer, S.E. (1999) Vision Science: Photons to Phenomenology (Cambridge, MA: The MIT Press). Pani, J.R. (2000) “Cognitive Description and Change Blindness” Visual Cognition 7, 107–26.

References 259 Papineau, D. (2014) “Sensory Experience and Representational Properties” Proceedings of the Aristotelian Society 114:1, 1–33. Papineau, D. (2016) “Against Representationalism (About Conscious Sensory Experience)” International Journal of Philosophical Studies 24:3, 324–47. Pasnau, R. (1999) “What Is Sound?” The Philosophical Quarterly 49, 309–24. Pautz, A. (2010) “Do Theories of Consciousness Rest of a Mistake?” Philosophical Issues 20, 333–67. Pautz, A. (forthcoming) “Naïve Realism and the Science of Sensory Consciousness” Analytic Philosophy. Payne, K. (2001) “Prejudice and Perception: The Role of Automatic and Controlled Processes in Misperceiving a Weapon” Journal of Personality and Social Psychology 81:2, 181–92. Peacocke, C. (1983) Sense and Content (Oxford: Oxford University Press). Peacocke, C. (1992) A Study of Concepts (Oxford: Oxford University Press). Peacocke, C. (2001) “Does Perception Have a Nonconceptual Content?” The Journal of Philosophy 98: 239–64. Peacocke, C. (2008) “Sensational Properties: Theses to Accept and Theses to Reject” Revue Internationale de Philosophie 243, 7–24. Petkova, V.I., H. Zetterberg, and H.H. Ehrsson (2012) “Rubber Hands Feel Touch, but Not in Blind Individuals” PLoS ONE 7, e35912. Philipona, D.L. and J.K. ORegan (2006) “Color Naming, Unique Hues, and Hue Cancellation Predicted from Singularities in Reflection Properties” Visual Neuroscience 23, 331–39. Phillips, I. (2016a) “Naïve Realism and the Science of (Some) Illusions” Philosophical Topics 44:2, 353–80. Phillips, I. (2016b) “Consciousness and Criterion: On Block’s Case for Unconscious Seeing” Philosophy and Phenomenological Research 93:2, 419–51. Pitcher, G. (1971) A Theory of Perception (Princeton: Princeton University Press). Price, H.H. (1932) Perception (London: Methuen). Prinz, J.J. (2006) “Beyond Appearances: The Content of Sensation and Perception” in T. Gendler and J. Hawthorne (eds.) Perceptual Experience (Oxford: Oxford University Press), 434–60. Prinz, J.J. (2010) “When Is Perception Conscious?” in B. Nanay (ed.) Perceiving the World: New Essays on Perception (New York: Oxford University Press), 310–32. Putnam, H. (1975) “The Meaning of “Meaning”” in K. Gunderson (ed.) Language, Mind and Knowledge, Minnesota Studies in the Philosophy of Science, VII (Minneapolis: University of Minnesota Press), 131–93. Putnam, H. (1981) Reason, Truth and History (Cambridge: Cambridge University Press). Pylyshyn, Z. (1999) “Is Vision Continuous with Cognition?” Behavioral and Brain Sciences 22, 341–65. Pylyshyn, Z. (2003) Seeing and Visualising: Its Not What You Think (Cambridge, MA: The MIT Press). Raftopoulos, A. (2001) “Reentrant Neural Pathways and the Theory-Ladenness of Perception” Philosophy of Science 68, S187–99. Raftopoulos, A. (2009) Cognition and Perception: How Do Psychology and Neural Science Inform Philosophy? (Cambridge, MA: The MIT Press). Raleigh, T. (2014) “A New Approach to ‘Perfect’ Hallucinations” Journal of Consciousness Studies 21, 1–29. Ramachandran, V.S. and E.M. Hubbard (2001) “Synaesthesia – A Window into Perception, Thought and Language” Journal of Consciousness Studies 8, 3–34.

260  References Ratcliffe, M. (2012) “What Is Touch?” Australasian Journal of Philosophy 90:3, 413–32. Reid, T. (1785/2011) Essays on the Intellectual Powers of Man, (Cambridge: Cambridge University Press). Rensink, R. (2000) “The Dynamic Representation of Scenes” Visual Cognition 7, 17–42. Richardson, L. (2013a) “Bodily Sensation and Tactile Perception” Philosophy and Phenomenological Research 86:1, 134–54. Richardson, L. (2013b) “Flavour, Taste and Smell” Mind & Language 28:3, 322–41. Richardson, L. (2018) “Odors as Olfactabilia” in T. Crowther and C. MacCumhail (eds.) Perceptual Ephemera (Oxford: Oxford University Press), 93–114. Richardson, L. (forthcoming) “The Epistemological Power of Taste” Journal of the American Philosophical Association. Robinson, H. (1994) Perception (London: Routledge). Roll, J-P., R. Roll and J-L. Velay (1991) “Proprioception as a Link between Body Space and Extra-Personal Space” in J. Paillard (ed.) Brain and Space (New York: Oxford University Press), 112–32. Rolls, E.T. (2008) “Top-Down Control of Visual Perception: Attention in Natural Vision” Perception 37, 333–54. Rosenberg, G. (2004) A Place for Consciousness: Probing the Deep Structure of the Natural World (New York: Oxford University Press). Rosenthal, D. (1990) “A Theory of Consciousness” in N. Block, O. Flanagan, and G. Güzeldere (eds.) The Nature of Consciousness, (Cambridge, MA: The MIT Press), 773–88. Roskies, A. (2008) “A New Argument for Nonconceptual Content” Philosophy and Phenomenological Research 76:3, 633–59. Roxbee-Cox, J.W. (1970) “Distinguishing the Senses” Mind, 79, 530–46. Russell, B. (1912/1967) The Problems of Philosophy (Oxford: Oxford University Press). Ryle, G. (1949/1990) The Concept of Mind (London: Penguin). Schellenberg, S. (2011) “Perceptual Content Defended” Noûs 45:4, 714–50. Schellenberg, S. (2013) “Externalism and the Gappy Content of Hallucination” in F. Macpherson and D. Platchias (eds.) Hallucination (Cambridge, MA: MIT Press), 291–311. Schellenberg, S. (2016) “Perceptual Particularity” Philosophy and Phenomenological Research 93:1, 25–54. Schellenberg, S. (2018) The Unity of Perception: Content, Consciousness, Evidence (New York: Oxford University Press). Schwenkler, J. (2013) “Do Things Look the Way They Feel?” Analysis 73:1, 86–96. Scruton, R. (2009) “Sounds as Secondary Objects and Pure Events” in M. Nudds and C. O’Callaghan (eds.) Sounds and Perception: New Philosophical Essays (Oxford: Oxford University Press), 50–68. Searle, J. (1983) Intentionality: An Essay in the Philosophy of Mind (Cambridge: Cambridge University Press). Segal, G. (1989) “Seeing What Is Not There” Philosophical Review 97, 189–214. Sellars, W. (1975) “The Adverbial Theory of the Objects of Sensation” Metaphilosophy 6:2, 144–60. Shams, L., Y. Kamitani, and S. Shimojo (2002) “Visual Illusion Induced by Sound” Cognitive Brain Research 14, 147–52. Shepherd, G.M. (2011) Neurogastronomy (New York: Columbia University Press). Shevell, S.K. and F.A.A. Kingdom (2008) “Color in Complex Scenes” The Annual Review of Psychology 59, 143–66. Siegel, S. (2004) “Indiscriminability and the Phenomenal” Philosophical Studies 120, 90–112. Siegel, S. (2006) “Which Properties Are Represented in Perception?” In T.S. Gendler and J. Hawthorne (eds.) Perceptual Experience (Oxford: Oxford University Press), 481–503.

References 261 Siegel, S. (2007) “How Can We Discover the Contents of Experience?” Southern Journal of Philosophy 45, 127–42. Siegel, S. (2008) “The Epistemic Conception of Hallucination” in A. Haddock and F. Macpherson (eds.) Disjunctivism: Perception, Action, Knowledge (Oxford: Oxford University Press), 205–24. Siegel, S. (2009) “The Visual Experience of Causation” The Philosophical Quarterly 59, 519–40. Siegel, S. (2010) The Contents of Visual Experience (New York: Oxford University Press). Siegel, S. (2011) “Cognitive Penetrability and Perceptual Justification” Noûs 46:2, 201–22. Siegel, S. (2014) “Affordances and the Content of Perception” in B. Brogaard (ed.) Does Perception Have Content? (Oxford: Oxford University Press), 51–75. Siegel, S. (2015) “Epistemic Evaluability and Perceptual Farce” in J. Zeimbekis and A. Raftopoulos (eds.) The Cognitive Penetrability of Perception: New Philosophical Perspectives (Oxford: Oxford University Press), 405–24. Siewert, C.P. (1998) The Significance of Consciousness (Princeton: Princeton University Press). Simons, D.J. and C.F. Chabris (1999) “Gorillas in Our Midst: Sustained Inattentional Blindness for Dynamic Events” Perception 28, 1059–74. Simons, D.J. and D.T. Levin (1998) “Failure to Detect Changes to People During a Real-World Interaction” Psychonomic Bulletin and Review 5:4, 644–9. Small, D.M. (2012) “Flavor Is in the Brain” Physiology & Behavior 107, 540–52. Smith, A.D. (2002) The Problem of Perception (Cambridge, MA: Harvard University Press). Smith, A.D. (2011) “Tastes, Temperatures, and Pains” in F. Macpherson (ed.) The Senses: Classic and Contemporary Philosophical Perspectives (Oxford: Oxford University Press), 341–54. Smith, B. (2015) “The Chemical Senses” in M. Matthen (ed.) The Oxford Handbook of Philosophy of Perception (Oxford: Oxford University Press), 314–49. Smith, B. (2020) “Tasting Flavors: An Epistemology of Multisensory Perception” in B. Brogaard and D.E. Gatzia (eds.) The Epistemology of Non-Visual Perception (Oxford: Oxford University Press), 29–52. Snowdon, P. (1980) “Perception, Vision and Causation” Proceedings of the Aristotelian Society 81, 175–92. Snowdon, P. (1990), “The Objects of Perceptual Experience” Proceedings of the Aristotelian Society, 121–50. Snowdon, P. (2005) “The Formulation of Disjunctivism: A Response to Fish” Proceedings of the Aristotelian Society 105, 129–41. Soteriou, M. (2000) “The Particularity of Visual Perception” European Journal of Philosophy 8, 173–89. Soteriou, M. (2013) The Mind’s Construction: The Ontology of Mind and Mental Action (Oxford: Oxford University Press). Speaks, J. (2015) The Phenomenal and the Representational (Oxford: Oxford University Press). Spence, C. (2017) Gastrophysics: The New Science of Eating (New York: Viking Press). Spence, C., M. Auvray, and B. Smith (2014) “Confusing Tastes with Flavours” in D. Stokes, M. Matthen, and S. Biggs (eds.) Perception and Its Modalities (Oxford: Oxford University Press), 247–74. Spence, C. and T. Bayne (2014) “Is Consciousness Multisensory” in D. Stokes, M. Matthen, and S. Biggs (eds.) Perception and Its Modalities (Oxford: Oxford University Press), 95–132. Spiller, M.J., C.N. Jonas, J. Simner, and A. Jansari (2015) “Beyond Visual Imagery: How Modality-Specific Is Enhanced Mental Imagery in Synaesthesia?” Consciousness and Cognition 31, 73–85. Stevenson, R.J. (2011) “Olfactory Illusions: Where Are They?” Consciousness and Cognition 20: 1887–98.

262  References Stokes, D. (2015) “Towards a Consequentialist Understanding of Cognitive Penetration” in J. Zeimbekis and A. Raftopoulos (eds.) The Cognitive Penetrability of Perception: New Philosophical Perspectives (Oxford: Oxford University Press), 75–100. Stokes, D. (2018) “Attention and the Cognitive Penetrability of Perception” Australasian Journal of Philosophy 96:2, 303–18. Stokes, D. (2020) “On Perceptual Expertise” Mind & Language, 1–23. https://doi.org/10.1111/ mila.12270 Strawson, P.F. (1959) Individuals: An Essay in Descriptive Metaphysics (London: Methuen). Strawson, P.F. (1979) “Perception and its Objects” in G.F. MacDonald (ed.), Perception and Identity: Essays Presented to A.J. Ayer (London: Macmillan), 41–60. Strawson, P.F. (1992) Analysis and Metaphysics (Oxford: Oxford University Press). Sturgeon, S. (2006) “Reflective Disjunctivism” Proceedings of the Aristotelian Society, Supplementary Volume 80, 185–216. Thompson, E. (1995) Color Vision: A Study in Cognitive Science and Philosophy of Science (London and New York: Routledge). Thompson, B. (2009) “Senses for Senses” Australasian Journal of Philosophy 87:1, 99–117. Travis, C. (2004) “The Silence of the Senses” Mind 113:449, 57–94. Treisman, A.M. (1996) “The Binding Problem” Current Opinion in Neurobiology 6:2, 171–8. Treisman, A.M. and G. Gelade (1980) “A Feature-Integration Theory of Attention” Cognitive Psychology 12, 97–136. Tye, M. (1975) “The Adverbial Theory: A Defence of Sellars against Jackson” Metaphilosophy 6:2, 136–43. Tye, M. (1982) “A Causal Analysis of Seeing” Philosophy and Phenomenological Research 42:3, 311–25. Tye, M. (1984) “The Adverbial Approach to Visual Experience”, The Philosophical Review 93:2, 195–225. Tye, M. (1992) “Visual Qualia and Visual Content” in T. Crane (ed.) The Contents of Experience: Essays on Perception (Cambridge: Cambridge University Press), 158–76. Tye, M. (1995) Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind (Cambridge, MA: The MIT Press). Tye, M. (2000) Consciousness, Color and Content (Cambridge, MA: The MIT Press). Tye, M. (2002) “Representationalism and the Transparency of Experience” Noûs 36:1, 137–51. Tye, M. (2003a) “Blurry Images, Double Vision, and Other Oddities: New Problems for Representationalism?” in Q. Smith and A. Jokic (eds.) Consciousness: New Philosophical Perspectives (Oxford: Clarendon Press), 7–32. Tye, M. (2003b) Consciousness and Persons: Unity and Identity (Cambridge, MA: The MIT Press). Tye, M. (2007) “New Troubles for the Qualia Freak” in B.P. McLaughlin and J. Cohen (eds.) Contemporary Debates in Philosophy of Mind (Oxford: Blackwell), 303–18. Tye, M. (2009) “The Admissible Contents of Visual Experience” The Philosophical Quarterly 59, 541–62. Tye, M. (2012) “Cohen on Color Relationalism” Analytic Philosophy 53:3, 297–305. Ungerleider, L.G. and M. Mishkin (1982) “Two Cortical Visual Systems” in D.J. Ingle, M.A. Goodale, and R.J.W. Mansfield (eds.) Analysis of Visual Behavior (Cambridge, MA: The MIT Press), 549–86. Von Helmholtz, H. (1867) Treatise on Physiological Optics, trans. by J.P. Southall (New York: Dover). Vroomen, J. and B. de Gelder (2000) “Sound Enhances Visual Perception: Cross-Modal Effects of Auditory Organization on Vision.” Journal of Experimental Psychology: Human Perception and Performance 26, 1583–90.

References 263 Wager, A. (1999) “The Extra Qualia Problem: Synaesthesia and Representationalism” Philosophical Psychology 12:3, 263–81. Ward, J., R. Li, S. Salih, and N. Sagiv (2007) “Varieties of Grapheme-Colour Synaesthesia: A New Theory of Phenomenological and Behavioural Differences” Consciousness and Cognition 16:4, 913–31. Weiskrantz, L. (1986) Blindsight: A Case Study and Implications (Oxford: Oxford University Press). Weiskrantz, L., E.K. Warrington, M.D. Sanders, and J. Marshall (1974) “Visual Capacity in the Hemianopic Field Following a Restricted Occipital Ablation” Brain 97, 709–28. Wilson, D.A. and R.J. Stevenson (2006) Learning to Smell: Olfactory Perception from Neurobiology to Behavior (Baltimore: Johns Hopkins University Press). Whyte, J.T. (1990) “Success Semantics” Analysis 50, 149–57. Wright, C. (2002) “(Anti-)Sceptics Simple and Subtle: G.E. Moore and John McDowell” Philosophy and Phenomenological Research 65, 330–48. Wright, C. (2008) “Comment on John McDowell’s “The Disjunctive Conception of Experience as Material for a Transcendental Argument”” in A. Haddock and F. Macpherson (eds.) Disjunctivism: Perception, Action, Knowledge (Oxford: Oxford University Press), 390–404. Wu, W. (2017) Shaking up the Mind’s Ground Floor: The Cognitive Penetration of Visual Attention” The Journal of Philosophy 114:1, 5–32. Young, B.D. (2020) “Perceiving Smellscapes” Pacific Philosophical Quarterly, 101:2, 203–223. Zampini, M. and C. Spence (2004) “The Role of Auditory Cues in Modulating the Perceived Crispiness and Staleness of Potato Chips” Journal of Sensory Studies 19:5, 347–63.

Author Index

Akins, K. 192 Allen, K. 91, 117, 128, 133, 180–2, 192 Alter, T. 190 Antony, L. 123, 133 Armstrong, D. 22, 28, 43–8, 67 Aristotle 3, 9, 41, 192, 193, 200, 201 Austin, J. 19, 34, 102, 115, 116 Auvray, M. 188, 189, 190, 191, 192, 222, 224 Bach-y-Rita, P. 195 Barnes, W. 21–2, 34, 72 Baron-Cohen, S. 184, 185, 188 Batty, C. 184, 205, 218–20, 222, 223, 234 Bayne, T. 155, 225, 227, 228, 229, 230, 238, 239, 245 Beck, O. 124 Berkeley 6–7, 9, 24, 29 Bermúdez, J-L. 59 Block, N. 86–7, 89, 90, 91, 92, 99, 112, 155, 160–2, 163 Bourget, D. 52, 67, 69 Brewer, B. 63, 68, 69, 102, 113, 124, 133, 184, 222 Brentano, F. 36 Briscoe, R. 228, 239, 240, 241, 242, 243, 245 Brogaard, B. 174, 176, 183, 192 Brooks, R. 146 Bruner, J. 148, 156 Budek, T. 220 Burge, T. 55, 58–9, 62, 129–31, 133, 137 Butchvarov, P. 82–3

Butterfill, S. 155 Byrne, A. 23, 50, 69, 90, 91, 97, 123, 133, 173, 180, 181, 182–3, 192 Campbell, J. 66, 103, 105, 111, 113, 124, 130, 133, 176, 180, 232–3, 234 Carrasco, M. 112 Casati, R. 206, 209 Cassam, Q. 112, 133 Cavedon-Taylor, D. 220 Chalmers, D. 38, 52, 66, 69, 85, 90, 93, 96, 105, 181, 197 Chemero, A. 143 Child, W. 17, 105, 126, 127, 128 Chirimuuta, M. 75, 83, 84, 99, 177–9, 181, 183, 184, 192 Chisholm, R. 22 Clark, Andy 139, 141, 142, 162 Clark, Austen 198, 210 Coates, P. 105, 125 Cohen, J. 174–5, 176, 177, 178, 179, 182, 191, 192, 234 Cornman, J. 75 Crane, T. 37, 38, 49, 61, 68, 85, 86, 91, 99 Cytowic, R. 185 D’Ambrosio, J. 80 Dancy, J. 26 Davidson, D. 75 Davies, M. 54 Dennett, D. 85, 127, 143, 144, 161, 162 Deroy, O. 188, 189, 190–1, 192, 229–31

266  Author Index

Descartes, R. 3–6, 7, 102, 115, 130 Dinges, A. 80 Dokic, J. 117, 122, 133, 206, 219 Drayson, Z. 140, 163 Dretske, F. 62, 69 Driver, J. 241 Ducasse, C. 72 Evans, G. 59, 233 Farkas, K. 98, 99, 220 Feyerabend, P. 153 Firestone, C. 147–50, 163 Firth, R. 41–2 Fish, W. 9, 34, 103, 105, 111, 113, 119, 120, 122, 124, 133, 155, 162 Fodor, J. 62, 138, 147 Foster, J. 29 French, C. 113, 124 Fulkerson, M. 210–11, 212, 224 Fumerton, R. 105 Gallace, A. 209 Gallagher, S. 232, 236 Gazzaniga, M. 227 Gelade, G. 202, 230 de Gelder, B. 199, 237 Genone, J. 103, 124, 133 Gert, J. 83–4, 99, 179–80, 181, 182 Gibson, J.J. 142–6, 162, 163 Glüer, K. 48, 69 Goodale, M. 158–9 Goodman, C. 148, 156 Gray, R. 215 Grice, H. 30–1, 32, 34, 194–200, 202, 210, 224 Grzankowski, A. 96 Haffenden, A. 159 Hahn, M. 192 Hanson, R. 153 Hardcastle, V.G. 93 Hardin, L. 172–3, 183 Harman, G. 62, 64, 69, 85, 87, 91, 92 Harrison, J. 184–5, 188 Heck, R.G. 68 Held, R. 234–5

Hellie, B. 104, 117 Hering, E. 169–170, 171, 172 Hilbert, D. 173, 180, 181, 182–3, 192 Hill, C. 86 Hinton, J. 107 Hobson, K. 105, 125 Hohwy, J. 138, 139–42, 162 Horgan, T. 96, 98 Hume, D. 7, 9, 103, 111 Hurley, S. 195 Hurvich, L. 170 Jackson, F. 24, 26, 34, 76, 79–80, 84, 96, 99 James, W. 213 Jameson, D. 170 Johnston, M. 103, 115–17, 133, 174, 176, 179, 180, 183, 191, 192 Kalderon, M. 124, 182, 184, 192, 206 Kanwisher, N. 161 Keeley, B. 191, 193, 196, 197, 199, 203, 224 Kind, A. 91–2, 99 Kon, M. 207 Kriegel, U. 28, 66, 71–2, 95–9 Kuhn, T. 153 Langsam, H. 104, 133 Leddington, J. 206 Levin, J. 235 Lewis, C.I. 26–7 Lewis, D. 31–3, 34, 46, 127 Loar, B. 98 Locke, J. 5–6, 7, 8, 9, 174, 232 Loewer, B. 69 Logue, H. 102, 103, 119–20, 123, 133 Lycan, W. 51, 62, 69, 216, 217, 218, 220, 221 Lyons, J. 153, 154–5, 157, 163 Macpherson, F. 89, 112, 117, 133, 148, 152, 163, 190, 192, 203–4, 224, 227, 228, 230, 231, 245 Marr, D. 39, 138, 162 Martin, J-R. 117, 122, 133

Author Index  267

Martin, M.G.F. 37, 49, 103, 104, 108, 110, 116, 118, 119, 120, 121–2, 133, 212, 214, 224 Massin, O. 203, 213, 214 Matthen, M. 183, 184, 202, 204, 217, 223, 224, 245 McDowell, J. 60, 103, 115, 123, 130–1, 133 McGinn, C. 55, 176 McGurk, H. 236, 238, 244 McLaughlin, B. 179, 13 Mendelovici, A. 98, 99 Miller, K. 207 Milner, A. 158, 159 Mishkin, M. 158 Mizrahi, V. 220, 224 Moore, G.E. 34 Moran, A. 114, 128 Müller, J. 165 Nagel, T. 2, 197 Neisser, U. 39, 138 Newen, A. 144 Noë, A. 105–6, 133, 144, 145–6, 163, 194, 195 Nudds, M. 207, 208, 223, 224 O’Callaghan, C. 190, 191, 206, 208, 224, 225–6, 228, 229, 230, 231, 232, 233, 238, 239, 244, 245 O’Dea, J. 194, 195–6, 224 O’Regan, K. 145–6, 163, 172, 194 O’Shaughnessy, B. 211, 212, 215 Orlandi, N. 146 Palmer, S. 166 Pani, J. 145 Papineau, D. 66, 69, 85, 87, 94, 95, 99 Pasnau, R. 204 Pautz, A. 50, 131–2, 133, 169 Peacocke, C. 37, 59, 87, 89 Phillips, I. 124, 199 Pitcher, G. 43, 45, 47, 114 Price, H.H. 13–14, 19, 27, 28, 34, 35, 40, 43

Prinz. J. 155, 157, 199 Putnam, H. 58, 161, 162 Pylyshyn, Z. 138, 147, 149, 150, 151, 152–3, 157, 163 Raftopoulos, A. 151–2, 153, 158, 163 Ramachandran, M. 185–6 Ratcliffe, M. 203, 205, 210, 240 Reid, T. 7–8, 9, 39 Rensink, R. 145 Richardson, L. 213–14, 216, 217, 218, 220–1, 222, 223, 224 Robinson, H. 14, 21, 22, 24, 34, 42, 73, 116, 123 Rosenberg, G. 189 Rosenthal, D. 51, 69 Roskies, A. 61 Roxbee-Cox, J. 201–2, 224 Russell, B. 34, 54, 56, 233 Ryle, G. 22, 25 Schellenberg, S. 48–9, 57, 63, 69, 109 Scholl, B. 147–50, 163 Schwenkler, J. 233, 234–5 Scruton, R. 209 Searle, J. 55 Segal, G. 59 Sellars, W. 79, 81 Siegel, S. 37, 38, 49, 69, 110, 111, 122, 133, 149, 152, 153–4, 155–7, 163 Siewert, C. 96 Simons, D. 144 Smith, A.D. 19, 56, 57, 123, 212–3, 214, 215, 216–7, 224 Smith, B. 215, 216, 221–2, 224 Snowdon, P. 109, 110, 126, 127 Soteriou, M. 55, 113 Speaks, J. 243 Spence, C. 209, 216, 221–2, 224, 228–9, 230, 237, 238, 239, 241, 245 Strawson, P.F. 24, 49, 81, 126, 155, 208

268  Author Index

Stevenson, R. 132, 220, 221 Stokes, D. 152, 154, 224, 245 Sturgeon, S. 120–1

de Vignemont, F. 203, 213, 214, 238 von Helmholtz, H. 137–8, 149, 162, 165, 170

Thompson, B. 179 Thompson, E. 178, 183 Travis, C. 63, 102 Treisman, A. 202, 230 Tye, M. 23, 33, 46–7, 51, 55–6, 57, 60, 62, 65, 67, 69, 75, 79–80, 81, 88, 91, 175, 182–3, 184, 220, 227, 228

Wager, A. 189 Weiskrantz, L. 66 Wilson, D. 132, 220 Whyte, J. 62 Wittgenstein, L. 22 Wright, C. 115 Wu, W. 150, 152

Ungerleider, L. 158

Zampini, M. 237

Young, B. 220

Subject Index

abstract content 51, 55, 219 accuracy conditions 37–8, 49, 99 achromatopsia 159, 202 acquaintance relation, the 102, 105–6, 112–14, 181, 210 actionism 105 admissible contents 155–7, 200 adverbialism 71–86, 95–6, 215; color 83–4, 177–9; and metaphysics 75, 86; and adverbial translations 73–4, 76–80 affordances 143 afterimages 14–15, 74, 76–80, 84, 89, 92, 170–2, 188, 190 agnosia, types of 159–60 akinetopsia 160 alternate visual systems 181–2, 184 ambient optic array, the 144–5. 147 ambiguous figures 89–90 Anstoss, the 213, 215 apperceptive agnosia 158–9 argument from hallucination, the 16–19, 27 argument from illusion, the 16–19, 27, 44 attention: and cognitive penetration 149–50, 151, 152; and inattentional blindness 144; overt and covert 150; and phenomenology 90, 91–2, 112; shifts explaining apparent phenomenal unity 229, 231; and transparency 64, 66, 91 audition see hearing Bayesian inference theory see predictive processing belief acquisition theory 43–8, 49–50, 72, 149; and the persistence of

illusions 47–8; and justification 45; and unchanging scenes 45 binding problem, the 88, 230 binocular rivalry 140–1 binocular vision 24, 89 blindsight 66, 199 blurred vision 75, 89, 91, 112–13, 124 bodily awareness 211–14, 237, 241–2; and tactile perception 212–14 body map theory 214 bottom-up explanations 63, 139–41, 151 brain before the eyes, the 31, 32, 46 capacitism 63 causal objection, the 116, 128–9 causal theory of perception, the 29–34, 46–7, 50; and naïve realism 125–8 causation: counterfactual theory of 33; the visual experience of 155, 157 change blindness 143–4, 161, 202, 229 characteristic experience view, the 196–9, 226 characteristic property view, the 200–3, 205; the key feature version of 201–3 characteristic sense-organ view, the 194–5 characteristic stimulus view, the 195–6 chemical senses, the 215–23; as exteroceptive experience 218; philosophical theories of 217–18, 219–22; subjectivism about 216–18 chemesthesis 215 clock on the shelf, the 30–1

270  Subject Index

cognitive penetration 147–55; and epistemology 153–5; and expert knowledge 154 cognitive science 137, 147, 152–3 color 165–91; adverbialism 83–4, 177–9; coarse-grained (see rough color); the components of 169; constancy 174, 181, 182, 184; dispositionalism 6, 174–7; eliminativism about 173–4, 175, 180; fine-grained 60, 83, 175, 179; inter-species variation 181, 182, 184; inter-subjective variation 112, 172–3, 181, 184; intra-subjective variation 175, 177, 181, 184; naïve realism about 181–2; the science of 166–73; pluralism 175, 182, 184; primitivism 179–82; reductive realism 182–4; relationalism 174–7; relativism 176; rough 83, 175–6, 179; synesthesia 184–91 Common Factor Principle, the 17–21, 44, 50, 72, 75, 101–2, 106–7, 125–8; as a conceptual truth? 126–7 common sensibles 200, 232, 235 complement objection, the 84 composite snapshot conception, the 225–7, 228; see also the decomposition thesis concepts: acquiring new 61; belief and 59; demonstrative 60; nonconceptual 52, 59–61, 68; of an external world 25; spatial 208, 233 conceptual content 60–1, 68, 149 concurrent 184–5, 186, 187, 188, 189, 190, 191 content 36–38; abstract 51, 54–5, 219; admissible 155–7, 200; conceptual/nonconceptual 51–2, 59–61, 68, 147, 158; everyday notion of 38; Fregean 54, 56–7, 63, 179; gappy 56–7, 61, 109; internalism/externalism about 57–9; naturalistic theories of 58, 61–3, 190; possible world theory of 54; schema 56; singular 54–7, 63, 108–9

continuity, perception to illusion 19, 56–7, 123 cross-modal illusions 236–9, 243–4; and cross-modal interactions 238–9 decomposition thesis, the 225–7, 228, 232, 235, 243 demonstrative concepts 60 depth perception 24, 41, 42 desynchronized audio/video 231 determinable-determinate relation, the 96–7, 182 direct perception 5–8, 14, 23, 25, 26, 29, 82–3, 102–3, 123, 137–8, 200–2, 205, 215, 220, 232 disjunctivism 106–11, 114–15, 123–5, 126–7, 129–31; about content 108–9; epistemological 109–10; and illusion 123–5; about metaphysics 108; multi- 117; about phenomenology 109–11 dogmatism within epistemology 153–4 dorsal stream, the 158–9 doxastic theory see belief acquisition theory duck-rabbit, the 149 early vision 147–8, 149, 151, 153, 154, 158 Ebbinghaus illusion, the 159 ecological psychology 142–3, 145, 146 epistemological hat, the 2; adverbialism and 82–3; belief acquisition theory and 45–6; intentionalism and 67–8; and Modern philosophy 6–9; naïve realism and 114–15; phenomenal intentionalism and 98–9; qualia theory and 94–5; sense datum theory and 25–8; sensory core theory and 39–41 embedded cognition see 4E cognition embodied cognition see 4E cognition enactive cognition see 4E cognition encapsulation see informational encapsulation

Subject Index  271

exing 50 explanation, about color 174 extended cognition see 4E cognition externalism: about content 57–9; about justification 45–7, 67; and Molyneux’s question 233–4; about phenomenology 58, 234 exteroceptive experience 218 feature binding 202, 230; intermodal 229–32, 235, 236, 242 feed-forward sweep (FFS) 151 flavor see tastes and flavors foundationalism 26–7, 82 4E cognition 145 Fregean contents 54, 56–7, 63, 179 fusiform face area 161 gappy contents 56–7, 61, 109 generative multisensory integration 240–2 global recurrent processing (GRP) 151 gustation 215–16, 222–3, 228, 239 hallucination 14–19, 27, 31–2, 44, 50, 56–7, 63, 72, 75–6, 84, 101, 106–11, 114–25, 128, 130, 140, 189–91, 208; perceptual theories of 106; positive theories of 115–17; epistemic/eliminativist theories of 117–22; multidisjunctive theory of 117; veridical 31–4, 46, 55, 76, 125–6 hearing 193–5, 200–1, 204–9, 220, 236–8; the objects of 205–7; philosophical theories of 208–9; the spatiality of 208; and synesthesia 185, 190 higher-level properties 155–6 higher-order belief theory see higher-order thought theory higher-order experience theory 51–3, 67 higher-order perception theory see higher-order experience theory higher-order thought theory 51–3, 67

hues: location of unique 112, 172–3, 181, 184; unique-binary distinction 169–71 hylomorphic physics 3 idealism 6–7, 9, 29 ideas 5–8 illusion 15–19, 27, 43–4, 46, 47–8, 50, 56–7, 63, 75–6, 101, 109, 122–5, 149, 159, 174, 184, 189–90, 219, 221–3, 236–9, 240, 242, 243–4; color 174, 184; cross-modal 236–9, 240, 243–4; the Ebbinghaus 159; as a case of hallucination (V vs. IH disjunctivism) 123; the parchment skin 237; as a case of perception (VI vs. H disjunctivism) 123–5; the McGurk effect 236, 238–9; the Müller-Lyer 47, 149, 238; of object motion induced by vibration 242; the rubber hand 237–8; smell and taste 219, 221–3; the sound-induced flash 236–7; the vase-face 149; the ventriloquism effect 236, 238 inattentional blindness 144, 161–2, 229 inclination to believe 47–8 incorrigible, definition of 27 indirect perception 5–7, 25, 82, 103, 200, 205 indiscriminability: and the characteristic experience view 197–8, 210; the concept of 17, 120–22; as identity of cognitive effects 119, 122; -based approaches to hallucination 117–20; and phenomenal character 110–11 indistinguishability see indiscriminability inducer 184–5, 187, 188, 189, 190 indubitable, definition of 27 infallible, definition of 27 information processing paradigm in cognitive science, the 137–9 informational encapsulation 147–9, 151–4

272  Subject Index

intentionalism 49–68; first-order versions of strong 51, 53, 64–7; higher-order versions of strong 51, 53; and the information processing paradigm 139; and the Mirroring Thesis 52–3, 87; and the perceptual attitude 50; and psychosemantics 61–3, 190; strong 53, 58, 62–3, 68, 87; and theories of perceptual content 53–61, 155–7; and transparency 64–6; and veridicality 50, 76; weak 51, 53, 87–8 intentionality 21, 36–9, 44, 71–2, 95–9; and Brentano 36; and intentional objects 72; phenomenal 95–9, 173 intermodal feature binding 229–32, 235, 236, 242; and generative multisensory integration 242 internalism: about content 57, 59; about justification 26–8, 45, 67–8, 82; and Molyneux’s question 232, 234; about phenomenology 58 intuitions, clashes of 157, 197 inverted Earth 91 inverted spectrum 90–1, 208–9 isomers 166 key feature view, the 201–3 kind properties and perceptual contents 155–7 kinesthesia 211 lateral geniculate nucleus (LGN) 158 light meter, the 31, 32, 46 localized recurrent processing (LRP) 151 low-level properties 150, 155–6, 186, 234 many property problem, the 76–80, 84, 88, 96–7 many relations problem, the 80 McCollough effect, the 188 McGurk effect, the 236, 238–9, 244 mechanistic physics 3–4, 6, 8, 137, 165 mental objects see sense datum theory

meta-character 197 metacognitive error 117 metamers 166–7, 169, 180, 182 method of phenomenal contrast, the 156–7 minimal multimodality, the thesis of 227–9, 231 Mirroring Thesis, the 52, 63, 87–91, 97 misrepresentation 56, 58–9, 222 Modern Philosophy, perception and 3–9 modesty, as an argument for disjunctivism 110–11 Molyneux’s question 232–5, 243; and cross-modal phenomenology 233; empirical studies of 234–5; and spatial awareness 233–4 multidisjunctivism 117 multimodality 222–3, 227–44 multimodally accessible features 210, 221–2, 239–42 multisensory: integration 230, 239, 240–2; processing 230 naïve realism 102–32, 142, 146, 166, 173, 177, 180–2, 183–4, 209, 215, 221, 222, 243; and the causal theory of perception 125–8; about color 180–2; and direct realism 103; and disjunctivism 106–8, 109–10; and empirical science 128–32, 234; and hallucination 115–22; and hearing 209; and illusion 122–5; and the chemical senses 221, 222; and predictive processing 142; and the Representational Principle 102; and synesthesia 189–90; and touch 215; and theories of color 166, 173, 177, 180–2, 183–4; as the theory of the naïve 104 naturalistic theories of content 58, 61–3, 190 Neckar cube, the 149 nonconceptual content 51–2, 59–61, 68, 147; and concept-less creatures 59; and concept learning 61 nonconscious experiences 199

Subject Index  273

object unity 229–31, 238 odor 198, 201, 216–7, 219–22; objects 132; philosophical theories of 219–21 olfaction 132, 215–16, 222–3, 239; orthonasal 216, 240; retronasal 216, 222, 240 olfactory epithelium, the 216, 222, 223, 240 olfactory qualities 220 opponent processing system 170–2, 183 optimizing multisensory integration 239, 240 orthonasal olfaction 216, 240 PANIC 51–2, 55, 65, 67 paradox of phenomenal judgment, the 93–4 parchment skin illusion, the 237 particularity 108–9 percept theory 41–3 perception and cognition 41, 147–57, 160–2 peripersonal space 214 Perky experiments, the 191 phenomenal character 23; and the admissible contents of experience 155–7; and cognitive penetration 148–50; and cross-modal interactions 238; eliminativist theories of hallucinatory 117–20; as identical to content 65; and indiscriminability 110–11; and gappy contents 56; hybrid theories of 88; and the Mirroring Thesis 53, 97; modality specific 194, 196–9, 226, 231–2, 235, 243; and Molyneux’s question 232–3; multimodal 233, 239–42; and naïve realism 103–4, 108–14, 124, 131–2, 182, 184, 209; and phenomenal intentionality 96–8, 173; positive theories of hallucinatory 115–17, 128; and qualia theory 85–8; and representationalism 53, 65, 189–90; and the screening off problem 116; according to sense datum theory 40; unifying 228

phenomenal contrast, the method of 156–7 phenomenal intentionality 95–9, 173 Phenomenal Principle, the 13–16, 20, 27, 39, 41, 42–3, 44, 65–6, 72–3, 75–6, 81, 101–2, 106, 208, 218 phenomenal unity 227–9, 231 phenomenism 87 phenomenological hat, the 2; adverbialism and 80–1; belief acquisition theory and 47; intentionalism and 52–3; and Modern philosophy 4–6, 7, 8–9; naïve realism and 111–14; phenomenal intentionality and 98–9; qualia theory and 92–4; representationalism and 64–7; sense datum theory and 23–4; sensory core theory and 40 phosphenes 89, 92 photoreceptors 167–73, 181 pop-out experiments on synesthetes 186–7 possible world contents 54 posterior piriform cortex (PPC) 132 prediction error 139, 141–2 predictive processing 139–42 presentational character 23; according to naïve realism 104, 111; according to qualia theory 87–8; according to representationalism 64–5; according to sense datum theory 23; according to sensory core theory 40 primary qualities 6–7 primary visual cortex 66, 158 primitivism 179–82, 183, 184; color 179–82; flavor 221; odor 220; sound 209 proper sensibles 200, 210, 243 propositional attitude 49–50, 59 proprioception 211, 240, 242 prosopagnosia 160 proximality principle, the 129 psychosemantics see naturalistic theories of content pure representationalism 52, 53, 67, 190, 243

274  Subject Index

qualia: uses of the term 85; dualist theories of 93–4; physicalist theories of 92–3 qualia theory 84–95; and adverbialism 86; and weak intentionalism 87, 88, 94; and phenomenal intentionality 98 qualities, primary and secondary 6–7 realism: metaphysical 6–9 re-entrant processing 151 reflectance types see surface spectral reflectance (SSR) relationalism see naïve realism reportability and phenomenal consciousness 158–62 Representational Principle, the 36, 38, 39, 41, 43, 44, 48, 50, 82, 87, 98; and naïve realism 102 representation 36–8; in cognitive psychology 39; and content 38; symbolic / linguistic 37; vehicle of 37 representationalism 51–3, 64–8, 97, 105; and admissible contents 155; and theories of color 166, 173, 176, 179, 181; and cognitive psychology 39, 139; and direct realism 103; defenses of the Mirroring Thesis 90, 92; and ecological theories 146; and hearing 209; impure 51–2, 243; and nonconceptual content 60–1; and predictive processing 140, 142; and psychosemantics 62–3, 190; pure 52, 53, 67, 190, 243; and smell 222; and synesthesia 189; and touch 215; and transparency 64–5 retronasal olfaction 216, 222, 240 revelation, about color 176–7, 179–81 robots see Tom and Tim rough colors 83, 175–6, 179 rubber hand illusion, the 237–8 satisfaction 30–1, 38, 46 screening off problem, the 116–20, 128 secondary qualities 6

sensations 4–5, 7–9, 13, 79, 147, 151, 165, 214, 215; of bodily awareness 212–14; and the chemical senses 216–17; and qualia theory 87 sense datum theory 16–29, 35–6, 75, 81, 82, 84, 101, 102, 104; and the Anstoss 213; the causal theory of perception 29–34, 76; and theories of color 173, 174; and externalist theories of justification 46; of hallucination 116; and hearing 208; the metaphysics of 21–2, 105; and percept theory 41–3; and sensory core theory 39–41; and smell 217–18; and synesthesia 189; and touch 212–13, 215 sense organs 137, 194–5, 205, 230; functional views of 194–5 senses: individuating the 193–205, 222–3, 240, 242; characteristic experience view 196–9; multidimensional view 203–4; perceptual systems view 204–5; characteristic properties view 200–3; characteristic sense organ view 194–5; characteristic stimulus view 195–6; number of 193 sensing, the relation of 22–3, 104–6 sensorimotor: skills 105–6; contingency theory 145–6 sensory core theory 39–41 sensory/cognitive distinction, the 41, 147–57, 160–2 simultagnosia 160 simultaneous contrast 175, 177, 181, 184 singular contents 54, 55–7, 63, 108–9 skepticism: external world 7, 103, 114–15, 213; about multimodality 228–9, 235, 236, 239; about synesthesia 186 smell 131–2, 193, 194, 196–7, 198, 200, 201, 215–23, 226, 237, 239, 244; as exteroceptive experience 218; illusions of 221–2, 237; and object-directedness 8; philosophical

Subject Index  275

theories of 217–18, 219–20; as properties of ideas 6; subjectivism about 216–18, 221; and flavor 222–3, 240 Sounds 205, 209, 229, 231, 233, 241; the components of 201; mislocating 236–7, 238–9, 244; theories of 205–8 sound-induced flash illusion, the 236–7 split-brain patients 227–8 Stroop effect, the 187 surface spectral reflectance (SSR) 182–4; see also color, reductive realism synesthesia 184–91; concurrents and inducers 184; cross-modal 185, 198, 227; Cynthia case 189–90; as phenomenal enrichment 190–1; lower vs higher 186; and the McCollough effect 188; as nonperceptual 190; and pop-out experiments 186–7; in popular culture 185; projector vs associator 186; and theories of visual perception 189–90; the robustness of 188; and the Stroop effect 187 tactile-visual substitution system (TVSS) 195–6 taste, the sense of 193, 200, 201, 202, 215–23, 244; as exteroceptive experience 218; illusions of 221, 237; the objects of 216; philosophical theories of 217–18, 221; the physiology of 216, 223; subjectivism about 216–17, 221; and synesthesia 185; unimodal vs multimodal 222–3, 239–40 tastes and flavors 216, 239; as properties of ideas 6 teleosemantics 62–3, 190 temperature: phenomenology 215; and taste 221; and touch 200, 201, 203, 209–10 temporal perception 207

theory-ladenness of observation, the 153 thermometer view, the 45 time lag argument, the 20, 114 Tom and Tim 33–4, 46–7, 55–6 top-down explanations 148, 151 touch 193, 197, 201–3, 209–15, 226; and bodily awareness 211–14; distal 213–14; haptic 211, 240; and Molyneux’s Question 232, 235; number of senses involved in 210–11, 240, 242; passive / cutaneous 211; philosophical theories of 215; physical stimuli involved in 195, 196; the physiology of 209; the spatiality of 214–15, 233–4, 241; and taste 216–17, 222; and TVSS 195–6 transparency about color 176 transparency of experience, the 64; objections to 88–9, 92; and phenomenal intentionality 96; and representationalism 64–6 trichromatic vision 168; as “normal” 181; and opponent processing 170–1 trigeminal nerve, the 215 Twin Earth 58 Two Hats, the 1–2; see also the epistemological hat; the phenomenological hat two visual systems hypothesis, the 158, 199 unconscious inference 138, 149 underdetermination problem, the 138, 143, 146 unilateral neglect 160, 161 unimodality 231, 239–42 unique hues 169–71; the location of 112, 172–3, 181, 184 unity, about color 183 unity of consciousness, the 227–9 V1 see primary visual cortex vase-face illusion, the 149

276  Subject Index

veil of perception 25, 26, 45, 212 ventral stream, the 158–9 ventriloquism effect, the 236, 238 veridical perception, definition of 15 veridical hallucination 31–4, 46, 55, 76, 125–6

vestibular system, the 211 vibration-induced illusions of object motion 242 visual experience: definition of 15; the boundaries of 113 visuo-spatial extinction 161 vomeronasal sense 199, 203